Coprtribrctors to Volume 9

P. G. DRAZIN
L. N. HOWARD
N. N. MOISEEV
A. A. PETROV
PIOTRPBRZYNA
R. M. ROSENBERG
 ADVANCES IN
APPLIED MECHANICS

Edited by
G. G. CHERNYI W. OLSZAK
H. L. DRYDEN W. PRAGER
P. GERMAIN R. F. PROBSTEIN
L. HOWARTH H. ZIEGLER

Managing Editor
G. KUERTI
Case Institrzte of Technology, Cleveland, Ohio

VOLUME 9

1966

ACADEMIC PRESS NEW YORK AND LONDON

COPYRIGHT0 1966. B Y ACADEMICm S S INC.
ALL RIGHTSRESERVED.
N O PART OF THIS BOOK MAY BE REPRODUCED I N ANY
FORM, B Y PHOTOSTAT, MICROFILM, OR ANY OTHER MEANS,
WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS.

ACADEMIC PRESS INC.

111 Fifth Avenue, New York, N e w York 10003

United Kingdom Edition pub&hrd by

ACADEMIC PRESS INC. (LONDON) LTD.
Berkeley Square House, London W.l

LIBRARY
OF CONGRESS CATALOG CARD 48-8503
NUMBER:

PRINTED I N THE UNITED STATSS OF AMERICA

 List of Contributors
P. G. DRAZIN,Department of Mathematics, University of Bristol,
England
I.. N . HOWARD, Mathematics Department, Massachusetts I%stitute of
Technology, Cambridge, Massachzcsetts
N. N . MOISEEV, Computing Center of th.e U.S.S.R. Academy of Sciences,
Moscow, U.S.S.R.
A. A. PETROV,Comfizcting Center of the U.S.S.R. Academy of Sciences,
Moscow, U.S.S.R.
PIOTR PERZYN A , Institute of Basic Technical Research, Polish Academy
of Sciemes, Warsaw, Poland
R. M. ROSENBERG, Department of Mechanical Engilteering, Division
of Applied Mechanics, University of California, Berkeley, California

V
 In preparing this ninth volume of the “Advances in Applied Mechanics”
we still had the invaluable aid of Dr. Dryden who, unfortunately, died on
December 2, 1966. His advice and judgment will be sorely missed in the
future.
Stability theory in two different fields of mechanics is the subject of two
contributions in this volume. The volume also contains a detailed report on
the sloshing of liquid in containers; the basic analysis of the sloshing problem
was set forth in Volume 8. A survey of the present theories of elastic-visco-
plastic behavior concludes the volume.
It is intended to publish at least the next volume of the “Advances” in
the form of several successive fascicles, in order to present the material as
rapidly as possible.
THE EDITORS
May, 1966

vii
 Hydrodynamic Stability of Parallel Flow of Inviscid Fluid
BY P . G. DRAZIN AND L. N . HOWARD

Department of Mathematics. University of Brisiol. England

and
.
Mathemgiacs Department. Massachusetts Institute of Technology Cambridge. Mass .

Page
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
I1 . Inertial Instability . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1. Eigenvalue Problem for Inertial Modes . . . . . . . . . . . . . . . . 3
2. General Stability Characteristics of Plane Parallel Flow . . . . . . . . 10
3 . The Initial-Value Problem and the Stability of Non-parallel Flow . . . . 22
4 . Stability Characteristics of Various Basic Flows . . . . . . . . . . . . 32
I11. Waves and Stability of Plane Parallel Flow of Inviscid Fluid under the Actions
of Various Force Fields . . . . . . . . . . . . . . . . . . . . . . . 43
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2. Internal Gravity Waves and Stability of a Fluid of Variable Density . . . 44
3. Sound Waves and Stability of Compressible Fluid . . . . . . . . . . . 47
4 . Planetary Waves and Stability in a Rotating System . . . . . . . . . 49
5 . Rossby Waves and Stability of Fluid in a Rotating System with Variable
Coriolis Parameter . . . . . . . . . . . . . . . . . . . . . . . . . 51
6. Magnetohydrodynamic Waves and Stability of an Electrically-Conducting
Fluid in a Magnetic Field . . . . . . . . . . . . . . . . . . . . . 62
IV . Heuristic Theory of Instability . . . . . . . . . . . . . . . . . . . . 54
1. Dimensional Analysis . . . . . . . . . . . . . . . . . . . . . . . 54
2. Physical Arguments . . . . . . . . . . . . . . . . . . . . . . . . 57
V. Instability of an Incompressible Fluid of Variable Density . . . . . . . . 80
1. General Stability Characteristics . . . . . . . . . . . . . . . . . . . 60
2 . Stability Characteristics of Various Basic Flows . . . . . . . . . . . . 68
VI . Stability of Other Parallel Flows . . . . . . . . . . . . . . . . . . . 79
1. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2 . The Semicircle Theorem for General Parallel Flow . . . . . . . . . . . 79
3. Inertial Instability of Axisymmetric Jets . . . . . . . . . . . . . . . 80
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

I . INTRODUCTION

Stability of parallel flow of inviscid fluid was first studied in the last
third of the nineteenth century. notably by Helmholtz [l]. Kelvin [2] and
Rayleigh [S] . They considered the inertial instability of a homogeneous
incompressible fluid. and its modification- Kelvin-Helmholtz instability-
when there is variation of density of the fluid transverse to the basic flow.
Subsequent authors have continued this work and gone on to consider
1
2 P. G. DRAZIN AND L. N. HOWARD

other modifications of inertial instability, such as those due to compress-

ibility of the fluid, to rotation of the system, and to magnetohydrodynamic
effects. There is a wide class of such problems, which have been considered
piecemeal by research workers ranging from sanitary engineers to astro-
physicists. General and particular results by hundreds of authors have
appeared in dozens of journals, and there has been much duplication of
work on the same mathematical problems in different physical contexts.
Our approach will be the fluid dynamical one of studying the phenomenon
of instability rather than its practical applications or natural occurrence.
In this way we shall emphasize the unity of the various problems discussed.
We begin in Section I1 with the fundamental theory of inertial instability
of plane parallel flow of inviscid fluid. Euler’s equations of motion are
linearized with respect to small perturbations of the basic parallel flow.
We first discuss the method of normal modes, whereby it is assumed that
each perturbation can be resolved into dynamically-independent wave
components. A linear eigenvalue problem is then posed to determine the
typical component. Squire’s theorem shows that the most rapidly growing
component in an unstable flow is two-dimensional. Thus, in seeking a
criterion for instability, one may assume that the typical wave component
is two-dimensional, and thereby simplify the eigenvalue problem. The
eigenvalue problem is singular, and the singularity admits solutions with
discontinuous derivatives and a continuous spectrum of eigenvalues in ad-
dition to well-behaved solutions with a discrete spectrum. All these solutions
are necessary to form a complete set to represent an arbitrary initial dis-
turbance. The eigenvalue problem for an inviscid fluid is related to that
for a slightly viscous fluid, though the two are formally independent, this
relation being discussed briefly. Many general properties of the eigenvalue
problem are given, the most notable being Rayleigh’s necessary condition
for instability that the basic velocity profile has a point of inflection. We
describe these properties both mathematically and physically before giving
details of stability characteristics for particular velocity profiles.
In Section 11.3 we discuss the stability problem from the point of view
of the initial-value problem posed by the linearized equations of motion.
Rayleigh’s inflection point theorem is reconsidered, and obtained in a quite
general form. Somewhat more detailed results for particular velocity profiles
can be obtained by taking the Laplace transform with respect to time of
the equations, and in this way the relation between the initial-value problem
and the equivalent normal mode solution is brought out explicitly.
In Section I11 we briefly pose the analogous eigenvalue problems when
various external force fields act on the inertial instability. We also give
the solutions for two important basic flows, those of static equilibrium and
of a vortex sheet. Unbounded disturbances of static equilibrium are
neutrally-stable waves, and those of a vortex sheet are instabilities which
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 3

in some cases may be stabilized by the force field. We add a survey of results
and literature for each force field. The fields we consider are compressibility
in a fluid of variable temperature, buoyancy due to variations of density,
Coriolis force due to rotation of the system in which the parallel flow is
placed, variations of this Coriolis force transverse to the flow, and magneto-
hydrodynamic forces.
The similarity of all these problems is brought out in Section IV. Mech-
anisms of instability are discussed, and analyzed dimensionally to give some
general stability characteristics. Some of these dimensional arguments are
elaborated by physical ones.
Lack of space and time prevent our treating in detail the case of each
force field, so we have picked the single case of buoyancy due to variations
of density for detailed study in Section V. This case is as typical as any,
and has the advantages of practical importance and of advanced theoretical
development. We discuss this case in Section V much as we did inertial
instability in Section 11. Finally, in Section VI, we give some results on non-
planar parallel flows.

11. INERTIALINSTABILITY

I . Eigenvalue Problem for Inertial Modes

The first work on instability of parallel flow seems to be a physical
remark of Helmholtz [I] in 1868, though he and others had studied neutrally-
stable waves previously. In 1871 Kelvin [2] gave a complete analysis of

Y = Y2 &Y”” “/‘”’”’- ”””

I
lb1

FIG. 1. (a) Channel of flow. (b) Velocity profile of basic flow.

the instability of a vortex sheet of inviscid incompressible fluid, allowing

for surface tension and a discontinuity of density at the sheet. Later
Rayleigh [cf. 31 wrote a series of fundamental papers on hydrodynamic
stability, and by the beginning of this century the theory was well formed.
4 P. G. DRAZIN AND L, N. HOWARD

A wide range of problems has been solved since, the theory has been extended
to viscous fluids, and applications of hydrodynamic stability are numerous.
This abundance of work makes a chronological account impractical, so we
develop the subject logically, referring to authors where appropriate. Strictly
speaking, a logical account should begin with the formulation of the problem
for an arbitrary, not necessarily parallel, basic flow. However, since the
vast majority of work in stability theory has been on the parallel flow case
and only a few general results are known otherwise, we shall defer our remarks
on non-parallel flow to Section 11.3, and begin here in the traditional manner.
We consider the stability of a basic plane parallel flow of inviscid fluid
with given velocity
fi+ = (@+(Y+)~O~O) cYi+ <Y+ <YaJ.
This is illustrated in Figure 1. We take the flow bounded by the two planes
y+ = ylS,yaI parallel to the flow. Each of these planes is either rigid or free,
so that either the normal velocity of the fluid is zero or the pressure con-
stant there. However, we allow one or both of the planes to be at infinity.
In this section we suppose the fluid is incompressible and homogeneous.
It is convenient to choose some velocity scale V of the basic flow zi,(y,)
and some length scale L of its transverse variations in order to introduce
dimensionless variables. Thus we define in the usual way the dimensionless
time, position vector, velocity, basic velocity and pressure as
t= t,V/L, r =r+/L u =u,/V, = ii,,/V,
ii = (6,0,0)
P =P*/PVa
respectively, where p is the density of the fluid. Now the Euler equations
of motion can be written
+
aulat U . PU = - vp.
Also the equations of continuity and incompressibility give
P*u=O.
The basic flow satisfies these equations and the boundary conditions
with uniform pressure fi. To study the instability of this flow one puts
+
u(r.4 = i(y) u'(r,t), $(r,t) = p'(r4 +
and neglects quadratic terms in the perturbations, denoted by primes.
Linearizing the equations of motion in this way for small perturbations,
we find
+ +
&#/at aa%#lax virtc2/dy = - apllax,
+
a q a t navllax = - apllay,
awllat + nawi/ax = - api/az,
+
atqax avl/ay awi/az = 0.+
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 6

The stability problem involves the growth of an arbitrary infinitesimal

perturbation, but it has been generally assumed that such perturbations
can be resolved into independent wave-like components. Each component
is supposed to satisfy the linearized equations of motion and boundary
conditions separately. So we consider a typical wave component with

for some functions i, 3, where a,y are real wave-numbers and c = c, ic, +
is a complex velocity. I t is to be understood that real parts are to be taken
to get physical quantities, this being permissible for the linear problem.
Thus each component travels as a wave in the direction of (a,O,y) with
+
phase speed ac,/(a2 y2)1/2and grows or dies away in time like exp (acit).
Therefore the wave is unstable when aci > O and stable when aci Q 0. It
is said to be neutrally stable when aci = 0.
This assumption that any disturbance can be represented by wave
components, according to the method of normal modes, serves to separate
the variables and reduce the linearized equations of motion from partial
to ordinary differential equations. They now become

(2.1) ia(G - c)zi + ddG/dy = - i a j ,

(2.2) ia(n - c)C = - d$/dy,

(2.3) ia(G - c)& = - iy6,

(2.4) ia4 + dd/dy f iy& = 0.

The boundary conditions are that B or fi vanishes on each boundary according
as it is rigid or free. We shall usually consider rigid walls, so that
(2.6) 0 =0 (y = ypy*).
These four equations and the two-point boundary conditions in general
pose an eigenvalue problem to determine an eigenvalue or values c for
<
given a,y,C(y).If aci 0 for every wave-number vector (a,O,y), then the
basic flow G(y) is stable to any wave disturbance, and is said to be stable.
Some care must be taken in this method of normal modes because of the
occurrence of "improper" modes associated with concentrated layers of
vorticity and of the corresponding continuous part of the c-spectrum, as
well as the occurrence of ordinary stable or unstable waves with the discrete
c-spectrum. The singularity of the equations where J = c leads to a contin-
uous spectrum of eigenvalues c, whose eigenfunctions can be found in terms
of generalized functions. These real eigenvalues are in addition to the
discrete spectrum of eigenvalues, which may be real or complex. The eigen-
functions for all these eigenvalues are needed to form a complete set to
represent an arbitrary initial disturbance with bounded vorticity. The
6 P. G. DRAZIN AND L. N. HOWARD

initial disturbance may be represented as an integral of components, some

of which separately have infinite vorticities. The existence of the continuous
spectrum was known to Kayleigh [3, pp. 391-4001 but its importance and
connection with the initial-value problem has been widely appreciated
only recently. We shall discuss this matter more fully in Section 11.3,
and assume the method of normal modes meanwhile.
Squire [4] has proved that for each unstable three-dimensional wave
component there is a more unstable two-dimensional one (i.e. one with
y = 0). His proof runs essentially as follows for inviscid fluid. First define
~2 = a2 + y2, EG = a& +yb, $1=
~ $/a, V' = a, C' = c.
Then equation (2.1) plus the product of y / a and (2.3), and equations (2.2),
(2.4) give

(2.6) i q n - c)a + v'ddldy = - i E j ,

and boundary conditions (2.5) give

(2.9) v' = 0 (r = ypy2).
I t can be seen that the eigenvalue problem (2.6)-(2.9) has the same form
as (2.1)-(2.5) when y = 0, I = 0. Thus C(E) has the same functional form
as c(a) when y = 0. Thus to each three-dimensional wave, growing in
time like exp (a@) for given a,y, there corresponds a two-dimensional wave,
growing like exp ( E C , ~ ) .Now E a > a a if y # 0. Therefore to each unstable
three-dimensiopal wave there is a faster-growing two-dimensional one.
Lin [ti, pp. 3-41 has described this result qualitatively. A three-di-
mensional wave travels in the direction (a,O,y), making angle 0 = tan-' (yla)
with the x-axis. If the coordinate frame is rotated about the y-axis so that
the new Z-axis is in the direction of the wave, then the basic flow has
components
-
ii = ( J ( y )cos O,O, - n(y)sin 0 ) .

The wave now propagates in the %-directionand is independent of Z. Further,

the equations governing 6 , 5. p are independent of g, t3, so that we have
essentially a two-dimensional wave-disturbance of a basic flow (G cos t3,O.O).
Thus the velocity of the basic flow is effectively reduced by the factor cos 8,
and the growth rate of the three-dimensional is less than that of a two-
dimensional disturbance by the same factor.
Henceforth, in seeking a sufficient criterion for instability, we shall
confine our attention to two-dimensional disturbances. Although the
fastest growing small disturbances are two-dimensional, it should not be
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 7

forgotten that three-dimensional disturbances may be of practical sig-

nificance. In particular, three-dimensional effects are known to be important
in determining the nonlinear growth of a disturbance.
With y = 0 = 6 , equation (2.4) may be integrated by use of the stream
function of the disturbance,
(b'(x,y,t)= y,cY)ek(z- dl
such that
Mi = a+'/dy, = - a+l/ax,
i.e.
fi=D~q, 6 = -iag
where D = d/dy. We may also write ti = w ( y ) without ambiguity hence-
forth. Then equation (2.1) gives
(2.10) 6 = yDw - (w - c)Dy.
Now equation (2.2) gives
(2.11) (w - c)(D2- a 2 ) v- (D2w)y= 0 ,
which is in fact the perturbation of the vorticity equation of inviscid fluid
in two-dimensional motion,

Equation (2.11) is called the Ruyleigh stability equation. Its generalization

for viscous fluid is the Orr-Sommerfeld equation,
(Dz - a2)2p,= iaR{(w - c)(DZ- aZ)y- (D%)p},
where H = V L / Y is the Reynolds number of the basic flow of fluid of kine-
matic viscosity Y.
Boundary conditions (2.9) give
(2.12) acp = 0 (r = Y1J4-
The eigenvalue problem comprises the singuIar second-order linear
differential equation (2.11) and the two-point boundary conditions (2.12).
The equation has two independent solutions q ~y2 ~ which
, are analytic func-
tions of y , ore, c over domains in which the equation is non-singular through-
out the field of flow, i.e. over domains in which c lies outside the range of
<
W ( Y ) for Yr y <Ye. Thus

'p = Aly,Cy ;a2,c) + A,y2(r ; a 2 4

8 P. G. DRAZIN AND L. N. HOWARD

for some complex constants A,, A,. Substitution into boundary conditions
(2.12) and elimination of A , , A , gives the eigenvalue relation

In general this can be inverted to give a many-valued function c = c(a2),

continuous in any domain excluding the range of w .
If w or Dw is discontinuous, a t yo say, then the pressure must be continuous
a t the material fluid interface with mean position y = y o . Therefore, to
first degree in the small perturbation,
(2.13) [ ( w- c ) D -
~ ( D w ) ~=] 0
a t y = yo, where square brackets here and henceforth denote the “jump”
of their contents at a possible discontinuity. Also the normal velocity of
the fluid must be continuous a t the material interface. Let this interface
have equation
y =yo +q(x,t), where q = -ct).

Then
V’ = Dq/Dt = aq/at + waq/ax = ia(w - c)V,
to first degree in the perturbation. But v = iav, so it follows that
(2.14) [d(W - 41 = 0
at y = yo.
Conditions (2.13), (2.14) can alternatively be proved mathematically.
We may take yo = 0 without loss of generality. Then stability equation
(2.11) has integral

[ ( w - c ) D v - ( D W ) ~ ]=? a2
1
~ (w - c)pldy.
-8

On taking the limit E 0

--+ +,
condition (2.13) follows. Division by (w - c ) 2
and further integration gives

[ d ( w - 4Ih. = a2
i
-e
dy(w -
-e idyl(wl - c ) ~ , ,

and thence condition (2.14) as E -0 +.

These ideas may be extended to
generalized functions w ( y ) , functions which do not represent real flows but
do approximate some properties of real flows and give sensible stability
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 9

characteristics easily. For example, when w * ( y ) = d(y - yo), the Dirac

delta function, we get from the above
(2.15) c2 [OF]
= a 8 (F)y,y,, [ F ]= 0
at y = yo, where F = V / ( w - c), a function which has no worse behavior
than a discontinuous derivative at y = yw
Explicit solutions of the eigenvalue problem are difficult to find in
practice for smoothly-varying functions w ( y ) . However, when w ( y ) is a
piecewise linear function, one can find exponential solutions of the stability
equation piecewise, and join them up by use of conditions (2.13), (2.14).
This method, first used by Kelvin, has led to explicit eigenvalue relations
for many problems. This use of profiles which do not vary smoothly can be
justified [6; 7, p. 221 ; 81 sometimes as an approximation to similar smoothly-
varying profiles when the wave-number is small.
Note the symmetry of the stability equation and boundary conditions
in a and I- a). So we shall henceforth take a 0 without loss of
generality. Then the criterion for instability of the basic flow is that there
be a solution with ci > O for some a >O.
Note further that for each eigenfunction q, with eigenvalue c. for given
a there is another complex conjugate eigenfunction v* with eigenvalue
c* = c, - ici for the same a. Thus to each damped stable wave. there
corresponds an amplified unstable wave, and vice versa. This expresses the
time symmetry of the problem, comprising periodic motion of inviscid
fluid with steady boundaries. It follows that the condition for stability
is that c is real, and for instability that c is complex. So we shall write
c, 3 0 when there is instability and ignore the associated conjugate eigen-
value, though it should be remembered that on the inviscid theory the
result ci < 0 equally implies instability.
We emphasize that we are entirely concerned with hydrodynamic stability
of inviscid fluid. Since the Orr-Sommerfeld equation is not invariant under
complex conjugation like the Rayleigh stability equation, stability for
viscous fluid does not necessarily correspond to real c. Indeed, the relation
to the solutions of the Orr-Sommerfeld equation is complicated. For this
we refer to Lin’s book [9, Chap. 81, where it is shown that at least for analytic
w ( y ) a solution of the Rayleigh equation for ci > 0 is a limit of some solution
of the Orr-Sommerfeld equation, though its complex conjugate may not be,
throughout the domain of flow. Further, when solution of the Rayleigh
equation gives a stable basic flow of inviscid fluid, solution of the Orr-
Sommerfeld equation may give instability of the same flow of viscous fluid,
in accord with Heisenberg’s criterion. These and other subtle mathematical
questions raised by the inclusion of viscosity are important in some cases,
but are not considered here, so our results must be taken with this in mind.
Nevertheless, a full understanding of the inviscid theory is a desirable prelim-
inary to any study of the viscous theory.
10 P. G. DRAZIN AND L. N. HOWARD

The fundamentals of this section are due to Rayleigh [3]. For more
recent accounts, the book of Lin [9, Chaps. 4,8] and chapter of Stuart [lo]
are recommended. These two works, and Reid's survey [Ill, also place
the inviscid theory in its viscous context.

2 . General Stability Characteristics of Plane Parallel Flow

Rayleigh [12] proved in 1880 that a necessary condition for instability
is that the velocity profile w ( y ) should have a point of inflection. This can
be proved by multiplying the Rayleigh stability equation (2.11) by Q* and
integrating from y1 to y8. Thus, after use of the boundary conditions (2.12)

(2.16)

The imaginary part of this equation gives

(2.17)

If c, > O it now follows that Dew must change sign at one or more points
in the field of flow. On assuming that Dew is continuous, there must be at
least one inflection point on (y l ,y2)and indeed an inflection point a t which
the velocity profile crosses its tangent, i.e. a relative maximum or minimum
in the basic oorticity i3 = - Dw. With the (weaker) assumption that i3
is only piecewise continuously differentiable (which has really been tacitly
made anyway in writing down the stability equation) we can still say that
6 must have a relative maximum or minimum.
In 1960 Fjmtoft [13, p. 261 proved the stronger necessary condition
for instability that (D%)(w - tos)< 0 somewhere in the field of flow, where
ys is a point at which D f vanishes, and where w, = w(y,). A proof comes
from the real part of equation (2.16),

Adding

J
YI
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 11

to the left-hand side, we get

The result follows. In particular, if w ( y ) is a monotone function and D8w

vanishes only once in the field of flow, a necessary condition for instability
<
is that (D%)(w - w,) 0 throughout the flow, with equality only where
y = ys; this result is depicted in Figure 2.
Y Y

FIG2. (a) Stable: D*w < 0. (b) Stable: Dgw> 0. (c) Stable: D’w = 0 at ys, but Dw in-
creases where w < ws. (d) Possibly unstable: D*w = 0 at ys and (w - ws)Dsw< 0 elsewhere.

Fj~rtoft’sextension of Rayleigh’s theorem can be shown to be equivalent

to the statement: If 6 is piecewise continuously differentiable, a necessary
condition for instability is that 11 should somewhere have a maximecm
6
inside the flow domain.
Neither condition for instability is sufficient in general. We shall present
a counter-example (c) to sufficiency with w = sin y in Section 11.4. How-
ever, Tollmien [14] proved sufficiency in 1935 for symmetric velocity
profiles in a channel and for boundary layers. The basis of Tollmien’s ar-
gument is first to show that there exists a neutrally stable eigensolution
with c = ws, and then to construct unstable solutions for which c -c w s
as ci + 0 through positive values.
12 P. G. DRAZIN AND L. N. HOWARD

Friedrichs [16] has given an elegant alternative proof of the existence

of the neutrally stable eigensolution,
Q =ips. a =a,>O. c = ws,

say. For the proof we suppose that K ( y ) = - D 2 w / ( w- w,) is integrable

over the field of flow, and put c = w,, A = - a2 in the Rayleigh stability
equation to get
DZp, + K(y)p,+ Ip, = 0,
a real non-singular equation which makes up a Stunn-Liouville problem
with boundary condition (2.12). The associated variational principle gives
the least eigenvalue

(2.18)

the minimum being for functions f that vanish a t the walls and have square-
integrable derivatives. Therefore a neutral eigensolution with positive
a = a, = (- As)1'2 exists if and only if I, < 0. There may be a finite number
of other eigensolutions for larger eigenvalues I, provided these are negative.
Also there may be other series of eigensolutions when c = w, for other
values of w, a t other points of inflection, and sometimes eigenfunctions with
real c not equal to the value of w at an inflection point, though these have
slightly singular behavior.
The existence of the neutral eigensolution with I, < 0 follows easily
when K ( y ) > n 2 / ( y 2- y J 2 everywhere on account of the well-known in-
equality,

i
Yn

(Y2 -YP! (war2 nz PdY.

Yl YI

Again, when K ( y ) > O over the field of flow and w vanishes a t the walls
but not between, trial of / = w shows that A,< 0.
Tollmien also demonstrated heuristically the existence of unstable waves,
whose limit as ci + 0 is the neutral s-wave above. This has been considered
alternatively by Lin [7, pp. 223-224; 9, pp. 122-1231 as follows. The
stability equation for Q, gives
(2.19) D*p,, + {I, - (D2w)/(w- w,)}p,, = 0.
Multiplying this by p, and subtracting vS times the equation for Q,we get
D(Q& - f@pls) - ( A - Is)Q% - ( o w { ( w - c)-l - (w - ws)-'}plqJ, = 0.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 13

Integrating from y1 to ya, we now find

YI I1

-
To find the unstable solutions near the neutral one, we take the limits
a +as, c w,, q +vs. Then

as ci -+ 0 through positive values, provided (Dw),= ys # 0, where 9’denotes

the principal value of the integral. Therefore
Y:

(w - wJ2
Yl Y.
(2.20)

For a known neutral eigensolution this gives eigenvalues c(a)near wsin the
complex c-plane and thence the associated eigenfunctions in the limit
ci -+ 0 f . If K ( y ) > 0 throughout the field of flow, then the imaginary
part of the denominator is positive, and it follows that (dci/daa)a=. as < 0,
with instability for a just less than a,.
Further, it can be proved as follows that there is instability only when
a < as. Suppose K ( y ) > O throughout the field of flow. Then, when cj # 0,
the real part of equation (2.16) plus (w, - c,)/ci times equation (2.17)
gives
14 P. G. DRAZIN AND L. N. HOWARD

Therefore

I t follows that there is stability (ci = 0) when a a,.

This argument can be extended to prove the following result, applicable
to flows for which the function K exists and is non-negative [16]. Let
&,A2,. . . be the eigenvalues of the Sturm-Liouville problem f" Kf + +
Af = 0, f(yJ = f(y2)= 0, arranged in increasing order. Then there can
be no more than rt - 1 linearly independent unstable eigenfunctions of the
stability problem if az 2 - A,,, Thus if a2 is larger than the absolute value
of the lowest (negative) eigenvalue Al, the flow is stable; if az lies between
- jlz and - A, there can be only one unstable mode, if it lies between - A:,
and - Lz,at most two, and so on. Of course eventually A,, becomes positive
and thereafter the relevance to the stability problem ceases. In particular
if A, 2 0, which may happen even for K 0 if the boundaries are suffi-
ciently close, then the flow must be stable even though it has an inflection
point with w f f ( w - w8)< 0. (For a different and interesting approach to
a related result in the case of a monotone w( y ) , see [17].)
We shall next consider the energy of a disturbance. If one multiplies
the Rayleigh stability equation by y*, integrates from y1 to yz, and uses
the boundary conditions, one finds

5
Ys

YI
(w - c)((DylS + a2(ollS) + ( D W ( q I 2- y * ( D 4 ( D q ) d y= 0.
The imaginaiy part of this gives

(2.21) ci l
YS

YI
+
~Dvla a2lqlady =
l !
(Dw)(vDv*- pl*Dq)dy.

This is in fact the x-average of the energy equation of the disturbance,

YI Y1

Foote and Lin [18] noted that the average of the Reynolds shear stress
over a wavelength

5
errla
1
= - &If = - (a/2n) u'vfdx = -ia(qDy* - y*Drp)e*it.
4
n
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 16

The stability equation now gives

Since t = 0 a t the boundaries, the integral of D z over (yr,y2) is zero:

this is in fact just the integral (2.17) used to prove Rayleigh’s theorem.
But consider now a neutrally stable mode which is adjacent to unstable

-.
modes, that is, suppose we have an unstable mode for some a, and as a
say, the corresponding c approaches a real value: ci(a) 0. Then for this
-.
as,

neutral mode, (2.22) shows that D t must be zero everywhere, except possibly
at y = y,(w(y,) = c) where the limit of the right-hand side of (2.22) does
not exist. By consideration of the nature of the singularity in D t which
appears as c, --r 0 + it can be shown that at the “critical layer” y = yc (or
layers) z has (in the limit ci --+ 0 -1) a jump [t],of magnitude

Because of the boundary conditions satisfied by t, the algebraic sum of all

such jumps must be zero. If the profile is monotone, there can be only one
jump, which must thus be zero, and this implies that D% vanishes at yc
since it can be shown that the alternative v(yc)= 0 is impossible. Thus
for monotone profiles the neutral value of c must be the value of w at the
inflection point. This is true also for some non-monotone profiles, for example
the symmetrical jet w = sech2y, but is not always the case. For many
non-monotone profiles, notably most non-symmetrical jets, there is no
possibility that D2w = 0 at all places where w = c. A neutral c adjacent
to unstable modes doubtless exists in such cases, but it is not the value of
w at any inflection point, and the corresponding eigenfunction must exhibit
a certain weak singular behavior so that t can have two compensating
jumps. This must in fact be regarded as the typical case for non-monotone
profiles; it is fortunate that many profiles of interest are either monotone
or sufficiently symmetrical so that the neutral c can be identified at once
as the value of w a t the inflection point. In general both the neutral c (ad-
jacent to unstable modes) and the corresponding a have to be determined
by numerical solution of the equation taking proper account of the sin-
gularities at the critical layers. Some further discussion .of the neutral
eigenfunctions is given in Section 11.3.
In 1915 Taylor [19, pp. 23-26] gave a physical interpretation of Ray-
leigh’s necessary condition for instability. Taylor noted how the frictionless
slipping of the fluid a t the boundaries prevented the transfer of x-momen-
turn necessary to maintain an unstable disturbance when D% is always of
one sign. Essentially the momentum is transferred by the Reynolds stress,
which must vanish near the walls and whose gradient can only vanish at
a point where D2w vanishes. (Lighthill [20] has applied these physical
16 P. G . DRAZIN A N D L. N. HOWARD

ideas to the instability of wind whereby ocean waves are generated.) Taylor
went on to note that viscosity allowed momentum to be diffused from the
boundaries, and suggested that a given basic flow might thus be stable for
inviscid but unstable for viscous fluid. This suggestion has since been veri-
fied, for plane Poiseuille flow as an example.
Lin [7, pp. 226-2271 also has interpreted physically the mechanism of
inertial instability by consideration of the migration of vorticity. He regarded
the flow due to a neutrally-stable disturbance in Kelvin’s “cat’s-eye” di-

FIG. 3. Kelvin’s cat‘s-eye diagram. The streamlines viewed by an observer moving

with the neutral wave.

agram [21], the pattern of streamlines viewed by an observer moving with

the phase velocity c of the wave (Figure 3). This observer sees a stationary
Y
flow, with 4= wdy and 4’ = v(y)eiax. (It should be remembered that
YC

the physical quantity 4’ is understood to be the real part of its complex

representation.) Let us assume that the critical layer y = y c lies within
the field of flow, and that cp does not vanish in that layer. Then there will
+
be some closed streamlines, and the streamline $ = 4 +’ = 0 will intersect
at points on y = yc periodically separated by 2n/a. Now the flow is inviscid
and two-dimensional. Therefore the total vorticity w = d w’ = - Dw+
- au‘/ay + av’lax is uniform on each streamline, and in particular
on the intersecting streamline. But aw/ay = 0 at the points of intersection.
Therefore aw/ay = O(1o‘I) at the critical layer, i.e. (D2w),=yc = 0 to zeroth
degree in the perturbed quantities, or aw‘/ay is singular. It follows that
it is possible to find a non-singular neutral disturbance in inviscid fluid only
if D2w = 0 where w = c. In reality a singular disturbance with large
a21u’I-law’/ay would be damped by viscosity. Lin [7; 9, pp. 66-68] has
also gone on to discuss the two-dimensional motion of vortices during in-
HYDRODYNAMIC STABILITY O F PARALLEL FLOW OF INVISCID FLUID 17

stability, and a more complete discussion of this physical mechanism has

been given by Gill [22].
Rayleigh limited the possible range of eigenvalues in the complex c-
plane, proving that wmin< c, < wma, when ci # 0. Howard [23] generalized
this result with his semicircle theorem. For its proof, suppose F = v / ( w - c)
is non-singular, and rewrite the Rayleigh stability equation as
(2.23) D{(w - C ) ~ D F-} a9(w - c)*F = 0.
Multiply this equation by F* and integrate from y1 to ye, using the conditions
that F vanishes on the boundaries. Then
Y,

(2.24) +
(w - c)s{1DFl2 aalFla)dy = 0.
YI

This equation implies that c cannot be real when F is non-singular and

therefore that c cannot lie beyond the range of w. Next suppose ci # 0 and
take the real and imaginary parts of (2.24). This gives

1
YI

- c,)' - ci2}Qdy = 0, 2ci (w - c,)Qdy = 0,

YI YI

where Q G lDFla + aalFI2> O . Therefore

But
j.1
YI
UQdy = &dy,
YI r j .
Y,
dQdy =
Y1
(Q' + c,')Qdy.

0 2
1 YI
(W - wmin)(w - wmax)Qdy

-j
- { (ci2
YI
+ G') - (wmin + wmax)C, + wminwmax}Qdy#
the maximum and minimum being taken over the field of flow yl < y < y2.
Therefore
cia + cva - (wmin + ~max)cr+ w&wmx <0,
i.e. for unstable waves c lies in the semicircle
(2.26) {c, - +(wmin + wmax)}' + ci2 < {i(wmax - wmin))' (ci> 0)-
18 P. G. DRAZIN AND L. N. HOWARD

This shows that any eigenvalue c, real or complex, must lie in or on the
circle with center t(w,, +
wmin) and radius i(wmx - wmin).
Again, with G = q/(w. - c)Y2, the stability equation can be written
(2.26) D((w - c)DG} - (40% + U'(W - C) + )(Dw)*/(w- c)}G = 0.
This has an integral

j.
Y.
+ aalG12}+ &(D2w)IG/'+ ) ( D W ) -~ (c+)lC/(w
(w - c)(lDGI2 ~ -c)l*}dy=O,

whose imaginary part gives

because lw - C I-~ = {(to - c,)% + < ci-'. It follows that

aci < amax IDwl.
This result is due to Hailand [24, p. 111, this proof to Howard [23]. A more
general analogue will be given in Section 11.3.
The stability problem has certain symmetries when the basic flow is
symmetric, i.e. when it is possible to choose coordinate axes so that y1 = - ya
and w ( y ) is an even function. In that event, if ~ ( y is ) an eigenfunction
with eigenvalue c for any given a, it follows that the even part

9 8 = Hvpcy) + d- Y))
and the odd part

vo = H d Y ) - 9-4- Y ) }
of v are also eigenfunctions for the same c, a. This can be seen a t once from
the symmetric stability equation and boundary conditions. It can be shown
further that either v0 or 9' is identically zero. To show this, we multiply
the stability equation (2.11) for pe by yo, and subtract ye times the same
equation for vo. This gives

vPv0 -v p v . =Q
where w # c. Therefore
@hpo - q&pS = constant = value at wail

= 0.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW O F INVISCID FZUID 19

Therefore (p',~,, are linearly dependent in general, which is only possible

if one of them is identically zero. Thus we have proved that an eigenfunction
is either odd or even, except possibly when c is real.
In fact both even and odd eigenfunctions are found for the same symmetric
basic flow, but they have different eigenvalues c for each a. An even eigen-
function is associated with a disturbance named sirruous by Rayleigh, the
pattern of streamlines being antisymmetric about the line y = 0. Similarly,
an odd eigenfunction is associated with a varicose disturbance, the stream-
lines being symmetric. This oddness or evenness of 9 allows one to assume
that (p is even (or odd) and reduce the effective field of flow to the half
< <
range, 0 y ya. applying the symmetry condition Dg, = 0 (or Q = 0 )
at y = 0 and the original boundary condition v = 0 at y = ya. This is a
convenient method to find eigenvalues for the sinuous and varicose modes
of instability. It can be seen from the variational principle (2.18) with
even K ( y ) that the least eigenvalue 1 corresponds to an even function f ,
and that therefore the first sinuous submode is more unstable than any
varicose mode of a given basic flow.
Next we suppose that the profile is antisymmetric, with y1 = - y o
and odd w ( y ) . Then for each eigenfunction ~ ( ywith ) eigenvalue c there
is an eigenfunction y * ( - y) with value - c* = - c, + icj for the same a.
When the eigensolution is unique, this Hermitian symmetry implies that
c, = 0 and g,*(- y ) = (p(y). Otherwise, there may be a pair of eigen-
solutions with phase velocities f c,(a) and the same ci(a),one function the
Hermitian conjugate of the other. Howard [26] gave a physical argument
for a situation when the latter must occur. In Section 11.4.1 we give an
example of a discontinuous shear layer for which it occurs. We also
know that it may occur for the s-eigensolution at least when the profile
has a point of inflection, other than that at y = 0. where w, # 0. A t any
rate for the neutral eigensolution with yo = 0. K ( y ) is an even function.
and the variational principle (2.18) gives the greatest wave-number a, for
an even eigenfunction (ps; Lin's argument to deduce the perturbation formula
(2.20) gives (dc,/daa)a-as-o= 0. So one might conjecture that for this mode
associated with the point of inflection y , = 0 there is exchange of stabilities
<
such that c, = 0 when cj # 0, i.e. when 0 a< a,. This can in fact be
proved for monotone antisymmetric profiles with K ( y ) >O; cf. [16].
Let us now revert to general basic flows, not necessarily with any sym-
metry. Equation (2.24) was derived on the assumption that F = v / ( w - c)
had a square-integrable derivative over the interval [yl,yo]. I t shows that,
when c is real and F not identically zero, c lies in the range of w and either
(i) a = 0, F = constant = A, say, or (ii) DF is not square-integrable. In
the latter case, Q itself might be singular where w = c or it might have a
lower order zero than (w - c). Our previous work now shows that as ci 0 +
+
-+

either a - 0 or a + a , -.
20 P. G . DRAZIN AND L. N. HOWARD

If a = 0 and F = A, we get the trivial dgensolutim with $' = Q =

A(w - c). This is really a form of the basic flow, for the total x-component
of the velocity of the perturbed flow is

and the y-component is ZI = - a#/& = 0. Thus the trivial solution is

really the basic solution displaced laterally by the small distance A. In
fact, for any solution it is readily seen that the vertical displacement at
(x.t) of the material surface with mean level y is

q(x,t) = F(y)eiacX- c t ) ,

The trivial solution appears as the first term in a power series expansion
of Q for small a. Heisenberg [S] found formally two solutions of the Ray-
leigh stability equation :

+
q+(y;a*,c)= (w - c ) { q i d y , c ) aaq,l(y,c)+ . . . + a*q~,d.y,c)+ . . .} (i= 1,2)
where

qlo(y',c)= 1, qeo(y,c)=

In these formulae the lower limit of integration is arbitrary, but may con-
veniently be taken as y l . The zeroth approximation for small a2 gives the
eigenvalue relation

(2.27)
1
YI
(w - c)-%y = 0.

However this result depends only on heuristic analysis and is equivocal

[7, pp. 220-2211. In fact Heisenberg used the series chiefly for the viscous
solution at high Reynolds numbers.
Heisenberg was not concerned with the case of one infinite boundary,
for which it can be seen that his series are not uniformly convergent. On
taking the limit as a + 0 for fixed w b ) , the stability equation in the form
(2.23) gives

(w - c)*DF = constant = value at boundary = 0,

HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 21

and therefore F is constant between critical layers y = yE. Therefore F = 0

from infinity down to the largest value of yc. On the other hand, the stability
equation and boundary condition (2.12) at infinity give

F-constant x e-=Y as y +w

for any positive a, however small, provided that w .+ constant smoothly.

Thus the order of the limits a + O , y w cannot be changed without
--+

changing the limit of the eigenfunction F, which, like c a y , does not admit a
power series expansion in aB uniformly for y large.
In 1962 Drazin and Howard [S] considered long-wave disturbances of
unbounded and semi-bounded flows. The basis of their work for unbounded
flows is as follows. The stability equation (2.23)has two solutions F, (y;a,c)
defined by their asymptotic properties
F, WeFay, DF, N Fa e F a Y as y+ f 00.
These solutions are defined by the stability equation for given a,c,w(y),
and are in general independent. However, for an eigenfunction which
vanishes at y = f w,
F E K+F+ K-F- (- W < Y < W)

for some complex constants K,, the solutions F, being linearly dependent
when c is an eigenvalue corresponding to given a. Therefore the Wronskian

(2.28) F+DF- - F-DF.+ = 0

at each and every point y , and at y = 0 in particular. This is the exact

eigenvalue relation. At this stage one may assume a is small and seek to
expand F, as power series in a. To avoid the non-uniformity of convergence
at y = & w we put
m

F, = eTaY 2 (& a)"X+,,(y,c)

r=o
(0 < fy < 4,
the two series being used in semi-infinite intervals which just overlap at
y = 0. The coefficientsx, ,,can be found formally from the stability equation
as repeated integrals of w h ) , c etc. Now the eigenvalue relation (2.28) can
be expanded in powers of a, the coefficients involving c, w ( y ) in explicit
integrals. This method can be shown to give one mode for which

(2.29) c + Q { w ( w ) + w ( - a)}+ i i l w ( w ) - w ( - w)I as a -0.

Thus there is instability when w(- w) # w ( w ) , or the flow is of shew-

kryc* or halfjet type. On the large scale of the long wave (with small a)
22 P. G. DRAZIN AND L. N. HOWARD

a general smoothly-varying shear layer behaves like the vortex sheet with
basic velocity

Indeed the limit (2.29)of c gives the exact result for this vortex sheet, as is
given in Section II.4.d. In fact there is also instability (but of smaller
growth) when w ( - 00) = ~ ( o o ) ,i.e. when the flow is of jet by@, the next
approximation then giving

Other modes were also considered 181, it being found that

as a 4 0 , where

at y = y,,,.
In discussing the eigenvalue relation for a general profile, bounded or
unbounded, we mentioned that c(a) may be a many-valued function. The
variational principle (2.18)suggested that there might be many values
of a for each c = ws and many values of c = w,. For symmetric profiles
we mentioned sinuous and varicose disturbances, when c is at least double
valued. We shall meet many-valued c in several examples of the Section 11.4,
finding that each branch of c(a) is well behaved and corresponds to a distinct
mode of instability. By continuity in a one might expect that for each
neutral eigensolution with c = ws, and therefore for approximately each
point of inflection, there is one mode of instability. Drazin and Howard [8]
considered unbounded flows and associated heuristically the neutral eigen-
solutions having c = w, for each zero ys of D'Jw with the small-a eigen-
solutions having c = w(y,,,) for each zero y,,, of Dw. However the general
problem defies oversimplification, and the modes have not been satis-
fact orily classified.

3. The Initial-Value Problem, and the Stability of Non-parallel Flow

Hydrodynamic stability theory is by far most highly developed for the
case of a parallel basic flow, but there are a few more general results and we
insert here a brief description of some of these. We shall also in this subsec-
tion regard the problem as an initial-value problem, though elsewhere
in this article we generally follow the more usual normal mode approach.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 23

In the parallel flow case, the flow domain is taken as the strip
- do < <
< x < do, y1 y y,, and the boundary conditions are usually
taken to correspond to rigid walls at y = y1 and y,. In the general case,
it seems appropriate to take the flow domain to be some region R in the
plane or in space, and to allow the possibility of flow across the boundary
B of R. In this case, more attention has to be given to the boundary con-
ditions. We shall assume that R is fixed and that we have given in R a basic
flow U, which is steady, incompressible, and inviscid. We shall also
not consider any body forces. The stability problem is then formulated as
follows: given some initial conditions which differ from U, by a small
amount, in terms of some appropriate measure (e.g., the L , norm of the
difference), we find the time dependent flow U determined by these initial
conditions and some suitable boundary conditions which are satisfied by
U,. The flow is stable if U continues to differ from U, by a small amount
in terms of the selected measure. By “suitable boundary conditions” we
mean such as assure that the initial-value problem for the flow equations
will have a solution, and a unique one. The mathematical questions of
existence and uniqueness of the initial-value problem for the inviscid flow
equations with various boundary conditions do not appear to have received
as much study as they deserve, but it is not appropriate here to embark
on such a discussion. We shall give only some brief heuristic remarks. The
most familiar case is that the boundary B consists entirely of a rigid wall,
so that Us n = 0 on it. This condition probably is, in itself, a “suitable
boundary condition” in the above sense. More generally, if the flow crosses
B , we should expect to prescribe U n as a boundary condition on B , sub-
ject only to the requirement that its integral over B should vanish (other-
wise the cohtinuity equation alone would have no solution). However,
this condition alone is in general not sufficient to insure uniqueness of the
solution to the initial-value problem-consider for instance the plane
flows in the annulus 1 r < < 2 (polar coordinates) whose radial velocity
component is l / r and whose azimuthal component is ( l / r ) f ( r -
2 2t) where
f is a function which is zero for values of its argument 2 1, but is otherwise
arbitrary (it may be as smooth as desired). It is easily checked that these
velocity fields do satisfy the flow equations, they all have the same normal
component on the boundaries of the annulus, and are identical at t = 0.
In the example just given (two-dimensional and axisymmetric flow in
an annulus), uniqueness can be insured by prescribing, in addition to the
normal velocity component on the complete boundary, the tangential com-
ponent on the part (r = 1) of the boundary through which the fluid enters
the flow region, for all t > O . I t is clear that this additional information
is just sufficient to determine the function f . Note also that this example
shows that one may not in general prescribe the tangential component
where the fluid leaves the region (r = 2 ) .
24 P. G. DRAZIN AND L. N. HOWARD

We shall now show that these boundary conditions, namely the pres-
cription of the normal velocity component over the complete boundary
(in a manner consistent with the continuity equation) and of the tangential
velocity components over that part of the boundary where the flow is inward,
together with one additional assumption, are sufficient to insure the unique-
ness of the solution to the initial-value problem for general three-dimen-
sional incompressible inviscid flow.
Let U, be such a flow, in a region R with boundary B, and let
+
U = U, u be another. We assume that U and U, satisfy the same bound-
ary conditions, so that u * n = 0 on B and u = 0 on that part B, of B on
.
which U, n < 0. Write B, for the rest of B , on which U, n 2 0. Consider
+
now the deformation tensor D of U, , with components Dij = U&,j U , , .
<
Since Djj has zero trace, at least one of its eigenvalues is 0; let C(t) be
the supremum over R of the absolute value of the most negative eigenvalue
of D-we call C(t) the maximum shear of U,, and we assume that C(t) is
+
finite, initially and thereafter. Writing for the difference of the pressure
fields of the flows U and U, , divided by density, one obtains the following
equation for the “perturbation” u by subtracting the momentum equation
for U, from that for IT:
(2.31) Ut + U-Vu + U* m, + v+ = 0.
Note that the “perturbation” u is not necessarily small. Multiplying (2.31)
by u and using V oU = 0 we get

Integrating this over R , applying the divergence theorem to the last term,
and using the boundary conditions, we get

Write E(t) J )luladV; we call this the “energy of the perturbation,”

R
and shall use it as a measure of how large the perturbation is. Since U,. n
<
2 0 on B, and - u - Do u C(t)Iu),, (2.33) gives

(2.34)

and thus
t

(2.36)
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 26

Inequality (2.36) shows that if the perturbation is initially zero, i.e. if U

satisfies the same initial conditions as U,, , then it remains zero, and so the
solution to the initial value problem is unique.
(2.36) can be applied in another way. Suppose U, is a steady flow, whose
stability we are studying. In this case C(t) is constant and we see that the
growth of the energy of any perturbation is limited by an exponential with
growth rate C, and this is true not only for the initial growth in the range
of the linear stability theory, but also for any subsequent non-linear
development of the initial perturbation. In the particular case of parallel
plane flow with velocity profile w ( y ) , the quantity C is easily seen to be
max IDwJ;for a linear perturbation with exponential growth rate aci, the
energy E (per wave-length, say) has exponential growth rate 2ac,, and the
<
result given by (2.36) thus reduces to Hpriland’s estimate ac, 4 max IDwl
given in Section 11.2.
Though we do not have the existence theorem that ideally should ac-
company it, this uniqueness theorem suggests rather strongly that the boundary
conditions of prescribed U - n over all of B and prescribed U over the “in-
coming” part of B are “suitable boundary conditions” in the sense of the
formulation of the stability problem given above. However, this is not to
say that these are the only suitable boundary conditions; in particular,
it is probable that instead of giving the tangential velocity components on
the incoming part of B one might equally well prescribe instead the tan-
gential components of the vorticity vector there. This becomes particularly
clear in the case of plane flow. Formulating the problem in terms of the
stream function Y(U = W x k), and eliminating the pressure by going
over to the vorticity equation we have the pair of equations:
(2.36) AY+$.R=O
(2.37) Rt + u *VrR = 0.

Now one might imagine the following step-by-step process (similar to

one used in numerical weather forecasting) for computing the solution to
the initial-value problem: Using the initial values of the velocity field for
U,integrate the first-order equation (2.37) to find Q at a slightly later time.
It is clear from the structure of (2.37), which says that the vorticity field
moves with the fluid particles, that what is needed to do this is the initial
values of R in R (which follow from those of U), plzls the values of R carried
by the new fluid particles which enter the region. Having found the vorticity
at the slightly later time, we then calculate the new flow field by solving
the Poisson equation (2.36); to do this, we need a boundary condition,
and the most natural one is to prescribe Y on B , which is equivalent to
giving U. n on B . Thus this hypothetical computation scheme suggests
quite definitely that suitable boundary conditions, with which one might
expect to be able to prove existence and uniqueness of the solution of the
26 P. G . DRAZIN AND L. N. HOWARD

initial-value problem for equations (2.36) and (2.37), are the prescription
of Y (or U n) on B , and of R on B,. In fact, with these boundary conditions
and the assumptions that (U,(and R, remain finite, one can prove unique-
ness by a method similar to that used above for the case of the tangential
velocity components being given on B,.
We now consider the stability problem for plane flow, using this vorticity
boundary condition, to establish a result which may be regarded as giving
a generalization to non-parallel plane flow of Rayleigh’s inflection point
theorem. Let the basic steady flow be U, = vYo x k. Its vorticity 51,
= - AY, is constant along streamlines, and we shall write Ro = /(Yo),
though in some cases such a representation is not literally possible unless
f is regarded as multiple valued-different streamlines might carry the
same value of Yobut different values of a,. We are going to prove that
if f’(Y,)< 0 throughout the field of flow, then the flow is stable to two-
dimensional disturbances. As with Rayleigh’s theorem, this will be a suffi-
cient, but not necessary, condition for stability. The perturbation momen-
tum equation is:
(2.38) U$ + Qk x u + wk x U, + Vh = 0,
where h is the perturbation H - H , of the total head H = ilU21 + P.
From (2.38) we deduce:

(2.39) -”(’
atZ Iul2) + w u * (k x U,) + (uh) = 0.

The -perturbation vorticity equation is

(2.40) W# + u. v w + U * VR, = 0,
and from this we get:

(2.41)

Now
WU. VQ, = f’(Yo)wuoW, = f’w(u x k) (W,
x k) = /’COU* (k x U,).
If we now assume f’< 0 we can rewrite (2.41), using the fact that f’ is in-
dependent of t and constant along the basic streamlines, as:

Subtracting (2.42) from (2.39) and integrating over the flow region R , using
the divergence theorem and the boundary conditions U. n = 0 on B and
w = 0 on B, we get:
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 27

So far we have not assumed that the perturbation is small. We now

note that the second integral on the left in (2.43) is of third order in the
perturbation, while the remaining terms are of second order. Thus within
the framework of the linear stability theory, this second integral is to be
dropped relative to the other terms. Since U,= n 2 0 on 3, and f' < 0 we
thus obtain, for the linear stability theory :

(2.44) <
0 2 / f ' ] d A 0.

Since /' < 0, this integral is positive definite, and it follows that the energy
of the perturbation, though it may possibly increase somewhat over its
initial value at the expense of the term J - w2/2f'dA,must remain bounded;
R
thus the flow is stable. It is interesting to note that the restriction to the
linear stability theory is not necessary for the (rather special) class of basic
flows which have f' constant and negative.
The relation of this result to Rayleigh's theorem is easy to see. For a
parallel flow with velocity profile .I&) we have

If there are no inflection points, w" is of one sign, say w f f >O. By adding
a suitable uniform translation if necessary which obviously does not affect
the stability properties of a parallel flow, we can assume that w >0 through-
out, and so f'c 0 and the flow is stable. (If w"< 0 we can add a suitable
uniform translation so that w < 0; but if w" changes sign this is not possible.)
If there is just one inflection point we can assume that w = 0 there, and
our result thus implies stability if w"/w >O; we thus also obtain Fjertoft's
extension of Rayleigh's theorem.
The above argument is not applicable if f' E 0, i.e. for constant vor-
ticity (in particular irrotational) flows. However such flows, like their
parallel prototype the plane Couette flow, are always stable, a t least with
the boundary conditions we have assumed here. For when VQo = 0, the
vorticity equation (2.40) shows that the perturbation vorticity w is constant
following particles. Since no new perturbation vorticity is brought in by
entering fluid particles, w cannot grow, and since the perturbation stream
function is determined from the Poisson equation A$ +
w = 0 with t,4 = 0
on B , 4 cannot grow either.
28 P. G. DRAZIN AND L. N. HOWARD

We conclude this subsection with a brief account of some investigations

in which a direct attack on the initial-value problem for plane parallel flow
is made.
First we take the solution for plane Couette flow due essentially to Om
[26. pp. 2&27; cf. 16, p. 2091 in 1907. Here we put w ( = a) = y (- 1 <
<
y 1) in the perturbation of the vorticity equation for two-dimensional
flow to get

(;+.;)(%+!$)=a.
Therefore

for an arbitrary function F differentiable with respect to x . Now any given

well-behaved initial velocity distribution satisfying the equation of continu-
ity and the boundary conditions can be expressed in terms of the Fourier
integral in x and series in y ,

-m
.l'
) da cos a x
@ ( x , y , ~=
m
2 bn(a)sin +nn(y+ 1 ) .
n-1

This given initial distribution determines F ( x , y ) , and thence F(x - yct,y).

The resultant time-dependent equation for +'
above and the boundary
conditions can be shown to have the solution

#(x,y,t) = da
-w i 2 #,,(aB+ +fi2na)cosech 2a
ml:

* [(sinh 2asin {ax+ ( t a n- at)(y + 1))

- sinh a(l - y ) sin a x - sinh a(y + 1)
*sin {ax + 2 (inn - at)})/{a2 + (inn - act)'}
- (sinh 2a sin { a x - (inn + at) (y + 1)) - sinh a(l - y ) sin a x
- sinh a(y + 1) sin {ax - 2 (ifin+ at)))/{aa+ (inn+ #t)a)l.
*

Evaluation of this solution for large t shows that @ = O(t-1) and therefore
that plane Couette flow is stable.
A more systematic approach to the initial-value problem, more suitable
for application to basic flows other than plane Couette, comes from use
of the Laplace transform with respect to time. This approach has been
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 20

developed for stability problems by Miles [27], Carrier and Chang [28],
Case [29. 301 and Dikii [31. 321. It has been reconciled with the method
of normal modes, both for inviscid and slightly viscous fluid, by Case [33]
and Lin [34].
First let us take Case's [29] solution of plane Couette flow, for comparison
with the other solution above. Let the Fourier transform with respect to
x and the Laplace transform with respect to t give

p(y,a,t) = 1 e- %,h'(x.y,t)dx,

i
Y ( y ; a , p )G e- ~'!P(y,a,t)dt.
0

Equation (2.46) has the Fourier-Laplace transform

(2.46) (6 - a8) U/(y;a,p)=

P
1
+ iay
(G - a') p(y,a,o).

Next invert the Laplace transform, using the results that

where C is a Bromwich contour parallel to the imaginary axis and on the

right of all singularities of the integrand. Then

Given the conditions that +' = 0 on the boundaries for all x,t and therefore
that
!P,P=O ( y = *I),

we show in the usual way that the solution is

30 P. G. DRAZIN AND L. N. HOWARD

where the Green's function

I- sinh a(l + y ) sinh a(l - yo)

a sinh 2a (- 1 < Y <Yo)

sinhall + v,,)sinhall - v )
a sinh 2a \d" .d

In principle we know S/(y,a,O)as the Fourier transform of $'(x,y,O), the given

distribution of $' at 1 = 0. So we have q(y,a,t), and $'(x,y,t) on inverting
the Laplace transform, for all t 2 0. To study the behavior of $' for large
t , Case [29, p. 1451 evaluated the inverted transforms asymptotically and
showed that $' = O(l/t) for fairly general +'(x,y,O). This result agrees with
Orr's solution above and with the solution in Section II.4.a below, obtained
with normal modes.
To extend this approach for a general basic velocity profile w(y), we
follow the work of Case [29] and Dikii [32]. Here the Fourier-Laplace trans-
form of the linearized vorticity equation is

(2.48) = H(y,a)/($+ iaw),

say, where we may suppose H known in terms of the Fourier transform of
the initial distribution $'(x,y,O). The transforms of the boundary con-
ditions are

(2.49) W.Y;a,$) = 0 (Y = yl,ya).

The solution of (2.48), (2.49)

(2.50)
j.
PU(y;a,$)= G(y,yo)H(Yo,a)/($ iaw0)dro;
YI
+
where
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 31

(2.51)

such that

As in the case of plane Couette flow, we can now invert the transforms, etc.,
to find $'(x,y,t) as an explicit multiple integral of #'(x,y,O).
The crucial integral is the inversion of the Laplace transform,

(2.52)

We seek to evaluate this integral for large t to see whether the flow is stable.
When 1 is large, (ept(is large along C where Re p > 0. However, as we shall
see, cancellation of the large components of the integrand means that the
integral may not be large. The cancellation makes it advantageous to move
the contour close to the imaginary axis and thereby reduce the magnitude
of lePtI. Following the usual method of deforming the contour C, we must
study the singularities of Y, which is given by equation (2.50). Since Yl,!P,
satisfy equation (2.&51),their only singularities can be at the singularities
of this equation, namely where w ( y ) = ip/u = c. These singularities lie
only on the imaginary axis of the p-plane. Thus it is convenient to take
a new contour on the imaginary axis with indentations to the right of each
singularity or just to the right of the imaginary axis, say from E - ioo to
E + ioo for small E . To find P in terms of the integral along the new contour
we must know the other singularities of the integrand in the half-plane
Re $ > 0 ; these coming from zeros of the denominator of the Green's func-
tion G(y,y,), i.e. from the solutions of
W(Y1.Y2)
=0 (ReP>O).
This can be seen to be just the eigenvalue relation for the discrete spectrum
of normal modes with ci >O. [Of course, equation (2.51) is essentially the
Rayleigh stability equation with p = - iuc.] Supposing these eigenvalues
are already known, we may evaluate the residues a t these singularities of
the integrand in equation (2.52) and find

1
E+iW

exponentials
!P(r;u,t) = -
2ni Yb,a,+)ePtdt +
e-iw

The fastest growing component will be dominant for large t. It will come
from the discrete spectrum if there is one, being the term growing like
8' for the largest value of uc{. Case [29, p. 1481 and Dikii [33, p. 11801
32 P. G. DRAZIN AND L. N. HOWARD

have indicated that the integral above over the continuous spectrum decays
like l / t , so the discrete spectrum alone is associated with instability. Thus,
in seeking a criterion for instability, we h a y use the method of normal
modes and ignore the continuous spectrum.
4. Stability Characteristics of Variozcs Basic Flows
As will be seen, most of the basic flows with known stability charac-

zw
teristics are piecewise linear. However, the characteristics of many smoothly-

--tf (a) (b)

Y Y

FIG. 4. (a) Plane Couette flow: w = y. (a) Plane Poiseuille flow: w = 1 - ya.
(c) Sinusoidal flow: w = sin y. (d) Vortex sheet: w = VS(cd,2 y > 0). w = V,(O>
y2 - dJ. (e) Rectangular jet: w = I(lyl c I), w = O(lyl> 1). ( f ) Thin jet w = (d(y))l/*.
(9) (i) Symmetric jet in channel. (g) (ii) Antisymmetric shear layer in channel.
HYDRODYNAMIC STABILITY OF P A W E L FLOW OF INVISCID FLUID 33

(k)

(01 (PI

FIG. 4. (continued) (h) Unbounded symmetric-trapeziumjet. (i) Double jet. (j) S p -

metric separated double jet. (k) Shear layer. (m) Bickleg jet: IV = Sech' y.
(0) Hyperbolic-tangent shear layer: m = tanh y. (p) Boundary layer with auction:
-
w = 1 6-y.
34 P. G. DRAZIN AND L. N. HOWARD

varying basic flows are partially known, and the advent of electronic com-
puters is allowing many to be found completely in numerical terms. Diagrams
of the basic flows accompany the examples below.

(a) Plafie Cozcette Flow

We follow the treatment [35] of Fjertoft and Heiland. When
w=y (-l<y<l)
the Rayleigh stability equation becomes
(y - c ) ( P - aa)q= 0.
Unless c =y in the domain of flow, i.e. unless - 1 < c < 1, this gives
only
(D2- a2)g,= 0,

which has no solution vanishing at both y = f 1. Thus the basic flow is

exceptional in that it has no eigensolutions of the discrete c-spectrum.
< <
However, when - 1 c 1, the stability equation also gives

( D 2- a2)q= S(y - c ) ,
where 6 is the Dirac delta-function. This representation in terms of a general-
ized function is admissible because the typical component of wavenumber
a is really only one component in an integral. It gives a solution of the
continuous c-spectrum with eigenf unction

v= II
sinh a(c - 1)sinh a ( y

sinh a(c
a sinh 2u
+ 1)
+ 1)sinh cr(y - 1)
a sinh 2a
(- 1<Y<C)

(c <Y < 1)
for each value of a and for each value of c in the interval (- 1,l). The set
of eigenfunctions is complete so that an arbitrary initial disturbance of the
velocity field can be represented as a sum orintegral of them [cf.36,pp. 11-12].

(b) Plane Poisezlille Flow

When
w=l-yS (-l<y<l)
there is no point of inflection, so the flow is stable (though the flow is unstable
for viscous fluid at large Reynolds numbers by Heisenberg's criterion). It
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 36

seems that no further examination of the stability characteristics has been

made except through the viscous problem.

(c) Sinusoidal Basic Flow

When

the stability equation becomes

Now D%J = 0 where y = ys = mz (n = 0, f 1, f 2,. . .). If there is no

value y s in the interval (y1,y2)the flow is certainly stable by Rayleigh's
criterion. If there is at least one value, we may suppose ys = 0 without
< <
loss of generality, so y1 0 y,. To find the s-solution we put c = w, = 0.
Then
siny{& + (1 - as*)vs}= o
where
pla = 0 (y = Y l J 2 ) .

In finding the s-solution we ignore the factor sin y (and thereby discard the
stable eigensolution corresponding to c = 0 in the continuous spectrum)
to get
vs = sin M Y - Yl)/(Y2- Y1))>
as = (1 - nznz/(yz- y1)2}1/2
for each positive integer 12 < (y, - yl)/n. I t follows that the flow is unstable
if (r, - yl) > n , but stable otherwise although the point of inflection lies
at y = 0 in the field of flow. This counter-example to the sufficiency of
Rayleigh's necessary condition for instability is due to Tollmien [14; see
also 7, pp. 219-2201,

(d) Vortex Sheet

When

the eigenvalue relation is [2; 3, p. 3791

(e - V J 2coth ad, + (c - VJZcoth adl = 0
36 P. G . DRAZIN AND L. N. HOWARD

and the eigenfunction

Therefore
c = { V, coth ad, + V, coth adl
+ ilVa - V1l [coth ad1 coth ada]l/e)/{coth ad, + coth ad,}.
This gives complex c and instability for each a.
When d,,dg = m,V2 = 1 = - V,, this gives c = i. This is an example
of an antisymmetric flow with c, = 0. It also gives the limiting eigenvalue
(2.29) as a -P 0 of any smoothly-varying flow with w ( 00)= 1 = - w ( - cm).
The elevation of the material surface with mean level y = 0 is q =
+
If V,+ V , = V and dl,da = 00, one finds c = - V and q = (A Bt)e”(x- vl)
for arbitrary constants A , B . This has been described as the instability of
a flapping flag.

(e) Rectalsgzclar Jet

When

there is [3, pp. 380-3811 a sinuous mode with

c = (1 + i(coth a)1/2}/{1+ cotha}
and a varicose mode with
c = (1 + i(tanha)’/2}/{1 + tanha}.
Each mode is unstable for all a.

(f) Thin Jet

When
w = {s(y)}’/a (- -< y < 00)
W

the velocity is infinite at y=O. but the total momentum flux J w a d y = l .

-w
With piecewise solution of the stability equation and use of conditions (2.12)
at infinity and (2.16) at y = 0. we find
c = &a)W
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 37

and

Although w is an even function, there is only the sinuous mode above. It

is unstable for all a, and is in accord with limit (2.30) for a smoothly-varying
jet of the same momentum flux.

(g) Channel Flows

Rayleigh [3, pp. 386-3901 found eigenvalue relations for a general
continuous piecewise-linear velocity profile in a channel, there being two
discontinuities of the velocity derivative between the walls of the channel.
In each case the eigenvalue relation is a quadratic equation for c. We give
a few examples below.

(i) For a symmetric-trapezium jet with

w=( V (Irkib’)
Vb-’(b + $3’- Y ) (ib’ < IYI G b + 8 0
Rayleigh found
c=V - V{absinh a(2b + b’)}-l{sinh ab sinh a(b + b‘) f sinha ab}.
Thus both the sinuous and varicose modes are stable, as was to be expected
for this flow which approximates a smoothly-varying profile with curvature
of one sign.
If further the middle layer is absent, then b’ = 0 and
c=V - Va- b-1 tanh ab.

[
(ii) For an antisymmetric shear layer with
V +W Y - ib’) (ib‘ < y <b + i q
w= 2Vy/b’ (IY I< tb‘)
-V + nv(y + ib’) (+b’ 3 y 2 - b - gb‘)
Rayleigh gives [3, p. 3881
+ a sinh a(b + b‘)]*
cB= P {[(A - 2/b‘) sinh d sinh ab’
- aBsinhaab}/{assinh ab’sinh a(2b + b’)}.

Thus there is instability for some a only when - l i b < A < - l / b + 2ib’.
This result in some sense exemplifies Fjmtoft’s result for smoothly-varying
profiles represented in Figure 2.
38 P. G. DRAZIN AND L. N. HOWARD

(h) Unbo%nded Symmetric- Trapeziwn Jet

When

0 (IYl>1)
a)@ -4 (1 > IYI > 4
(a > IY I)
the eigenvalue relation is [3, p. 3971
4(1 - - 2(1 - a)ac{2(1 - a)a ‘f e-*(l - e-2(1-a)a)}
+ {- 1 +2(1 - a)a + r2(1-a)a} e-*{1 - [I + 2(1 - a)a]e-2(1-a)a}
= 0,

where the upper sign is for the sinuous and the lower for the varicose mode.
When a = 1 we get an example of the rectangular jet (Section II.4.e).
When a = 0 we get a triangular jet, with
2aW +a(1- 2a - e-%)c + (a(1 + e-2”) -1 +e-a) =O

for the sinuous mode, and

c = (2a)-1(1 - e-”)
for the varicose mode. Here the varicose mode is always stable, but the
+
sinuous mode is unstable for O < a < as 1.8. The logarithmic growth
rate aci of the sinuous mode is greatest when a = 1.2.

(i) Doltble Jet

When

0 (IYl>1)
w = I u (-l<y<O)

1 (O< Y < 1)
the eigenvalue relation is [S, p. 2691

+ (1 - c ) tanha}{c*
(1 - C ) ~ { C % ~ tanha + (U- c)~}
+ (U- tanh a + (1 - c)%}{ca + (U- c)%tanh a} = 0.
C)~(C%

This gives three modes unstable for each a, with

c = i(i(1 + Ua)a}l/Z+ . . ., 1 + i(ia)1’2+ .. .,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 80

or
U{1 + i(ta)'/*+ ...}
for small a.

(j) Syrrametric Sepuruted Dauble Jet

When
0 ( I r l > 2 or I r l a
w=( 1 Irl<2)
the eigenvalue relation is [8, p. 2691
{c/(l - c))*(l + {c/(l - c))* tanha) + 1+ {c/(l - c))*cotha = 0
for the sinuous modes and
{c/(l - c)}*({c/(l - c ) } ~+ cotha) + {c/(l - c)}* + tanha = 0
for the varicose modes. This gives four modes unstable for each a.

(k) S k a r Layer
When
= { rllrl (Irl>4
Yla (Irl<a)
the eigenvalue relation is [3, p. 3931
-2 4 *
c* = (4~*a*)-~{(1 - e-&}.
Only two special cases are of interest : a = 0 and a = 1. The former gives
Kelvin-Helmholtz instability of the vortex sheet (Section II.4.d). When
a = 1, c is pure imaginary for a< a, I 0.64 and real for a 2 a,. The
logarithmic growth rate acj is greatest when a = 0.4.

(1) Double Vortex Skeet

When

the eigenvalue relation is [a, p. 2701

{d + (1 - c*)*} tanh 2cr + 2 4 1 + c') = 0.
40 P. G. DRAZIN AND L. N. HOWARD

<
There are two modes unstable for each a. For a 4 tanh-1 (6l/*-- 2) both
unstable roots c are pure imaginary; for a >4 tanh-l (S1le - 1) both are
complex. Thus an antisymmetric profile does not neceSSarily give c, = 0.

(m) BickZey Jet

When
w = sech*y (- w < y < m)

the s-eigensolution for the sinuous mode is [36]

qs= sechz y, ws = Q, a, = 2
and for the varicose mode [37]
cps= sech y tanh y, ws = #, a, = 1.

The Lin perturbation (2.20) then gives [8, p. 2811

ac/aa2 = 0.0423 - i0.0278 (a = 2 - 0)
for the sinuous mode, and
ac/aas = - 0.0264 - iO.0836 (a = 1 - 0)

for the varicose mode. For small a the sinuous mode has [8, p. 2811
c =a + i{#a - a* - i d l o g (24a-1) - &azni}l/*+ . , ,
and the varicose [8, p. 2791
c =1 + e(2/a)d(&a)21a(l+ gC(Wa)"(#n2a2)1/8 + O(a)}.
Numerical values of q,c, for some a have been given by Lessen and Fox
[38; cf. 81. We give results of a somewhat more complete recent calculation
in Table 1.
Stability Characteristics of other jet or wake profiles have been computed
by Hollingdale [39], Haurwitz and Panofsky [40], and Sato [41].

(n) Astisymmetric Dozcble Jets

When
w = Sech"'y t a h y (- OO< y< 00, m >, - 1/2)
the s-eigensolution is [41]
pt = sechm+ly, w, = 0, q*= 2m + 1,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 41

It seems that solutions corresponding to inflection points other than y = 0

are not known. Numerical results are given in Table 2 for w = )V%sechay
tanhy. Note that the “propagating mode”, though associated with the

TABLE1. INSTABILITY
CHARACTERISTICS
OF THE B I C KL ~Y
JET

Sinuous Mode Varicose Mode

a cr ci Cr ci

0.06 0.030 0.169 0.931 0.079

0.1 0.086 0.216 0.889 0.104
0.2 0.166 0.267 0.826 0.121
0.3 0.229 0.267 0.780 0.119
0.4 0.280 0.263 0.746 0.108
0.5 0.323 0.262 0.719 0.092
0.6 0.362 0.236 0.700 0.074
0.7 0.396 0.218 0.686 0.056
0.8 0.424 0.198
0.9 0.461 0.170
1.0 0.475 0.169 0.667 0
1.2 0.620 0.121
1.4 0.669 0.086
1.6 0.597 0.053
1.8 0.632 0.024
2.0 0.667 0

-
TABLE2. INSTABILITY
CHARACTERISTICS
FOR UJ = 9 1 3 sech* y tanh y

Standing Mode (c, = 0) Propagating Mode

a ci a 6, ci

0 0 0 1.o 0
0.1 0.270 0.05 0.944 0.071
0.2 0.362 0.1 0.903 0.103
0.3 0.426 0.2 0.831 0.133
0.6 0.495 0.3 0.775 0.139
0.7 0.612 0.4 0.732 0.130
1.o 0.472 0.6 0.700 0.116
1.5 0.317 0.6 0.678 0.098
1.8 0.199 0.7 0.660 0.077
2.0 0.114 0.8 0.648 0.062
6 0 0.9 0.641 0.044
42 P. G. DRAZIN AND L. N. HOWARD

inflection point at w = 2-lI2, does not have c, + 2-lI2 as ci + 0. This is a

case where the Reynolds stress for the neutral mode has two compensating
jumps. There is of course another propagating mode, associated with the
inflection point at w = - 2-li2, for which the sign of c, is reversed.

(0) Hyperbolic- Tangent Shear Layer

The special case m = 0 of the example in Section I1.4.n gives the shear
layer

which has better known stability characteristics [S, p. 2811. The s-eigen-
solution is [43, 441

vs= sechy, is= 0, as= 1.

The Lin perturbation gives

ct = (2/n)(l - a) + 0(1- a)B as a + 1-

and for small a

~d =1 - 1.786a + 1.52W + .. . .
Numerical calculations for this case have recently been given by Michalke
[&I.
Stability characteristics of some other smoothly-varying profiles of
shear-layer type have been computed by Hollingdale [39], Carrier [cf. 461
and Lessen and Fox [38]. Their results are similar to those for the hyperbolic
tangent shear layer.

(p) Bounda~yLayer with Suction

When

Lin pointed out [47, p. 901 that the Rayleigh stability equation can be
transformed into the hypergeometric equation, a result used by Chiarulli
and Freeman 1471.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 43

111. WAVESAND STABILITY

OF PLANEPARALLELFLOW FLUID
OF INVISCID
UNDER THE ACTIONSOF VARIOUSFORCEFIELDS

1. Introrlzcction
The theory in this section consists chiefly of developments by Scandina-
vian meteorologistsin the 1920’s and 1930’s of Kelvin’s paper [2] of 1871. Their
work is collected in a book by Bjernes, Bjerknes, Solberg, and Bergeron [48]
of 1933. It may be more accessible to the reader in Chapter X of the book by
Godske. Bergeron, Bjerknes, and Bundgaard [49]. which also has a biblio-
graphy of relevant meteorological papers published by 1960. These me-
teorologists have considered the equations, boundary conditions, and eigen-
solutions for piecewise-constant velocity profiles under the influences of
combinations of density variation, compressibility, and rotation. Haunvitz
[SO] has considered the equations and boundary conditions under the same
combinations of force fields for smoothly-varying profiles also.
The very generality of these combinations and their meteorological
context has obscured some of the fluid dynamics and enabled other authors
to duplicate their work in ignorance of it. So we shall consider the force
fields separately in order to simplify the understanding and to compare
the effects of different force fields. Of course any combination of these
force fields can be considered, and. indeed, there are dozens of papers dealing
with various combinations. However these combinations need no special
techniques in their treatment, so we shall not describe then. If the reader
is interested in a physical problem of stability of parallel flow affected by some
combination, he may readily adapt the methods used for the fields separately.
Again, we have ignored the effect of surface tension on an interfacial
boundary condition. This may be physically important, but it is a simple
matter to apply the method of Kelvin [2] to cater for surface tension.
Treating one external force field at a time, and finding its effects on the
inertial instability discussed in the previous section. we first give the stability
equation and boundary conditions for two-dimensional wave-disturbances,
and comment on the validity or invalidity of Squire’s theorem. Then we find
the eigensolutions for two unbounded basic velocity profiles, that of static
equilibrium (w = 0), and that of a vortex sheet (w = y/lyl). It is found
that for static equilibrium a neutrally stable wave motion may occur for
most of the external force fields. For example, sound waves may occur
when the external “force” is due to compressibility of the fluid. Such a
wave motion is important both for its own sake and for its representation
of some stability characteristics of profiles with shear. The example of a
vortex sheet is also important, because a vortex sheet is the simplest flow
with shear, and because its stability characteristics represent those of any
smoothly-varying shear layer for long-wave disturbances. Finally we sum-
marize briefly the literature on stability problems for each force field.
44 P. G. DRAZIN AND L. N. HOWARD

Thus the section is a rapid survey of problems within our purview.

Their juxtaposition emphasizes their similarities. In the next section we
shall emphasize those similarities further, with dimensional and physical
arguments. Then we shall choose one example of the external force fields,
that of buoyancy of a fluid of variable density, as a prototype and discuss
its stability characteristics in detail.

2. Internal Gravity Waves and Stability of a Fl& of Variable Density

Following Kelvin [2], many authors have studied the stability of parallel
horizontal basic flows of incompressible fluid with piecewise-constant
velocity and density distributions (i.e. with velocity and density uniform
in layers but varying from layer to layer) under the action of gravity.
Rayleigh [61] was the first to consider the stability of fluid of smoothly-
varying density distribution p,,,(y*) at rest, y* being the height. For the
particular density distribution p* = po exp (- By) (- bo < y* < bo, con-
stants Po, fi >O), he found neutrally-stable internal gravity waves of phase
velocity
(34 c* = a, = (gp/a,*)'lS,
on neglect of a&, i.e. on neglect of the variation of inertia due to the
variation of density but not of the buoyancy.
Taylor [62], Goldstein [63],and Haurwitz [60] consideredtwo-dimensional
disturbances of parallel horizontal flow of incompressible fluid under gravity
with smoothly-varying profiles of velocity and density. Their angysis
leads to the stability equation with dimensionless form,
(w - c)(D*- a*)p,- (0%)~ +
J y / ( w - c) - K{(w - c)Dy - ( h ) p , } = 0;
(3.2)
where the local Richardson number of the basic flow is
(3.3) =
Jcy) - ( g L B d p * / d Y * ) / W ,
8

a measure of the characteristic ratio of the buoyancy to inertia; and

(3.4) K ( Y ) = - LdF*/ij*dY* ,
is a measure of the characteristic ratio of the variation of inertia due to
heterogeneity of the fluid to the inertia. (Note that the Froude number,
(3.6) F E gL/Vp= J ( ~ ) / W Y ) ,
is independent of y.) The stability equation can be seen to reduce to the
Rayleigh stability equation for a homogeneous fluid, i.e. for dii,/dy, = 0.
A t the walls (or infinity) we use boundary conditions (2.12) as before. At
a discontinuity of w , Dw or j the continuity of the normal velocity component
and of the pressure imply respectively that
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 46

In fact Squire’s theorem may be extended to this case [54], giving for
each three-dimensional wave making angle 0 with the basic flow a two-
dimensional wave of the same growth rate but effective Richardson number
Jcos*O and Froude number Fcos*0. Usually density variation with
dp,/dy, < 0 acts as a stabilizing agent, so we seek criteria of stability with
minima of J or F. Thus two-dimensional disturbances are the most unstable
usually, and anyway the stability characteristics of any three-dimensional
disturbance follow at once from knowledge of the characteristics of all
the two-dimensional disturbances,
The outstanding problem of this case is Kelvin-Helmholtz instability of
the basic vortex sheet with

(3.7)

Here the eigenfunction can be shown [a] to be

where the eigenvalue

Thus the flow is always unstable if there is heavy fluid above lighter
< A*). If there is heavy fluid below lighter, the flow is stable to those
waves with

(3.10)

but unstable to all shorter waves.

If V = 0, then we have
(3.11) ‘* = fk @ 8 * - Pl*)/a*(Pl* + P8*))”

for neutrally-stable internal waves at the interface of the fluids. Taylor [66]
recognized that the same analysis gives

c* = f{(g - g’)@** - Pl*)/a*(Pl* + P**)}’”

when the interface has constant acceleration g‘ downwards. Thus there is
instability when (g - g‘) and (pa* - pl*) have opposite signs, i.e. when the
resultant acceleration (g’ - g) is directed from the lighter toward the
46 P. G. DRAZIN AND L. N. HOWARD

heavier fluid. This is called Raylcigh-Taylor instability on account of

Taylor's work and the paper by Rayleigh [61] on waves in a heterogeneous
fluid at rest, which he showed unstable when dp*/dy, > O somewhere. For
an example, the surface of water in a bucket is subject to Rayleigh-Taylor
instability when moved downwards with constant acceleration greater in
magnitude than g.
If pa* = pl*, then eigenvalue (3.9) reduces to the value (Section II.4.d)
for a homogeneous fluid.
In many geophysical problems the effects of variation of inertia are
negligible though the buoyancy is not, so we may approximate K = 0 with
J # 0. Then the stability equation becomes
(3.12) (w - c ) ( P - aa)p,- (D%)q + J ( y ) y / ( w- c ) = 0.
When p* = p,, exp (- By,,), it can be seen that J ( r ) = g,dLa/Va, a
constant. When further we = 0, we get Rayleigh's internal gravity waves
with
y = constant, c = f (J/a*)1/2 = f a.
When p+ = po exp (- By*), w* = y*/Iy*J it can be shown [48, 60, 661
that the eigenfunction is
(c - 1) exp (- (1 - d / ( c - l)a)112ay)
(3*13) {
'P = (c + 1) exp ((1 - a*/(c+ l)a)l/Zay)
(r > 0)
(Y < 0 )
where
(3.14) (C - l)*{l- u*/(c - l)'}'" + + 1)'{1 - d / ( c + 1)*}'Ia = 0.
(C

The square-roots must be chosen with non-negative real parts in order that
the eigenfunction (3.13) is bounded at y = f 00. If a square-root is pure
imaginary. its sign must be chosen so that there is outward radiation of
energy at infinity; however, this occurs only for real c, in which case there
is stability anyway. It now follows from squaring up equation (3.14) that
(3.16) c=O and 8 2 1 ; c*/a*=l and a t = 00;

or
ca = - 1 + &as.
The second mode represents Rayleigh's internal gravity wave with I'= 0
and c,, = f a,. The first and second modes are isolated from one another,
and from the third mode, which is the only one that can give instability.
It can be seen that there is stability to all waves only when
(3.16) as 2 2, i.e. aa < +gp/Va.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW O F INVISCID FLUID 47

We shall not discuss the stability characteristics when K = 0 further

now, because we shall take them up in detail in Section V.
The case J = 0, K # 0 is also not without interest, and it may represent
instability of vertical flames. Kere the stability equation becomes
+
(w - c)(D*- a 2 ) p- ( D 2 w ) q K ( y ) { ( w- c)Dp - (Dw)pl}= 0.
For a vortex sheet with exponential density,
w = y / ( y l and jj = exp (- Ky) (constant K > 0),
the eigenvalue can be shown to be
+
c = (- K 2 / 2 a Z i ) / ( l + K2/2a2)1/e,
giving instability for all K . a .
Menkes [67] has considered such a problem for the smoothly-varying
shear layer w = tanhy.

3 . Souad Waves and Stability of Compressible Flzcid

Stability of a basic parallel flow of compressible perfect gas with
piecewise-constant temperature and velocity was first studied by Bjerknes
et al. [as] and Haurwitz [60]. Haunvitz also found the stability equation
for basic flows with smoothly-varying temperature and velocity. He in
fact considered external fields due to buoyancy and rotation as well as
compressibility, but in our special case for two-dimensional waves in adiabatic
motion it has the dimensionless form
(3.17) D({(w - c)Dpl - (Dw)p}/{aa- (w - c)") - a%r2(w - c)pl = 0.
Here pl is defined by the equation for the lateral velocity,
v' = iapl(y)exp {ia(z- ct)}.

because two-dimensional motion of a compressible fluid has no stream

function. Also the local inverse Mach number of the basic flow is u(y) =
a+(y*)/V,a, being the local speed of sound. In general a, vanes with the
basic temperature T,(y,) of the perfect gas so that a, = (yRT,)'l2,
where y is the ratio of its specific heat at constant pressure to that at constant
volume, and R is the gas constant. Note that the stability equation
(3.17) above reduces to the Rayleigh stability equation as a + 00, i.e. as
the fluid tends to be incompressible.
The boundary conditions (2.12), at the walls are valid for compressible
fluid as well as incompressible fluid. The boundary conditions at a discon-
tinuity of w , Dw, a , or Da can be shown in the usual way to be
(3.18) [pl/(w- 41 = 0,
[{(w - c)Dp - (Dw)pl}/{az- (w - c ) ~ } ]= 0.
48 P. G. DRAZIN A N D L. N. HOWARD

A generalization of Squire’s theorem for this case is valid, giving [68]

each three-dimensional disturbance of the basic flow with w+(y+), a*@+)
the same growth rate as a two-dimensional one for a basic flow with w+ cos 8,
a*, i.e. with velocity scale V cos 8 and therefore Mach number V cos 8/u, =
u-1 cos 8 < a-1. Thus to each two-dimensional disturbance there corre-
sponds a three-dimensional one of the same growth rate but higher Mach
number. It follows that if a flow of slightly compressible fluid is unstable
to some two-dimensional disturbances then the same flow is unstable at all
Mach numbers to some three-dimensional disturbances. Thus, although
we shall find the cushioning effect of compressibility a stabilizing one by
and large, it can never stabilize waves nearly perpendicular to the basic
flow. However, it is fruitful to examine the stability characteristics of com-
pressible fluids, and it is again sufficient to consider two-dimensional dis-
turbances only, because their characteristics trivially imply those of all
three-dimensional disturbances.
The important problem of a vortex sheet has been treated by Landau [69],
Hatanaka [60], and Miles [27]. With

the stability equation (3.17) solved piecewise with boundary conditions

(2.12). (3.18) gives eigenfunction
(c - 1) exp (- a{l - (c - l)s/ala}1/2y)
(c+ +
1) exp (a{l- (c 1)*/%*}1/2y)

al-*(c - l)S{l - (c - 1)*/a1a}-1/2

(3.20) + ae-a(c + 1)a{1 - (c + 1)a/a9*}-1/2= 0,

where the square-roots have non-negative real parts, etc.
For illustration, let us take the special case of uniform basic temperature.
Then ua = a,, = a say. Therefore

(3.21) (C - (c + 1)2)’’2 + (c 4- 1)*{aa- (C - 1)’}11” = 0

- l)a{~9
and it follows that
c*/a*>, 1 and a9= w;
(3.22)
c=O and ag<1;
or
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 49

The first mode represents a sound wave with c, = f a, when V = 0. The

second mode c = 0 gives a root of the unsquared equation (3.21) only when
the square-roots are pure imaginary and have opposite signs. In fact, it
represents the steady flow of supersonic streams on both sides of a thin
wavy rigid wall coincident with the stationary interface between the two
streams (cf. Liepmann and Roshko [61], art. 8.6); all streamlines have the
same shape and size, but wave crests are out of phase, lying on Mach lines.
These stable modes are isolated from one another and from the unstable
third mode, which exists for all values of a. As a 00 (which may be effected
by letting V + 0 or a, + m) the third mode gives c + i or f 21'2a. The
former limiting root corresponds to the instability of a vortex sheet in an
+
incompressible fluid. As a -,0, c = f 1 & a O(aB); these four roots
correspond to stable sound waves superposed on the upper and lower streams
of the basic flow. The vortex sheet is stable to two-dimensional disturbances
if and only if c is real for all a, i.e.
(3.23) a < 2-112.
Kuchemann [62] has studied the stability of a piecewise-linear profile
representing a boundary layer in a compressible fluid. Lin [63]has con-
sidered general stability characteristics, and particular ones of a shear
layer. Eckart [a] has generalized Howard's semicircle theorem for com-
pressible fluid.

4. Planetary Waves and Stability in a Rotating System

If the equations of motion of inviscid fluid are referred to a frame
rotating with constant angular velocity a,the Coriolis acceleration must
be addedrto Euler's equations, but the centrifugal force may be put with
the pressure, giving
&*/at, + u*. VU*+ 2 a x 11, = - VP,,

where p*P, is the pressure plus the centrifugal potential. Thus the vorticity
with respect to a non-rotating (inertial) frame is 2n plus the relative vorticity
O, = V x u*. Using the modified vorticity equation and the usual methods
of normal modes, Johnson [66] has found the stability equation of a three-
dimensional disturbance,
(W cos e - c ) % ( D-~ a9)cp - cos qrar
cos 8 - c ) ( D * w ) ~ ~
(3.24)
+ Ro-1 cos x(R,,-l + sin 8Dw)cp = 0 ;
cos x
where pl is defined by v' = iav(y)exp {i(ax + yz - act)}, x is the angle
between and the wave-number vector = (a,O,y),and the Rossby number is

(3.26) Ro --= V/2LR,

50 P. G. DRAZIN AND L. N. HOWARD

a characteristic ratio of the inertial to Coriolis forces of the basic flow.

When cos x/Ro and 8 are zero this equation becomes the Rayleigh stability
equation for two-dimensional disturbances in a non-rotating system. When
8 = 0, cos x/R, # 0, the equation has the same form as equation (3.12)
for a heterogeneous fluid. However, Squire’s theorem is invalid in this case,
the above equation for 8 # 0 differing essentially from its form for 8 = 0.
The boundary conditions (2.12)at a wall are still valid. At a discontinuity
of w or Dw, continuity of normal velocity and pressure at the material
interface give respectively,
[v/(wcos e - c)] = 0,
(3.26)
[(w cos 8 - c ) D -
~ cos B(Dw)tp + R,-l cos x tan @,I = 0.

When w = 0 (- 00 < y < 00) the solution is [cf. 49, p. 3361

(3.27) v = constant, c = f c o s ~ / a R ,= fa,

say, giving plarcetary (or iwertial) waves with phase speed

a, = 2Q cos x/a*.

For the vortex sheet w = y/lyl (- ao< y < ao) the eigenfunction is

(3.28) 9, =
1+
(c - cos 8) exp (- ay(1 - a2/(c- cos 8)z}1/2)
+
(c cos 8) exp (ay(1 - a’J/(c cos 8)2]1/2)
and the eigenvalue relation is
(c - cos e)yi - ay(c - COS e ) y
(3.29)
+ (C + cos e)z(l - u ~ / (+c cos 8)z}1’z= 2a sin 8.
, I

When 8 = 0 this eigenvalue relation has the same form as that of (3.13),
(3.14) for a heterogeneous fluid in a non-rotating system. On squaring up
(3.29) etc. for general 8, we get the cubic equation in cz,

(3.30)
o =~ ( C Z ) 4 cosz ec6 + (8 cos4 e - a2(i + 3 cOSae)}c4
+ ( ~ ~ ~ S ~ ~ - ~ U ~ C O S ~ O ( ~+ - UC ~O }S C~ O~ )+ U ~ ~ ~ ~ ~ ~ ( ~ ~ - C O S

As a + 0, the three roots are: cB= i a z tanz8 O(a4) or - cosz 8 f +

+
2ia sin 8 O(az). This gives only two admissible roots of the unsquared
equation (3.30) with square-roots having non-negative real parts, namely
c = & i cos 8 +
a sin 8 +
O(a8). As a --* 00, cz + - sin* 8, or

(a8/8C O S ~e){i +3 8 f sin e(i - 9 C O S ~8)1lz}.

COS~
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 51

The first root is inadmissible. The latter pair give only a complex conjugate
pair of roots c when cos2 8 > 1/9, implying instability. When cos2 8 \< 1/9
there is stability provided a sin 8 2 0 but not otherwise.
The Scandinavian meteorologists [cf. 48,491 have considered the stability
characteristics of various piecewise constant velocity profiles and Johnson
[66] has treated the shear layer w = tanhy.

6. Rossby Waves and Stability of Fluid in a Rotating System with Variable

Coriolis Parameter
For large-scale (w lo3 km) motions of the earth’s atmosphere or oceans
it is customary to neglect the kinematic effects of the earth’s curvature and
use rectangular Cartesian coordinates, but to retain the more important
dynamic effects of the variation of the Coriolis parameter f 2Q sin 1
with latitude 1. This is done in the 8-plane method of Rossby [cf. 661.
With this approximation it can be shown that the only modification to the
stability of an eastward horizontal basic flow fi,,, = w*(y*)i relative to the
earth is the addition of f i to the relative vorticity D,w,k, the earth having
+
angular velocity R(cos Aj sin A). Kuo [66] has shown that this leads
to the stability equation,
(w - c)(DB- a 2 ) v- (D% - 8 )=~0
by the usual methods for two-dimensional disturbances, where 8, = D*f
is usually approximated by a constant and y* by the product of the local
value of 1 and the radius of the earth.
The boundary conditions (2.12) at a wall hold in this case. A t a dis-
continuity of w or Dw conditions (2.13), (2.14) still hold.
Squire’s theorem gives each three-dimensional disturbance of the basic
flow with w,(y,),p, the same growth rate as some two-dimensional one of
the basic flow with wpI*cos O,p, cos 8. Hence it is sufficient to consider
two-dimensional disturbances only as in Section 11.
When w = 0 (- b o < y < a), we get the solution
(3.31) = constant, c = - 8/a23 - a.
say. This represents a Rossby wave of phase speed a* G fi*/a*2. Rossby
waves travel westwards and are dispersive. They are really a form of neutrally-
stable inertial oscillation on the rotating earth.
For the vortex sheet w = y/lyl (- 00 < y < bo) the eigenfunction is
re71
62 P. G. DRAZIN AND L. N. H O W A R D

and the eigenvalue relation is

(3.33) (c - 1)2{l + a/(c- 1)}1/2+ (c + l ) a { l + a/(c + 1)}ll2 = 0.
On squaring up, etc., it follows that c/a 2 - 1 and a = 00 or
(3.34) 0 = f(c) = c(c2 + 1 ) + a(3ca+ l ) / 4 .
The former mode is isolated, giving the Rossby wave with c = - a when
w* = 0 (i.e. V = 0). The cubic has one real root, admissible only if 2 < a < 00,
for which-there is stability with - 1 2 c 2 - fa. However, there is also
an unstable mode with complex conjugate pair of roots of the cubic, for
which c -, f 3-'/% as a + 00 and c + f i as a -+ 0. Thus the rotation
is a weakly stabilizing influence.
Kuo [66], Lipps [68] and Howard and Drazin [67] have considered
other problems of this case.

6 . Magnetohydrodynamic Waves and Stability of an Electrically-Conducting

Fluid in a Magnetic Field
Many problems of stability of parallel flow of an inviscid incompressible
electrically-conducting fluid in a magnetic field have been considered.
They may be classified by use of the magnetic Reynolds number,
RMG V L / 1 ,
an overall measure of the ratio of the convection of the magnetic field to
its diffusion, where 1 is the magnetic diffusivity of the fluid. Thus stabil-
ity problems may be specified by RM as well as the variation and magnitude
of the basic magnetic field H.
We shall restrict our attention to problems for which
(a) the basic magnetic field is uniform and steady, so that the variables
may be separated to yield a tractable stability equation;
(b) RM is zero or infinite, so that the stability equation is of second
order, like the other stability equations discussed in. this paper;
(c) the basic magnetic field is directed in the (x,y)-plane of flow, because
Squire's theorem is invalid otherwise,
With these restrictions we may state three eigenvalue problems typical
of magnetohydrodynamic stability of parallel flow of inviscid fluid.
( 1 ) When RM = do (i.e. the fluid is a perfect conductor) and the basic
magnetic field H = (H,,O,O) is parallel to the flow, the stability equation
can be shown [69, 70, 711 to be
(3.36) D({(w - c ) -~ aa}D{y/(w- c ) } ) - aa{(w - c ) z - as}{y/(w - c ) } = 0 ,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 63

where
(3.36) as E pHo2/4nVapp,,

a characteristic ratio of the magnetic to kinetic energy of the basic flow,

p being the magnetic permeability of the fluid.
The boundary conditions (2.12) at a wall hold as usual. At a discontinuity
of w, Dw. or a the conditions are [71]

(3.371
. .
[{l- u4/(w - c)2} { ( w - c)Dp - (Dw)q}]= 0.

When w = 0 (- do < y < cm) the solution is

(3.38) p = constant, c= fa.

This represents A l f v k (or magnetohydvodynarnic) wuves of phase speed

a* G (pH0a/4np*)'/2.
For a vortex sheet w = y/Jyl(- 00 < y < cm) the eigenvalue is n13

(3.39) c= * (as - 1)I".

Therefore the flow is stable if and only if aa 2 1.

(2) When RM = 0 and the basic magnetic field is parallel to the flow,
the stability equation can be shown to be [70]

where
(3.41) N pHo2LL/4np,/V.

The boundary conditions are the same as with no magnetic field. There
is no progressive wave possible when w = 0, and the vortex sheet is unstable
for all values of H,, however large, although the magnetic field reduces
the instability [cf. 721.

(3) When RM = 0 and the basic field = HJ perpendicular to the

basic flow, the stability equation can be shown [73] to be

(w - c)(DP - a2)p - ( 0 % )-~iNDaq = 0.

However, in this case the chief effects of the magnetic field on the stability
characteristics occur through change of the basic flow itself rather than
through change in the mechanism of instability.
64 P. G. DRAZIN AND L. N. HOWARD

IV. HEURISTICTHEORYOF INSTABILITY

1. Dimensional Analysis

The instability we have described is essentially a manifestation of three

mechanisms :
(a) the inertial instability of the basic flow, whereby the basic balance
of vorticity is upset:
(b) the kinematic constraints of the boundaries, which by and large
reduce instability;
(c) the external force field, such as buoyancy or the Coriolis force.
In Section I1 we discussed mechanisms (a), (b) extensively in our review of
inertial instability of parallel flow. The action of mechanism (c) alone is
also well known, for it gives wave motions, such as sound. I n this section
we shall discuss qualitatively the interaction of mechanisms (a), (c). We
shall exclude mechanism (b) because it is subsidiary and complicates the
discussion. To understand the interaction better we shall relate the stability
characteristics under both mechanisms to those under each separately by
use of dimensional analysis.
To illustrate the use of dimensional analysis it seems clearest to consider
one specific case, and we have chosen that of the stability of parallel flow
of a fluid of variable density under the action of buoyancy, with stability
governed by equation (3.12),

The methods we shall use for.this problem can be readily applied to the
other stability problems of Section 111, which have a similar form. The
stability equation above shows that gravity occurs only in the product
- gD,p+/p+, for J ( y ) = - g(D,p,)La/p,Va. Therefore the eigenvalue
problem (3.12), (2.12) gives eigenvalues of the form

for the class of similar profiles w(y),p(y), where Jo* is the value of J + 3
-gD,p,/p, at any specified point y . Now dimensional analysis implies
that
(4.2) c = c(a,Jo)
where Jo = Jo+La/V~ acts as a characteristic value of J ( y ) . To solve a prob-
lem we find this relation explicitly, and, in particular, find the values Jo(a)
for which ci(a,Jo)= 0 but for which ci(a,Jo)> O nearby. These values
Jo(a)define the curve of neutral stability (or ~eutralcurve or stability boundary)
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 55

in the (a,Jo)-plane. This neutral curve is important, because it separates

the unstable from the stable disturbances and hence shows at once whether

-
there is instability or not for any given wave and flow defined by the point
(a,Jo)*
Now consider the limit as Jo 0 for fixed a # 0. Then (some of) the
eigenvalues
(4.3) c - c(a,o),
which limit we suppose to exist and to equal the eigenvalues c(a) of the
Rayleigh stability equation (2.11), which is equation (4.1) with Jo 0.
It should be borne in mind that internal gravity waves exist for all Jo >O,

eigenvalue with Jo = 0 for each eigenvalue as Jo -

however small, but not for Jo = 0; so we cannot expect there to be an
0. However, know-

-.
ledge of c(a) from the theory of section 11 now tells us the behavior of
some of the branches of the eigenvalues c(a,Jo) as Jo 0, i.e. as the buoyancy
becomes small, as the velocity scale becomes large, or as the length scale
becomes small.

-
To consider mechanism (c) alone let us take the limit as w* + 0 for an
unbounded flow, i.e. as V 0 for fixed w ( y ) , p ( y ) , L, Jo*, u,,. In this limit
we suppose that c* tends to a function which is independent of w*, and
therefore of both L and V , because as w* vanishes its length scale and shape
cannot be relevant physically. Therefore c* is some function of a*, Joe
which has dimensions of velocity. This implies that
c*-kJi$/a*, as V 40,
where k is a (many-valued) dimensionless constant independent of wCy)
but dependent on pCy). But when V = 0 and the flow is unbounded it is
well known that there are internal gravity waves whose speeds do depend
on p b ) . These speeds will give k. For example, when
P* = Po* exp (- BY*),
it can be shown [Sl] that
cs = (gS)"%* :
therefore k = 1 if we choose Jo* = gfi. (Of course the arbitrary multi-
plicative constant in Joe affects k, because it is only the product k Ji:/+"
that
is determined physically.)
To determine mechanisms (a), (c) together let us again suppose that
the flow is unbounded. We can now let the length scale L of the velocity
profile tend to zero without altering the infinite domain of flow. Thus we
let L - 0 while a*, V , w ( y ) , g, p(y) are fixed. Then a = a,L + O and
Jo = - gL(Dp/p)o/Va+ 0, although Jo/a is fixed. Thus if we write

c = c(a,Jo/a)
66 P. G . DRAZIN AND L. N. HOWARD

and let L - 0 we find

c -c(O,Jo/a) as a 40,
for fixed smoothly-varying profiles w ( y ) , p ( y ) and for fixed Jo/a. In this
same limit we find
w * ( r * ) = WY*/L) (-- m< Y* <
cr* > 0)
cr* = 0)
Vw(- w) w*(- -4 (Y* < 0)
Vy*/ly*l (for shear layers)
=(0 (for jets),
on ignoring the isolated point y+ = 0 which can have no physical signif-
icance. It is understood that
w*(- w) = - w*(..) =- v
for profiles of shear-layer type and that

w*(- m)=O=w *(MI

for profiles of jet type, as can be effected without loss of generality by a

Galilean transformation if necessary. Thus for a shear layer w* represents
a vortex sheet in the limit as L - 0 and for a jet w* represents no flow in
the limit. Similarly we find

(Y* > 0)
P*(Y*) + P*o (Y* = 0)
Q
'::[ (Y*<O)

-
as L + O if these limits exist.
Now let us review what happens in the limit as L - 0 for a profile of

W ( Y ) ' P M and also w* -

shear-layer type. We have found that then c c(O,J,,/a)as a 40 for fixed
vY*/lY*l, P* +P+olcy* >O) or P*-a&* < 0) for
fixed a*. However, we know the value of c* for the vortex sheet from Kelvin-
Helmholtz instability (3.9), which gives
C* = f V(Jo/a-
on choosing Jo = (Lg/V2)[p*-, - ~ * ~ ] / j 3 p*J +
* - ~and neglecting the
variation of density (but not buoyancy) of the fluid in the inertia. We
conclude that for any profile of shear-layer type
HYDRODYNAMIC STABILITY OF PARALLEL FLOW O F INVISClD FLUID 67

c + c(O,J,,/a) = (Jo/a- 1)1/2 as a -0

for fixed Jo/a. We shall confirm this result analytically in the next section.
For a profile of jet type we similarly identify c(O,Jo/a)as the speed of
internal waves when w* = 0 and p* = pa&* >O) or p*-&y*< 0). as
given in equation (3.11). Thus
c -c(O,Jo/a) = f (Jo/a)*/2 as a +O.

We shall also confirm this result analytically in the next section.

Similar dimensional arguments for each of the force fields discussed in
the last section can be used to apply the results for zero basic flow and for
a vortex sheet to profiles of jet and shear-layer type respectively with long
waves. In each case, on the large scale of a long wave ( L << a*-') every
shear layer looks like a vortex sheet and every jet like no flow; so their
stability characteristics should correspond as a .+ 0. Unfortunately these
arguments do not seem quantitatively correct for all force fields. For example,
in the case (Section 111.5) of a rotating system with variable Coriolis parame-
ter, we expect that for each smoothly-varying shear layer w ( y )the eigenvalues
c(a,a)-+ c(0,a) as given by equation (3.33) for the vortex sheet, where
a = @/a2,the Rossby wave speed. Thus it would seem that each shear
layer is unstable as ic -+0 for fixed a, as a vortex sheet is. However, an
exact solution for the shear layer w = tanh y seems [67] to imply that the
neutral curve touches a = 1 as a + 0, i.e. that there is stability for u > 1
as a + 0. This type of inconsistency occurs for some other force fields and
has not been satisfactorily resolved. Possibly the resolution may come

-. -
from there being more than one mode of instability for a smoothly-varying
shear layer, yet only one for a vortex sheet; again the limits c, 0, a 0
may not be uniform.

2. Physical Argtcmnts
The mechanism of instability of a vortex sheet w* = Vy*/ly*I in a
compressible fluid at uniform temperature will now be described, essentially
in the way attributed to Ackeret [cf. 74, p. 2401. Consider a small irrota-
tional two-dimensional disturbance of the velocitv field in which the interface
between the two streams of speeds V , - V is distorted. Thus the interface
has small bends. If the streams are subsonic ( V < a*), then by continuity
the speed on the convex side of a bend has a small increase over its basic
value, and the flow on the concave side a small decrease. Now Bernoulli's
theorem for irrotational unsteady flow of barotropic inviscid fluid plausibly
suggests that the pressure decreases on the convex side and increases
on the concave side of the bend. This pressure difference induced across the
bend increases the curvature of the bend and thus causes instability of
the interface. By the theory of the Laval nozzle [cf. 74, 8 3.61 the speed
58 P. G. DRAZIN AND L. N. HOWARD

is decreased on the convex and increases on the concave side of a bend if

the streams are supersonic (a,< V). Hence the trend is reversed and the
flow is stable. This heuristic argument indicates that there is instability
for all a >1.
This sufficient condition for instability of a vortex sheet is confirmed
by an analytic argument of Lin [63],which gave this condition for a smoothly-
varying shear layer subject to two-dimensional disturbances. However the
condition apparently contradicts the result (3.23) that the vortex sheet is
<
stable for all a 2-112 and unstable for all a >2-l12. The disturbances
considered for equation (3.22) are in fact irrotational on either side of the
vortex sheet, the rotational disturbances being part of the continuous
spectrum. Any contradiction may be due to the difference between the modes
of instability of a vortex sheet and of a smoothly-varying shear layer in
the limit as a + 0, a difference similar to that for flow with variable Coriolis
parameter discussed at the end of the last section.
Another physical argument may be applied to jets. We take the qual-
itative argument of Backus [cf. 8, p. 2641 for inertial instability of a homo-
geneous fluid, make it quantitative, and generalize it for a fluid of variable
density under gravity. Let us suppose the jet has profile w,(y,) where
w,(f do) = 0 in a fluid of basic density p&,) such that (D,p,)+m = 0.
We shall consider only long-wave disturbances of this jet.
For long waves the effective width L of the jet is much less than a wave-
length 2n/a,. Thus the jet oscillates sinusoidally like a string. Far away
from the core of the jet the flow is irrotational, because the basic flow is
uniform and the disturbance of finite origin receives no vorticity. Therefore
the amplitude F , of the oscillation of a particle path dies away exponentially
at y* = f do with scale height l/a,; the height of a material surface

as y, -.
above its basic level is 11, =F,(y*) exp(ia,(x, - cat,)} where F, -FO*e-OL*IY*I
f 00. The density is j,-=below the jet. Close to the jet, within
a distance of order of magnitude L from the jet, i.e. much closer than a
wavelength, the long waves seem locally like a vertical translation of the
jet. Thus the jet oscillates like a string with form
110, = FO, exp {ia*(z* - c*t*)h
for q , is approximately constant on any vertical line near the jet.
In this motion the vertical mass-acceleration of the fluid on both sides
of the jet is in balance with buoyancy and the centrifugal force due to the
(small) curvature of the jet. The buoyancy comes from lifting fluid below
the jet a height 7, into space previously occupied by the lighter fluid above
the jet and vice versa. In this way the buoyancy gives rise to a pressure
disturbance across the jet
= g{p*-a, - p*m}qo*s
to first order for small q,.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 69

The centrifugal force exerted by a volume element of the jet of vertical

thickness dy, and unit horizontal area is the product of its mass, curvature,
and the square of its horizontal velocity. namely j&dy*(- apLll+/ax*2)w+*
to first order in q*. This builds up the pressure difference across the jet

s
m

= a*2 P*W*%*dY*.
-m

Now over the effective width of the jet q+ = qo*, because a* is small; in
the distant regions where q* -+ 0 exponentially w* is small anyway. Therefore
this pressure difference

I
m

= a*%* P*w*2dY*
-m

for small a*, qs.

The vertical mass-acceleration of the flow over unit horizontal area

Now /3, changes rapidly from its value at the origin to its values
at infinity, whereas q,, changes slowly like Fo,e-"*lY*l +"*(+* - c*t*) . There-
fore, for small a*, this expression for the mass-acceleration
m 0

= - a*cla{jkm + P*-m)qo*.
Finally the balance of pressure and mass-acceleration per unit area gives

1.e.
60 P. G. DRAZIN AND L. N. HC'YARD

as a* + O for fixed
g{P*-m - ij*m}/V8a*@*-c P*mh

This result will be verified analytically for a sinusoidal disturbance in the

next section. In particular, we now see that long waves are stable when

g{A-m - P * m ) 2 a*
.jP*w*'~Y**
-m

In this argument we have approximated the buoyancy force with only

the change of density from one side of the jet to the other. Thus we have
neglected the modification of the buoyancy due to the density structure
of the jet, which should be of order of magnitude a*L times our first ap-
proximation g(j5*-m - pmm)qo*.This modification may lead to the addition
to equation (4.4) of a term larger than the last included term unless the
first term of the right-hand side of (4.4) is of comparable or lesser magnitude
than the second term in the limit as a,L + O .
Holmboe [76] has given other physical descriptions of the instability
of parallel flow of fluid of variable density under gravity. In particular,
he has looked at the development of symmetric waves in terms of real var-
iables rather than in the usual way with normal modes.

V. INSTABILITY
OF AN FLUID
INCOMPRESSIBLE OF 1 ARIABLE DENSITY

1. General Stability Characteristics

In this section we consider the instability of a basic steady plane par-
allel flow of an inviscid incompressible fluid of variable density under the
action of gravity. We take the basic velocity ii* = w*(y,)i and density
&, = p*(y+)as before, y,, being the height. Also we neglect the variation
of inertia due to the variation of density of the fluid, i.e. we take K ( y ) =
- Ldp*/p*dy, = 0 but retain J ( y ) G - L8g(dj5,/dy,)/p'*V* # 0. This is
similar to the Boussinesq approximation and can be justified for many
practical applications of the theory in which K is small and J of order one.
We have shown in Section 111, by the usual methods of hydrodynamic
stability with normal modes, that the instability is described by the di-
mensionless eigenvalue problem :
(6.1) (w - c)(D8- aS)p,- (D8w)cp + J(y)p,/(w- c) = 0;

(6.2) cp = 0 (Y= YIJY').

We shall consider general and particular properties of the eigensolutions
in the two subsections of this section, following the methods of Section 11.
As in Section 11, we have dynamically independent two-dimensional waves,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 61

each having a stream function of the form = ~ ( yexp +'

) {ia(x - ct)}.
The eigenvalue problem is again invariant under complex conjugation, so
there is stability only when c is real, and instability when c is complex,
one of the conjugate solutions growing like exp ( w i t ) .
We can proceed to generalize Rayleigh's theorem and some other results
of Section 11.2 as follows. Assuming that ci >O. let W = w - c, H
W"-'V, some definite branch being chosen when n is not an integer. Then
the stability equation becomes

. .
(6.3)
+ W-zn(n(I - n)(Bw)2- J ) ) H = 0.

Multiply this equation by H* and integrate from y 1 to y 2 to get

(5.4)
Y, T We(l-*l(lDHI2 + a2lH12) + nWl-"(D%m)IHJ2
+ W-%{n(l -~ ) ( D w-)J))HI2dy
~ = 0.

This result of Howard [23] can lead to various properties according to the
value of n chosen.
When n = 1 we have

(5.6) +
[ ( D q ~ l ~oc21pl12 + W-'(D2w)(cp(' - W-2J1~\adr= 0.
YI

The imaginary part of this gives

(5.6) cij{D2w - 2(w - c,)JJWJ-B}JWI-2)91)2dy~

= 0.
Y1

Therefore
(5.7) Dew = 2(w - c , ) J ( y ) / ( (-~ c,l2 + ci*}
somewhere in the field of flow. If D2w # 0 in the field of flow we further
have

(6.8) ci < ( W (< max ((1 < max ]2J/D%(.

- ct)JI(WllDBwl}
These results (5.7), (5.8) are due to Synge [76]. When J = 0 they give
Rayleigh's necessary condition for instability that D2w = 0 somewhere
in the field of flow. Unfortunately when J # 0 they are not so simple,
because they involve the unknown c.
62 P. G. DRAZIN AND L. N. HOWARD

When n = 0 we have H = y/W = F , and

(6.9) +
r ( w - c)*((DFI* a*lF(*)- JlFl*dy = 0.
Yl

This leads to the proof of the semicircle theorem, as in Section 11.2. The
extra term in the present case only strengthens the inequalities used provided
J 0 everywhere. Thus, when ci > O and J ( y ) 2 0 in the field of flow,
(6.10) {c, - a ( ~ & + Wna.)}~ + cis < { a ~ ~ m -a s wultn)}**
Howard proved this result [23] for a heterogeneous fluid originally. Even
when J < 0 somewhere, it follows that wulti,< c,< wmx. When J ( y ) 0 <
everywhere, equation (6.9) shows that no non-singular neutral mode can
exist, i.e. that either ci # 0 or c lies within the range of w and F is therefore
singular. However, when J ( y ) > 0 somewhere, it is possible that non-singular
neutral modes exist with c outside the range [w-,w-] of w ( y ) ; these
isolated neutral modes in fact occur as internal gravity waves.
When n = 4 we have

j.,
Y1
- c){lDHl* + a*lHl*}+ i(D2w)lHl*+ W-l(t(Dw)*- filHl*dy= 0.
(6.11)

The imaginary part of this gives

(6.12) - IDHIa + a31HIB+ IWl-*{J - ,(oW)*}1Hl*dy = 0.

YI

Therefore, when ci >0,

(6.13)
Y,
P P
0 > - lDHlady = (aa
9,
+ {J - f(Dw)*}/lW(*)IHl*dy.
Therefore J C y ) < $(Dw)* somewhere in the field of flow. This gives Miles'
[77] sufficient condition of stability that J - t(Dw)*should be everywhere
non-negative. Further from inequality (6.13) we have

(6.14) a*ci' <a*lWl*<max {i(Dw)*- I@)},

a result due to Howard [23].
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 63

The results of Section 11.2 about s-eigensolutions in the case J = 0

have no simple generalizations for the present case J # 0, because the
singularities of the stability equations for the two cases differ. We cannot
now form a Sturm-Liouville problem to find ips after choosing some particular
value c = I , . However the neutral s-solutions which occur for J = 0 will
be modified as J increases and give some stability boundary of the form
Jo = J s ( a ) ;where Jo is some characteristic value of J ( y ) ,such as
+
gW*(Yi*) - P*(Y2*)ll(P'*(~i*)L(Ya*)}B
for dynamically similar basic velocity and density distributions. If wCy),
J ( y ) are analytic functions in the real interval bl,ya], then the solution
p(y;a,Jo,c)of the stability equation (5.1) will be an integral function of
a, Jo, c over any fixed domain of y within b1,ys] which excludes a neigh-
borhood of the singularity w ( y ) = c, if any. It follows that the eigenvalue
c is a continuous function of a and Jo [78, p. 2111. This result, together
with the semicircle theorem in the limit as ci +O, implies that a stability
boundary consists of singular neutral modes, i.e. modes for which ci = 0
and w = c in (y1,ya) [77, p. 6061. Further, for a certain class of basic velocity
and density distributions at any rate, every singular neutral mode has a
contiguous unstable mode in the (a.Jo)-plane [78, $41. In general the
stability boundary Jo = J,(u) is both many-valued and has many branches,
as will be indicated by examples in Section V.2. As yet there is no
general theory to find this stability boundary, but it has been found for
many special velocity and density profiles.
However, supposing the stability boundary to be known, Howard [26]
gave a heuristic method to perturb it and find neighboring unstable solu-
tions. The method generalizes the argument leading to equation (2.20),
which was for the case J = 0. Proceeding as in that argument but with
fixed J ( y ) # 0, we find

I j
Y,

(6.16) d d l d c = (2J - WD%)papV2dy/ @dy,

YI YI
where q~ is any eigenfunction with eigenvalue c. When c is real, care must
be taken in evaluating these integrals because Q is singular. Examination
of equation (6.3) shows that p behaves like W1-" when W + 0, i.e. near
y = yc, f i being a root of the equation n(l - n)(Dw)a= J evaluated at
y = yc. We have found that J / ( D w ) ~ <f somewhere in (y1,ya)in order that
a singular neutral mode should exist; in that event there are two roots n be-
tween 0 and 1, which coincide at n = & when J/(Dw)8= t at y = yc. Thus

(da*/dc)b= lim
e+a,
j. (2JW- 1--zn- w-ymY+
(0%)
P ~ ( 1 * -w a y ,
Yr YI
(6.16)
64 P. G. DRAZIN AND L. N. HOWARD

where H = W”-lg, behaves smoothly at y = yc as a -c a,, c + w, for a

fixed function J ( y ) . For definiteness let us suppose that - n < arg (w - c)
c 0 for ci >O, so that

Now it follows that the denominator of equation (6.16),

(6.18) lim
a+a#
i
YI
qp
w2(1-
i
d y = WF(1- n)Hsadr,
Yl

which we are supposing to be known. The integral of the numerator of

(5.16) diverges at y = yc in the limit, so care is need to approach the limit
with ci 4 0 through positive values. In this way Howard [26] was able
to evaluate the right-hand side of equation (6.16) and thence find
(dclda), = (da/dc),-’ = 2ec,(daa/dc),-’.

We can use this result to find further stability characteristics. In general

we seek the function c = c(a,Jo) and thence criteria of stability. But by
partial differentiation we see

We suppose that the stability boundary ci(a,Jo) = 0 is known and gives

Jo = JS(a). Thus, on that boundary,

(5.20)

In the previous paragraph we have shown how to determine (ac/aa),-from

knowledge of the neutral s-eigensolution, so we now can find c(a,Jo) on the
unstable side of the neutral curve. Howard [26] derived these results and has
applied them to two examples of shear layers.
The dimensionless Reynolds stress is t = faj(g,*Dq --‘g,Dq*)eeaCi’,as
for a homogeneous fluid, where here j varies with y but its derivative is
neglected except in the buoyancy term, i.e. KCy) = 0. Whence one finds
from the stability equation (5.1) that

(6.21)
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 66

The boundary conditions imply that T vanishes a t y = y1,y2. Therefore

&lay vanishes somewhere in between, which when ci # 0 gives Synge's
generalization (6.7) for heterogeneous fluid of Rayleigh's necessary condition
for instability. Equation (5.21) also shows that, when ci = 0, T is constant
except for possible discontinuities where w = c. These discontinuities do
not occur for non-singular modes with c outside the range of w ( y ) . For
monotonic profiles w ( r ) only one discontinuity y = yc is possible, but the
boundary conditions give t = 0 on either side of the possible discontinuity,
so t = 0 everywhere.
The occurrence of modes when J f 0 is somewhat like that for the case
when J 0 discussed in Section 11. Unstable modes that exist for J o = 0
continue to exist as J o increases from zero. By and large, increase of Jo
decreases their instability, as would be anticipated from the physical effects
of buoyancy. When J ( y ) / ( D w )> a $ everywhere all modes are stable. In
addition to modification of the modes present when J = 0, variation of
density gives rise to new modes. These are the internal gravity waves,
which are isolated modes not associated with instability. Profiles with
even functions w ( y ) , J ( y ) f 0 have sinuous and varicose modes as when
J 0. For odd functions w ( y )with even function J ( y )there is often exchange
of stabilities with c, = 0 when ci >O. However, there may be exceptions
when the unstable mode is not unique [26]. Similar arguments to those
valid when J 0 may be applied to problems when J $ 0. We shall
illustrate them by examples in the next subsection.
Drazin and Howard [79] have considered the stability characteristics

-
of unbounded flow for long waves. Their method is a natural generalization

-
of that for a homogeneous fluid with J = 0. The eigenfunction must be
such that p, constant x exp (Fay) as y f m, when DP 0 smoothly
-+

at infinity. Proceeding for this case J f 0 in the manner of Section 11.2,

one can show that the eigenvalue relation for small a is

(6.22)
0= a(W-,
2
+ Wma) - 2J0+ 1
--m
{a(W2- Wm2)

+ Jo(l - A)){a(W2- W t m )+ Jo(l+ A)}W-*dy + .. .;

where
(6.23) A(Y) {p-m + Fm - 2P(Y)}/{F-m - F m }
and
Jo gL(F--m - P m ) / W P - m + Pm),

so J(y) =JODI. For profiles of shear-layer type with wfao = f 1, this

relation gives
(6.24) c2 = Jo/a - 1 ,..+ as a ---* 0 for fixed Jo/a.
66 P. G. DRAZIN AND L. N. HOWARD

This result in the limit as a --+ 0 agrees with that of (3.9)for Kelvin-Helmholtz
+
instability of a vortex sheet when (p-m - Pm)/(P-m Pm) << 1 , i.e. when
K ( y ) 0. For profiles of jet type with wim = 0, relation (6.22) gives

1
m

ca = Jo/a - aa +
( W z - c ~ ) ~ / 2Jo(Wa
W ~ - ca)/aWa
--m
(6.26)
+ Joa(l - 12)/a2W2dy+ . . . .
When w E 0 everywhere this relation in turn gives the speeds of the internal
gravity waves. For w(r) not identically zero, these internal gravity waves
are modified and become isolated stable modes. There are also a t least
two unstable modes for sufficiently small a,Jo. When Jo is of order aa,
equation (6.26) gives the sinuous mode with

(6.26) as a -0,
-m

in agreement with the physically deduced result (4.4)when

So the sinuous mode has stability boundary with

m

(8.27) wady as a-0,

--m

there being stability to long waves when Jo is greater than this value.

as y -
It is physically realistic to have a model for which p ( r ) tends to constants
f 00. However, in many circumstances it is practical to suppose
p tends to zero or infinity. For example, one might suppose that P = po
exp (- By) for constant p > O to represent the atmospheric density in a
problem of instability that occurs effectively in a finite range of y ; in this
case J ( y ) = g p L Z / V ais independent of y . We have shown (3.1) that then
internal gravity waves have speed a = J1I2/a. I t can also be seen from the
stability equation (6.1) and boundary conditions (6.2) at infinity that

(6.28) q-constant x exp (T aL*y) as y --+ f 00,

y - +
where L , = (1 - U ~ / W ; , ) ' / ~
provided
, that w(r) --+w,, smoothly as
f m. Now a direct generalization of the method of Drazin and Howard
[79] described above gives for small a and fixed a the eigenvalue relation
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 07

0 = L+wma +L-w-, + a
2

J J
- w 0

- L+L-Wm'
-m
5
0

1 - Wt-mjw'dy} + ... .
For a shear layer with I-, = - w , = - I, this relation in the first ap-
proximation gives the same stability characteristics (3.15) as a vortex sheet.
Higher approximations are obscured by the difficulty discussed at the end
of Section IV.2 for a compressible fluid. For a jet with w + = ~ 0, relation
(6.29) gives

(5.30) 0 = 2Lcs +a
5
--oo
(Wa - ca)(1 - L*c2/Wa)dy + ... ,
+
where L = (1 - U ~ / C ~ ) ) ' /In
~ . the first approximation for small a this
gives an unstable mode with

(5.31) c--li(a/u)
2
1
W

--m
wady as a -0.

and an internal gravity wave with

(5.32) ca - a' N ..!

4
(ag/aa) ( 1w' - Sawdy as a -0.
--m

In this special case with constant J and unbounded flow, we can find
a sufficient condition for stability. We have required that the real parts
of L , be non-negative in formula (6.29) in order that (p does not exponentially
increase at inwity. In fact the solution of tb initial-value problem must
die down as y + f 00, so isolated modes may have non-zero or even un-
bounded eigenfunctions at infinity, but any dense set of waves must have
eigenfunctions which tend to zero there. Therefore, when c is real, only
isolated waves may be not exponentially damped as y + f 00. It follows
that in the limit as ci --* 0 , L+ and L- are real and non-negative. Therefore
68 P. G. DRAZIN AND L. N. HOWARD

1 - aa/(w, - c)a, 1 - aa/(w-= - c ) 2 ~ 0 for eigenvalues c on the stability

boundary. Therefore a* <(w, - C ) ~ , ( W - , - c)* on the stability boundary.
Therefore a sufficient condition for stability is that

(5.33) aa> max {(wm- c)~,( w - m - c)~}.

Now the semicircle theorem gives wmin< c < w-,. Therefore another
sufficient condition for stability is that
(6.34) a2> ( W m a - w*p.

2 . Stability Characteristics of Various Basic Flows

In Section 11.4 we have given some stability characteristics of several
basic flows of homogeneous fluid. In Section 111.2 we gave the stability
characteristics of heterogeneous fluid in a state of rest and in the motion
of a vortex sheet. In this subsection we shall exemplify the interaction of the
inertial instability of some other flows of Section 11.4 and the effects of
buoyancy due to various basic density distributions.

(a) InterHal Gravity Waves

The stability of heterogeneous fluid in a state of rest (w = 0 ) is
governed by buoyancy alone, there being no shear in the basic flow. This
problem is simpler, being a regular Sturm-Liouville problem. If the hasic
density anywhere increases with height there is instability. Otherwise
neutrally-stable internal gravity waves occur. Their structure and speeds
depend on their wavelength and the distribution jib) of basic density.
These waves also occur as isolated modes for velocity profiles with shear,
and are treated in detail in some of the papers we refer to. However, by
and large, we shall exclude any detailed treatment of internal gravity waves
from this review. Here we shall merely refer the reader to the classic paper
of Fjeldstad [SO] and to a more recent survey of the literature by Davis
and Patterson “1.

(b) Plane Couette Flow

The stability of a heterogeneous fluid in the basic flow with velocity

w = y(yl < <
y ya) was first considered by Taylor [SZ]. He took p =
po exp (- by) so that J ( y ) is a constant, and then showed that the stability
equation was essentially Bessel’s equation of order v (& - J)l/*. With
one of the boundaries at infinity, he found there were no eigensolutions
when O < J < 4 and only stable ones when J > &. Taylor [SZ,9 41 con-
sidered also the stability of plane Couette flow with three and four layers
of homogeneous fluids of different densities.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 69

In a more complete investigation Eliassen et al. [36] have shown that

when p = p,, exp (- By) plane Couette flow, whether bounded, semi-bounded,
>
or unbounded, is stable if J 0 and unstable if J < 0. They considered
the initial-value problem as well as normal modes. When 2 < J < if there -
is no discrete spectrum of normal modes. These results seem natural in
view of the known stability of plane Couette flow when J = 0 and of the
anticipated stabilizing influence of buoyancy.
Also Heriland [82] has considered stability of plane Couette flow when
J is a quadratic function of y.

(c) Sinccsoidal Flow

When
(5.35) w = sin y , J = constant (rl <y <ye)
certain exact neutral parts of the solution are known [83] for 0 Q a Q 1 .
The stability equation (6.1) has the following solutions.
(6.36i) c = 0, J = (1 - a2)1/z- 1+ az, q~ = lsin ~ l ( l l Z ) + ~ ;
(5.36ii) c = 0, J = 3(1 - aZ)ll2- 3 + a2, q~= cosy [siny1(1/2)+v;
(6.36iii) c = 0, J = )(aa - t), +*Ices 4 ~ 1 1 -( 1 / 2 ) v ;
q~ = Isin +yl(llO)
where Y G (& - J)’“.
For y 1 = O,ya = z the flow is known to be stable when J = 0. However,
solutions (i)-(iii) are all eigensolutions. It is presumed [83] that the associated
neutral curves in the (ct,J)-plane are not stability boundaries.
For y , = - z,yg = 3t there is instability for J = 0 when O< a < 3llZ/2;
the limiting eigensolutions being
c=O, a=O, g ~ = s i ny and
(6.37)
c = 0, a = 3‘l2/2, q~ = cos 4y.

Again (5.36 i-iii) are eigensolutions, but it seems [83] that none is a
stability boundary.

(d) Thin Jet

Applying the method of derivation of “jump” conditions (2.16) to the

stability equation (5.1), one can show that when
(6.38) w = (bCy))”Z (- -< y < w)

in a heterogeneous fluid, the conditions a t y = 0 are that

(6.39) cg [OF] + Jo [A]F= aaF, [F]= 0,
70 P. G. DRAZIN AND L. N. HOWARD

where

F q/W# J ( y ) J$A 1 5 {p-m + pm - 2p(y)}/{p-m- j m } .

With these conditions one can solve the stability equation (6.1)piecewise
for y >0 and y < 0, join up the solutions at y = 0, and find the eigenvalue
relation. When

the eigenvalue relation can be shown to be

(6.41) ca = - t a + Jola.
It can be seen that this result also follows exactly from relation (6.26),and
gives stability of the thin jet when Jo 2 ia*. This result is typical of the
stability of jets to long waves, and will have to serve for other results on
jets. which the literature lacks.

(e) SLar Layer

The stability of heterogeneous fluid with basic flow

-1 (Y< -1)
(6.42) W = l Y (-l<y<l)

1 (1< Y )
was first considered by Taylor [62, $31 and Goldstein [63,$53, 61. They
took essentially

i
8-m (ye-1)
(6.43) p= A (-l<y<l)

Pm ( 1 <Y )
and found the eigenvalue relation. When p,, = i@-= + p-) and
+
(p-- - pm)/@-aD pm)<< 1, there is instability if and only if
(6.44) 2a/(l + e-%) - 1< J0< 2 a / ( l - e-*) -1
where Jo -ggL@-, - &)/V*(p-, +
pm). When Jo = 0 this reduces
to Rayleigh’s result [cf. Section II.4.k] that there is instability if 0 < a < a, =k
0.64. For general values of Jo,a the stability boundaries etc. are shown in
Figure 6, after Goldstein. It can be seen that any wave unstable when
J o = 0 becomes stable when Jo is sufficiently large. However, other waves
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 71

r
are made unstable as Jo increases, some n a o w band of waves being unstable
however large Jo is.
Goldstein [63,5 61 further considered the shear layer with the continuous
density distribution
(Y< - 1)
(5.6) p= p-me-b(y+l) (- l < y < 1)
p- w (1< Y)l
where again j? is negligibly small except when multiplied by gravity.
Goldstein’s solution, involving Bessel functions, is complicated. In brief,

I I I
‘ I

it gives stability to all waves if and only if Jo g,8,La/Vn >, t. This condi-
tion is in agreement with Miles’ sufficient condition J/(Da)P>, for stability.
When a << 1, it can be shown from Goldstein’s work that is there is stability
when
J~ 2 - 2 - -as
4 - -a4
I6 -
(6.46)
3 9 46
...,
in agreement with relation (6.22). Recently Miles and Howard [a] have
clarified an obscure point in Goldstein’s paper and given some numerical
72 P. G. DRAZIN AND L. N. HOWARD

results for this example. The principal stability characteristics are shown
in Figure 6.
Holmboe [78] studied the model of Taylor and Goldst+ with density
distribution (6.43) by his method of symmetric waves. Holmboe also con-
sidered density distribution (6.40)with a single discontinuity at y = 0.

0.8

0.7 -

a
0.5 -

0.4
- UNSTABLE

0.3-

0.2-

JO

characteristics of the shear layer with w = y / l y l ( l y l > 1). w =

j = p-,&< - 1). j = j-me-fl(y+l)(lyl< 1). i = P-me-w*
2 4 16
=a--aB-- a8 - - a 4 . (b) Stability boundary.
3 46

Then the eigenvalue relation [76, equation (7.6) essentially] can be shown
to be
(5.47) 4 a V - c a{(2a - 1 ) 2 - ,-*a + +
4aJ0} (Jo/a)(2a- 1 e-k)a = 0. +
I t follows that ca is complex only when
(2a - 1 +3 e - 9 - 2{2e-k(2a - 1 + e-h)]l/g

(5.48) < 4aJo/(2a- 1 + e-&)

c (2a - 1 + 3e-&) + 2{2e-”(2a - 1 + e-k)}l/e.
The curves representing equalities above are shown in Figure 7. Both
curves touch Jo = a at the origin and Jo = a - 1 at infinity. For values
of Jo,a between these curves, ca is complex and therefore c,,ci # 0, i.e. there
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 73

is overstability. On the curves, cz = f (J0/2a3(2a - 1 e-ea). It follows +

that there is stability (c2 >O) between each curve and an axis except be-
tween the upper curve and the a-axis for 0 < a < as = 0.64. In that region
c, = 0, c, # 0, there being exchange of stabilities a t a = 0, a = a, when
Jo = 0. This is accordingly an example of a flow with odd function w ( y )

-
1 , 1 , 1 , 1 , I . I
0 0.2 0.4 0.6 0.8 1.0 1.2
JO

FIG.7. Stability characteristics of the shear layer with w = y/lyl(lyl> 1). w =

= y ( l y l < 1) and j = i O D ( y > 01, = j-=(y< 0).

and even J ( y ) where there is not everywhere exchange of stabilities, this

being possible because there is not a unique mode of instability.

(f) Double Shear Layer

Another flow for which w is an odd function and J even, yet for which
there is not exrchange of stabilities has been pointed out by Howard [26].
When
-1 PO(1 +4
(6.49)

cr> 1)
and J,, gLs/V*,s<< 1, it can be shown that
(5.60) (2c2 + 1 - J o / a ) g- 4c2(1 - e-kl) = e-"(1 - Jo/a).
Howard [26] found stability with four real roots c when

(5.61)

On the stability boundary, the curve with equality in the above,

c = f (1 - e-*a)@(l - e-h)1/2 + 2 - 2e-49-112.
14 P. G. DRAZIN AND L. N. HOWARD

There are waves with c = 0 ; in fact they occur when Jo = a ; however,

this locus lies within the stable region and is not adjacent to parts correspon-
ding to instability. These results are illustrated in Figure 8.

JO

FIG. 8. Stability characteristics for the double shear layer w = y/!y)(lyl> l),
I= O(lyI < 1) with density = pa(’ + e ) ( y c - l), = po(lyl < I), p =
po(l--E)@> 1).
(g) Sicliley Jet
Howard and Drazin [83] have found various exact parts of the neutral
eigensolutions for
(6.52) w = sechly (- oo<y< oo),
with various density distributions.
When J ( y ) is constant, the following solutions may be verified.
(5.52i) Sinuous mode c = (6 + as)/16, J = aa(4- aa)(Q- aa)/225,
p = (sechsy - c)h(sechy)m (0< aa < 4)
where
k +
3(4 - aa)/2(6 aa), m -= 6aa/(6+ aa).
(6.62ii) Varicose mode c = (3 + aa)*/3(3+ 5x3,
J = aa(l - a*)(9- aa)(3+ a*)”9(3 + 6aP)%,
tp = tanh y(sech’y - c)*(sechy)” (0 <aa < 1)
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 76

I ' ' where

k 3(1 - aa)/2(3 as), +
m = 4aa/(3 aa).+
1.5 -
The neutral curves are sketched
in Figure 9(a). Howard and
1.315- - Drazin [a]argued that the
curve for the varicose mode (ii)
Q
is a stability boundary, but that
- at most the upper part of the
curve for the sinuous mode (i)
is a stability boundary.
When J = J o sechay , eigenso-
- lutions may be verified as follows.
1
(6.63i) Sifiuous mode c = - a ,
3
1
Jo = 3aa(2 - a)(3- a),
STABLE

0 0.05
JO
0.1 .K)9 .I27
q = (sechy)=(sechPy- +a)*- (1/2)a
Fig. 9a.
(0 < a < 2).
(6.63ii) Varicose mode
+
c = (3 aa)/3(1 a). + STABLE

Jo = a ( l - a)
* (3 - a)(3 +
aa)/Q(l a ) , + 1.5 -
q = tanh y(sech y)'
.(sechay - c)(l@)(*-a)(0 < a 1). < UNSTABLE

The neutral curves, shown in 1.157-

Figure Q(b),are both thought [83] a

FIG. 9 (a). Stability characteristics

for Bickley jet w = sech* y when
J = const. (i) Neutral curve for
sinuous mode: J = aa(4 - a*)
-
( 9 a*)/226. (ii) Neutral curve for 5-
-
varicose mode: J = a*(l a*)(9 - as) *
+
(3 a9'/9(3 &*)a. + (b). Stability
characteristics for Bickley jet w =
sechs y when 1 = Josech*y. (i) Stabil-
ity boundary for sinuous mode: Jo =
-
-
a*(2 a)(3-a)/% (ii) Stabilitybound- 0 0.1 a156 a 2 a231
JO
ary for varicose mode: J o = a(l - a )
. - +
(3 a)(3 a*)/9(1 a). + Fig. 9b.
76 P. G. DRAZIN AND L. N. HOWARD

to be stability boundaries for their respective modes. Further information

on this example, and others, can be found by use of the formula (6.26)
for small a and the perturbation (5.20).

(h) Adisymmetric DoMble Jet

When
(6.64) w = sechy tanhy, J = Josechsy (- ce< y < ce),

JO

FIG.10. Stability boundary of one mode for to = sech y tanh y, J = J,, sech*y ( - 00
< y < do): Ja = a9(9 - a9)/9.

it can be seen [83] that a neutral eigensohtion is

1
c = 0, Jo = 3a2(3 - a*), 9 = (tanhy)l-(lls)='(sechy)l+(l/*)"
(0 < a* < 3).
This seems to be a stability boundary of one of the modes. It is shown in
Figure 10.

(i) Hy$erbolic-Tange& Shcav Layer

Various exact neutral solutions have been found for various density
distributions with velocity profile w = tanh y(- 00 < y < 00).
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 77

When J is constant, corresponding to basic density of the form p =

exp (- Py), Drazin [86] verified that an eigensolution is given by
c = 0. Jo = a2(l- a2), q = (sechy)@ltanhy1' -
(6.66)
(0 < a < 1).

,
It gives the stability boundary, shown in Figure ll(a).

a
0 UNSTABLE

0.I 0.2
J JO

FIG. Illa). Stability boundary for w = tanh y, J = constant (- m < y < a): J =
= aa(1 - a*) (b). Stability boundary for w = tanh y. J = Josech3 y(- ao< y < ao):
Jo = a(l - a).

When J = Jo sech2y , corresponding to ji = po exp (- fi tanh y), Holm-

boe [cf. 781 found the eigensolution

(6.67) c = 0, J o = a(1 - a ) , q = (sechy)altanhyJ1-a (0< a< 1)

for the stability boundary shown in Figure ll(b). It is somewhat similar

to the boundary of the example in Section V.2.e shown in Figure 6.
When J = 3Jo sech*y tanh2y, corresponding to density 13 = po
exp (- fi tanh3y), Garcia [cf. 76, 781 found the eigensolutions,

c = 0, Jo = +a(a+ 31, q = tanh y(sech y)";

(6.68)
c = 0, Jo = &(a- l)(a+ 2), q = (why)'.
78 P. G. DRAZIN AND L. N. HOWARD

These define a stability boundary, shown in Figure ll(c). It is somewhat

similar to the boundaries of the example in Section V.2.e shown in Figures 6
and 7. Some waves are unstable for each value of Jo, however large. Thus the

STABLE

0 1 2 3 4 1 6
JO

FIG. 1 l(c). Stability boundary for ru = tanh y , J = 3J0 sech' y tanh9 y ( - 00 <y< 00) :
+ +
J o = a(a 3)/3 and J o = (a - l ) ( a 2)/3.

flow cannot be stabilized. This occurs because J ( ~ ) / ( D W

vanishes
)~ where
w = c, i.e. at y = 0, and therefore cannot be everywhere larger than for
sufficiently large Jo.
Miles [78] has considered combinations of the above two density distribu-
tions, with local Richardson number J ( y ) = Jo(l- r 3r tanh8y) sechsy. +
Thus r = 0 corresponds to Holmboe's case above, and Y = 1 to Garcia's.
Miles showed that, when Y >i,the relation Jo = Jo(a) on the stability
boundary is no longer single valued, because the neutral curve turns away
from the a-axis. Further, when 0.896< I < 0.968, there are two distinct
branches of the stability boundary. As I + 1, the number of distinct
branches increases to infinity.

(j) Free Surface Flows

A class of flows with important applications comprises those with a

free surface, i.e. with p = 0 for y 2 0, say. Esch [86] has considered a few
examples of this class with further variation of p ,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 79

VI. STABILITY FLOWS

OF OTHERPARALLEL

1. Discussion

So far we have considered the stability of plane parallel flows, with

basic velocity ii*(y*)i, to which in fact the literature is largely confined.
There is a little work on the more general parallel flows, with basic velocity
tZ*(y*,z*)i, partly on the basis of known properties of plane parallel flow
[87], [88], partly by use of general properties like the energy and vorticity
integrals [89], [64]. We give an example of this (the proof of the semicircle
theorem) in Section VI.2. However it seems essential to reduce the linearized
partial differential equations of motion to ordinary ones in order to analyze
them thoroughly. For plane parallel flow, this is possible by the techniques
of transforms and normal modes, discussed in Section 11.1. It seems that
the only other class of parallel flows for which this is possible is those with
+
an axisymmetric basic velocity zi*(r*)i, where r+2 = Y , , ~ z,2 -the round
jets. Although their instability has some significant differences from that
of plane parallel flow, both the mechanisms and the mathematics are similar
in the two cases. It thus seems sufficient for our purposes to present only
a brief account of the instability of axisymmetric jets in an inviscid in-
compressible fluid. This is given in Section VI.3.

2. The Semicircle Theorem for General Parallel Flow

The most general discussion of the semicircle theorem appears to be

that given by Eckart [64], who has derived it for compressible flow, with
gravity, which is parallel (w(y,z)i) or circular (w(r,z)$). However since
Eckart’s notation is rather personal, we give here a sketch of another version
of the proof with more traditional terminology, restricting ourselves to
parallel incompressible flow though the compressible case is almost as
easy. Our proof is essentially the same as one constructed by H. Schade
and Howard (1963), and independently by Hocking [88]. We mention,
also that while Eckart’s proof seems to cover about as general a case as
one might expect to find, there is a t least one other case in which the semi-
circle theorem holds: non-parallel flow which is parallel and uniform in
layers, but varies both in magnitude and direction from layer to layer-for
example the Ekman boundary layer flow.
We assume that the flow is in a cylindrical region {- m< x < m,(y,z)
in S}, where S is some connected region in the (y,z)-plane with a sufficiently
smooth boundary. The basic flow is w(y,s)i; since the coefficients of the
stability equation are independent of x and t we look for normal modes
of the form f(y,z)exp ia(x - c t ) , just as in the plane case. The stability
equations become :
80 P. G. DRAZIN AND L. N. HOWARD

(6.1) ia(w - c ) d + v'. F'p + ia$' = 0

(6.2) ia(w - C)V' + P2$'= 0
(6.31 i a d + P2aV' = 0

where V2 is the transverse gradient operator, 9' is the perturbation pressure

divided by density, and I0 and v' are the longitudinal and transverse parts
of the perturbation velocity vector. The boundary conditions are v'. n =
O,n being the normal to the boundary B of S; from (6.2) this can be expressed
instead as

Using (6.2) to eliminate v' and (6.3) to eliminate zb', (6.1) isreadily transformed
into:

(6.6) I729 [(w - c ) - V 2 p ' ] - Ctyw - c)-2$' = 0.

Multiplying by p', integrating over S . and using the boundary condition

(6.4) we obtain

This is of the same form s(w - c)aQ = 0 with Q > O as in the plane parallel
case, and the semicircle theorem thus follows immediately, as before.

3. Ifiertial Isstability of Axisymmetric Jets

The work of this subsection is analogous to that of Section I1 on inertial

instability of plane parallel flow. By reference to the motivation and methods
of that section, we may state results briefly here. We take the basic axi-
symmetric parallel flow of inviscid incompressible fluid, with velocity
1* = U,(r,)i Q Y* Q r2*). This represents a jet between the rigid
coaxial cylinders r* = y1*,rB*, where y1* may be zero and yB* infinite. It
is again convenient to choose dimensional scales V of U*(Y*) and L of its
variation, and to render all variables dimensionless by scaling. Then the
basic flow of the jet is
(6.7) 1 = U(r)i (rl < <ra).
Y

With use of cylindrical polar coordinates ( x , ~ , ? ) and associated velocity

components ( N % , N , , U ~ ) ,the equations of motion may be linearized much
as before to give the perturbation equations,
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 81

+
at4;lat -I- vauzi/ax u;du/ar = - apilax,
aw;lat + uau,i/ax= - apt/&,

a q a t + uau;lax = - aptIraq,
a q a x + a(Yu;)lra + auci/ray = 0.

With the method of normal modes one assumes that

ull.u,',upl,p' = Re [{F(r),ZG(r),H(Y),P(r)}e'W+++--)],
where a is a real wave-number, c a complex velocity, '~tany integer which
represents the azimuthal Fourier component, and F , G, H , P are eigen-
functions to be determined. Then the linearized equations reduce to the
ordinary differential equations,
a(U - c)F + (DU)G= - aP,
a(U - c)G = DP,
a(U - c)H = - nP/r,
aF + DG + G/r + nH/r = 0 ,
where now D 3 dldr. On elimination of F , H , P one may get the single
linear ordinary differential equation for G:
+
(6.8) D{rD(rG)l(n2 a%')} - G - (rC/(U - c))D(yDU/(n2+ a%,)} = 0.
The boundary conditions are that the normal velocity u,' vanishes on
the coaxial cylinders r = vl,rp Therefore, in general,
(6.9) G =0 (Y = y1,rg).

However, when Y, = 00 we require that all perturbations vanish there in

order that the energy of the disturbance of finite origin be bounded. For
this it is sufficient that G --* 0 and is well behaved as Y + m; and thus
we may use condition (6.9) at infinity. When = 0 the continuity of
u,p implies that u,',$' are independent of ip and that u,',u+,' vanish at r = 0
(except for n = 1, when u,' and u i need only be bounded, but G(0) +
H ( 0 ) = 0, from the continuity equation). Therefore F(0) = P(0) = 0 (n # 0 )

-
and G(0) = H ( 0 ) = 0; and thus we may use condition (6.9) at Y = 0 except
when n = 0. In fact it can be seen that in general equation (6.8) gives
G N constant x r"-' (n# 0) and G constant x Y (n = 0) as Y + 0.
If U or DU is discontinuous, at yo say, then the pressure must be con-
tinuous a t the material interface with mean position r = y,. It follows that
(6.10) [(U- c)D(YG)- (DU)(rC)]= 0 (r = Yo).
82 P. G. DRAZIN AND L. N. HOWARD

Also the normal velocity must be continuous a t this material interface.

Therefore
(6.11) [G/(U - c)] = 0 (Y = YO).

Note that when U is piecewise constant it is easier to work with the amplitude
@ = JGdr of the velocity potential rather than with G directly [W].
The eigenvalue problem ( 6 4 , (6.9) is essentially due to Rayleigh [9l].
It has prompted surprisingly little later work, in view of the scores of papers
on the analogous problem (2.11), (2.12) of plane parallel flow. Perhaps the
similarity of the two problems and their methods of solution has discouraged
duplication of work, perhaps the greater physical importance of plane parallel
flows has overshadowed that of round jets. However, there is one important
difference between the two eigenvalue problems, namely the essentially
three-dimensional nature of instability of a round jet. Experience of plane
parallel flows suggests that varicose instability (axisymmetric disturbances
with n = 0) of a round jet should be less than sinuous instability (n = l),
so it comes as no surprise to find that there is no analogue of Squire’s the-
orem. In fact, in a recent examination of non-axisymmetnc disturbances,
Batchelor and Gill [go] found that a certain jet is most unstable to the
mode t z = 1.
The eigenvalue problem ( 6 4 , (6.9) is symmetric in a and (- a), so we
can again take a 2 0 without loss of generality. There is also a symmetry
in G,c and G*,c* for the same a, so c is real for stability and complex for
instability. Again, we write c, >0 when there is instability, bearing in mind
the initial-value problem and the inviscid limit of the viscous problem
P21, C90, §23.
Rayleigh [91] found a necessary condition for instability, analogous to
there being a point of inflexion in the velocity profile of a plane parallel
basic flow. Essentially by multiplying the stability equation (6.8) by
&*/(U - c), integrating from Y , to Y,, and taking the imaginary part, he
found that

(6.12) ci lglzDQdr = 0,
TI

+
where g E rG/(U - c ) , Q = r(DU)/(na a%*). Therefore a necessary
condition for instability (ci >O) is that DQ = 0 somewhere in the field of
flow. This is equivalent to U having a point of inflection with respect to the
+
variable p = s(n* aara)/rdr= n8 log Y +
)a%*. This reduces to Ray-
leigh’s condition for plane parallel flow if one regards the round jet as being
plane parallel flow locally when y1,r2-+ 00 and Y, - y1 = y , - y1 is fixed.
The following general stability characteristics are due to Batchelor and
Gill [go], who give details of the proofs.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 83

On multiplying the stability equation (6.8) by YG*/(U- c), integrating

from r1 to r,, and taking the real part, they found that

(6.13) 1
Is

rl
( g ( W- c,)DQdr < 0.
Therefore, when ci > O . equation (6.12) gives

(6.14)

U,being the value of U ( r ) at r = r,, where DQ = 0. Therefore, when DQ

changes sign only once, a necessary condition for instability is that
<
(U - U,)DQ 0 throughout the flow.
The semicircle theorem follows much as in Section 11.2, giving
(6.15) {cr - i(wmin + ~max)}' + ci2 < {i(~max - wmin)}a (Ci>O).

It can be shown that the Reynolds stress tensor, averaged over one
period 2n/a of x and one 2n/n of v, has orthogonal components

(6.16) tlr
~

y'u,' E (n, + u%a)-1/2(nu;+ aru,')u;

(6.17) = h-l(n2 + azra)1/eWe20LCi',
where
W = ir(n2 + aW)-I((rG*)D(rG)- (rG)D(rG*));
and

(6.18)

-
- nc,(DU)lGla
(6.19)
2a(n2 + aPra)l/z{(U - c,), +cia}
The stability equation (6.8) gives

2c,ra[GlaDQ
(6.20) DW=-
(U- c,)' + Cj' *

I t follows that, as ci - 0 through positive values, W is piecewise constant

and
(6.21) [W]= - ~z(Y'~G~'DQ/DU),,,~,
84 P. G. DRAZIN AND L. N. HOWARD

where U(r,) = c, in the limit, provided that (DU),=#, # 0. Now W = 0

at r = rl,ra, Therefore, in the limit as c,--+0,W = 0 at Y = r1,r2. There-
fore, in the limit as ci +O,W = 0 everywhere if U = c, at only one point
1= r,, or, in particular, if U ( r ) is monotonic. In that event [W] = 0, and
therefore either DQ = 0 or G = 0 at I = r,. The latter equality is compatible
with the stability equation (6.8) only if G f 0. Therefore DQ = 0 at Y = Y,,
i.e. r, = Y/ and c, = Up
If we put c = U,and look for neutral solutions that are limits of unstable
solutions, the task is more difficult than that of (2.18). However, Batchelor
and Gill [go] showed as follows that there is no such singular neutral solution
for sufficiently large B . Equation (6.8) now can be written as

Therefore, if

the solution (rC) of the stability equation will be monotonic and cannot
satisfy both boundary conditions. Thus a necessary condition for the exist-
ence of the singular neutral solution is that f i is not so large that
max {rDQ/(U,- U)}< 1. In fact this condition is quite restrictive.
Very few examples have been treated in- the literature. First we take
the exact solution of the Navier-Stokes equations for a viscous fluid as our
basic flow, namely Poiseuille flow in a pipe with
(6.23) U = Ara + B log r + C (rl < r < r2).
Rayleigh [91] investigated the stability of this basic flow in an inviscid
+ +
fluid. I t gives Q = (2Br4 C ) / ( n 2 a W ) , which varies monotonically
with r . Therefore the flow is always stable.
For the cylindrical vortex sheet,

(6.24)

Batchelor and Gill [go] used the velocity potential on each side of the
discontinuity to deduce that the eigenvalue is
(6.26) c = (1 + iLnl’z(a)}/(l + L ( a ) } ,
where &,(a) - K,,(a)I,,’(a)/K,,’(a)l,,(a)in terms of the modified Bessel
functions I,,,K, of the first and second kinds and their derivatives. This
flow is unstable for each pair of values %,a. As a 00 (i.e. as the radius
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 86

L of the vortex sheet-,oo for fixed a*),L,(u)

for a plane vortex sheet is recovered.
-. 1 and the result c = #(l +i )
For the profile of a realistic round jet,
(6.26) I7 = (1 + Y y (O<Y< w)

DQ does not vanish anywhere when n = 0,so the axisymmetric disturbances

are stable [W].Further, max (YDQ/(U,- V ) }< 1 when n 2 2, so the
O<r< m
only possibility of instability occurs when n = 1. In fact there is instability
when n = ,l[go], the singular neutral mode occurring for a = u, = 1.46
and (arc)*= 0.57; i.e. c = U,= 0.62. Thus there is instability only for the
sinuous mode with o r < 1.46.

Refemaces
1. HELMHOLTZ, H.,tlber diskontinuierliche Flussigkeitsbewegungen, Monats. konigl.
fireuss. Akad. Wiss. Berlin 215-228 (1868); translation by F. Guthrie. On
discontinuous movements of fluids, Phil. Mag. (4) 86, 337-346 (1868) ; also
,.Wissenschaftliche Abhandlungen“. vol. 1, J. A. Barth, Leipzig. 1882.
2. KELVIN,W., The influence of wind on waves in water supposed frictionless. Phil.
Mag. (4) 42, 368-374 (1871); also ”Mathematical and Physical Papers,” vol. IV,
“Hydrodynamics and General Dynamics,” pp. 76-83, Cambridge Univ. Press, 1910.
3. RAYLEIGEL J. W. S., “The Theory of Sound” vol. 11, chap. XXI. Dover, New York,
1945 (reprint of 2nd ed. of 1894).
4. SQUIRE,H. B., On the stability of three-dimensional disturbances of viscous flow
between parallel walls, Proc. Roy. SOC.A, 142, 621-628 (1933).
5. LIN. C. C., “Hydrodynamic stability,” Proc. Symposia in Appl. Math., vol. V,
McGraw-Hill, New York, 1-18, 1954.
6. HEISENBERG,W., uber Stabilltiit und Turbulenz von Fliissigkeitsstr6men. Ann.
Phys., Leipzig (4) 74, 577-627 (1924); translated as Nat. Adv. Comm. Aero.,
Washington. Tech. Memo., no. 1291 (1951).
7. LIN, C. C., On the stability of two-dimensional parallel flows. Part 11: Stability
in an inviscid fluid, Quart. AHZ. Math. 5, 218-234 (1945).
8. DRAZIN,P. G. and HOWARD, L. N., The instability to long waves of unbounded
parallel inviscid flow, J. Fluid Mech. 14, 257-283 (1962).
9. LIN, C. C., “The Theory of Hydrodynamic Stability,” Cambridge Univ. Press,
London and New York (1955).
10. STUART, J . T., Hydrodynamic stability, i n “Laminar Boundary Layers,” chap. IX,
(L. Rosenhead, ed.), Oxford Univ. Press, London and.New York, 1983.
11. REID, W. H.. The stability of parallel flows, i n “Basic Developments in Fluid
Dynamics,” (M. Holt, ed.) Academic Press, New York, pp. 249-307, 1965.
12. RAYLEIGH, J . W. S., O n the stability, or instability, of certain fluid motions, PYOC.
Lond. Math. SOC.9, 57-70 (1880); also “Scientific Papers,’’ vol. I. pp. 474-487
Cambridge Univ. Press, London and New York. 1899.
13. FJBRTOFT. R., Application of integral theorems in deriving criteria of stability of
laminar flow and for the baroclinic circular vortex, Geofys. Publ. 17, no. 6
pp. 1-52 (1950).
14. TOLLMIBN, W.. Ein allgemeines Kriterium der Instabltiit Laminarer Geschwindig-
keitsverteilungen, Nachr. Ges. Wdss. Gottingen, Math.- Phys. Klasse, 60, 79-1 14
(1935); translated as Nat. Adv. Comm. Aero., Washington. Techn. Memo.. No. 792
(1936).
86 P. G. DRAZIN AND L. N. HOWARD

16. FRIEDRICHS, K. O., “Fluid Dynamics,” chap. IV. mimeographed lecture notes,
Brown University, Providence, Rhode Island, 1942.
16. HOWARD, L. N., The number of unstable modes in hydrodynamic stability problems.
J . de Mdcanique 8, 433-443 (1964).
17. ROSENBLUTH, M. N. and SIMON,A., Necessary and sufficient condition for the
stability of plane parallel inviscid flow. Phys. Fluids, 7, 667-668 (1964).
18. FOOTE,J . R. and LIN, C. C., Some recent investigations in the theory of hydro-
dynamic stability, Quart. Appl. Math. 8, 266-280 (1950).
19. TAYLOR,G. I., Eddy motion in the atmosphere, Phil. Trans. Roy. Soc. A, B1Q.
1-26 (1816).
20. LIGHTHILL, M. J., Physical interpretation of the mathematical theory of wave
generation by wind, J . Fluid Mech. 14, 386-398 (1962).
21. KELVIN,W., On a disturbing infinity in Lord Rayleigh’s solution for waves in a
plane vortex stratum, Nature 88, 45-46 (1880): also “Mathematical and Physical
Papers,” vol. IV., Cambridge Univ. Press, pp. 186-187, 1010.
22. GILL, A., A mechanism for instability of plane Couette flow and of Poiseuille
flow in a pipe, J . Ffuid Mech. 81, 603-611 (1966).
23. HOWARD, L. N.. Note on a paper of John W . Miles, J. Fluid Mech. 10, 609-612
(1961).
34. HBILAND,E.,On two-dimensional perturbation of linear flow, Geofys. Publ. 18,
NO. 9 pp. 1-12, (1863).
26. HOWARD.L. N., Neutral curves and stability boundaries in stratified flow, J.
Fluid Mech. 16, 333-342 (1963).
26. ORR,W. M’F.. The stability or instability of the steady motions of a perfect liquid
and of a viscous liquid. Part I : A perfect liquid, Proc. Roy. Irish Acad. %7 A,
9-68 (1907).
37. MILES, J. W., On the disturbed motion of a vortex sheet, J. Fluid Mech. 4, 638-662
(1968).
28. CARRIER, G. F. and CHANG.C. T., On an initial vaiue problem concerning Taylor
instability of incompressible fluids, Quart. ApPI. Madh. 18, 436-439 (1969).
20. CASE,K. M.,Stability of inviscid plane Couette flow, Phys. Ffuids 8, 143-148 (1960).
30. CASE, K.M.,Stability of an idealized atmosphere, I: Discussion of results, Phys.
Ffuids 8, 149-164 (1960).
31. DIKII,L. A., On the stability of plane parallel flows of an inhomogeneous fluid,
J. Appl. Math. Mech. 84. 367-369 (1960). This is English translation of Russian
paper in Prik. Math. Mekh. B4, 249-260 (1960).
32. DIKII.L.A.. The stability of plane-parallel flows of a n ideal fluid, Sou. Phys. Dohlady
6, 1179-1182 (1961). Translated from Russian in Dokkrdy Akad. Nauk. S S S R
186, 1068-1071 (1960).
33. CASE, K. M., Hydrodynamic stability and the inviscid limit, J . Fluid Mech. 10,
420-429 (1961).
34. LIN, C. C., Some mathematical problems in the theory of the stability of parallel
flows, 1. Fluid Mech. 10, 430-438 (1961).
36. ELIASSEN,A.. HeILAND, E.. and RIIS, E., Two-dimensional perturbation of a flow
with constant shear of a stratified fluid, Inst. Weather Climate Res., Oslo. hb-
lication No. 1 (1063).
36. SAVIC,P.,O n acoustically effective vortex motion in gaseous jets, Phil. Mag. ( 7 )
8g, 246-262 (1941).
37. SAVIC.P. and MURPHY, J. W., The symmetrical vortex street in sound sensitive
plane jets, Phil. Mag. (7). 84, 139-144 (1943).
38. L~SSEN, M. and FOX,J. A.. The stability of boundary layer type flows with infinite
boundary conditions, ,,60 Jahre Grenzschichtforschung,” (H. Wrtler and
W. Tollmien, eds.), Vieweg und Sohn, Braunschweig, pp. 122-126, 1965.
HYDRODYNAMIC STABILITY OF PARALLEL FLOW O F INVISCID FLUID 87

39. HOLLINGDALB, S., Stability and configuration of the wakes produced by solid
bodies moving through fluids, Phil. Mag. (7)99, 209- 57, (1940).
40. HAURWITZ, B. and PANOFSKY, H. A., Stability and meandering of the Gulf Stream,
Trans. Amer. Geophys. Union 81, 723-731 (1960).
41. SATO,H.. The stability and transition of a two-dimensional jet, J. Fluid Mech. 0 ,
53-80 (1960).
42. CURLE, N., On hydrodynamic stability of unlimited antisymmetrical velocity
profiles. Aero. Res. Council, London, unpublished Rept. No. 18664 (1956).
43. GARCIA,R. V., Barotropic waves in straight parallel flow with curved velocity
profiles, Tellus 8, 82-93 (1966).
44. CURLE,N.. Hydrodynamic stability of the laminar mixing region between parallel
streams, Aero. Res. Council, London, unpublished Rept. No. 18426 (1956).
45. MICHALKE.A., On the inviscid instability of the hyperbolic-tangent velocity profile.
J. Fluid Mech.. 19, 643-656 (1964).
46. ESCH,R. E., The instability of a shear layer between two parallel streams, J . Fluid
Mech. 8, 289-303 (1957).
47. CHIARULLI,P.and FREEMAN, J. C.. Stability of the boundary layer. Headquarters
Air Materiel Command, Dayton, unpublished Tech. Hept. No. F-TR/l197-1A
(1948).
48. BJERNBS. V., BJERKNES. J., BBRGERON, T.. and SOLBERG.H., .,Physikalische
Hydrodynamik,” Springer, Berlin, 1933. Translated into French as “Hydro-
dynamique Physique.” Presses Universitaires de France, Paris. 1934.
49. GODSKE,C. L., BLRGERON, T.. B~ERKNES, J. and BUNDGAARD, R. C., “Dynamic
Meteorology and Weather Forecasting,” Amer. Meteor. SOC., Boston, 1967.
60, HAURWITZ. B., Zur Theorie der Wellen in L u f t und Wasser, Veriiff. Geophys. ins;.
Llniv. Leipsig, 6, no. 1 (1931).
51. RAYLEIGH, J. W. S., Investigations on the character of the equilibrium of an in-
compressible fluid of variable density. Proc. Lond. M d h . SOC.14, 17&177 (1883).
Also in “Scientific Papers,” vol. 11, pp. 200-207, Cambridge Univ, Press,London
and New York, 1900.
62. TAYLOR, G.I., Effect of variation of density on the stability of superposed streams
of fluid, Proc. Roy. Soc. A, 183. 499-523 (1931).
53. GOLDSTEIN, S.. On the stability of superposed streams of fluid of different den-
sities, Proc. Roy. SOG.A. 182. 524-548 (1931).
64. YIH, C. S.. Stability of two-dimensional parallel flows for three-dimensional dis-
turbances, Quart. A p p l . Math., 12, 434-435 (1965).
65. TAYLOR,G. I., The instability of liquid surfaces when accelerated in a direction
perpendicular to their planes. I, Proc. Roy. SOC.A, W l , 192-194 (1950).
68. ALTERMAN,2.. Kelvin-Helmholtz instability in media of variable density, Phys.
Fluids, 4, 1177-1179 (1961).
57. MENKES,J., On the stability of a shear layer, J. Fluid Mech., 6, 518-622 (1969).
68. FEJER,J. A.. and MILES. J. W.. On the stability of a plane vortex sheet with respect
to three-dimensional disturbances, J. Fluid Mech., 16, 335-336 (1963).
59. LANDAU, L., Stability of tangential discontinuities in compressible fluid, Akad.
N a r k . S S S R . Comptes Rendus (Doklady), 44, 139-141 (1944).
60. HATANAKA, H., O n the stability of a surface of discontinuity in a compressible
fluid, J . SOC.Sci. Culture, Japan, e, 3-7 (1947).
81. LIEPMANN. H. W.and ROSHKO, A., ”Elements of Gas Dynamics,” Wiley, New York.
1957.
62. K ~ C H B M A ND., N , Sarungsbewegungen in einer Gasstr6mung mit Grenzschicht.
ZCit. angew. Math. Mech. 18. 207-222 (1938).
83. LIN, C. C., On the stability of the laminar mixing region between two parallel
streams in a gas, Naf.Adv. Comm. Aero., Washington, Tech. Note No. 2087 (1968).
88 P. G. DRAZIN AND L. N. HOWARD

64. ECKART. C., Extension of Howard’s circle theorem for adiabatic jets, Phys. Fluids
6, 1042-1047 (1963).
66. JOHNSON, J . A., The stability of a shearing motion in a rotating fluid, J. Fluid
Mech. 17, 337-362 (1963).
66. Kuo, H. L., Dynamic instability of two-dimensional non-divergent flow in a baro-
tropic atmosphere, J . Meteor., 6 , 105-122, (1940).
67. HOWARD. L. N. and DRAZIN,P. G., On instability of parallel flow of inviscid fluid
in a rotating system with variable Coriolis parameter, J . Math. and Phys. 48,
83-99. (1064).
68. LIPPS, F. B., The barotropic stability of the mean winds in the atmosphere, J.
Fluid Mech. 12. 397-407 (1962).
69. MICHAEL, D. H.. Stability of plane parallel flows of electrically conducting fluids,
Proc. Camb. Phil. Soc., 49, 166-168 (1953).
70. STUART, J . T., On the stability of viscous flow between parallel planes in the presence
of a coplanar magnetic field, Proc. Roy. SOC.A, 391. 189-206 (1964).
71. MICHAEL, D. H., The stability of a combined current and vortex sheet in a perfectly
conducting fluid, Proc. Camb. Phil. SOC.,61, 628432 (1956).
72. DRAZIN, P. G., Stability of parallel flow in a parallel magnetic field at small magnetic
Reynolds numbers, J . Fluid Mech., R, 130-142 (1960).
73. LOCK,R. C., The stability of flow of an electrically conducting fluid between parallel
planes under a transverse magnetic field, Proc. Roy. SOC.A, 388. 10.5-125 (1965).
74. LIEPMANN, H. W. and PUCKETT, A. E., “Introduction to Aerodynamics of a Com-
pressible Fluid,” Wiley, New York, 1947.
76. HOLYBOE,J., On the behavior of symmetric waves in stratified shear layers.
Geofys. Publ., 94, no. 2, 67-113 (1962).
76. SYNGE,J . L., The stability of heterogeneous liquids, Trans Roy. SOC. Canada, 27,
111, 1-18 (1933).
77. MILES, J. W., On the stability of heterogeneous shear flows, J. Fluid Mech., 10,
496-508 (1961).
78. MILES, J. W., On the stability of heterogeneous shear flows. Part 2. J . Fiuid Mech.,
16, 209-227 (1963).
79. DRAZIN, P. G. and HOWARD, L. N., Stability in a continuously stratified fluid,
Proc. Amer. SOL. Civil Engvs. E . M . Div. R7, 101-116 (1961). Also Trans A m r .
SOC. Civil h g r s . 19A. 849-864 (1963).
80. FJELDSTAD, J . E., Interne wellen, Geofys. Publ., 10, no. 6, pp. 1-35 (1933).
81. DAVIS,P. A . and PATTERSON, A. M., The creation and propagation of internal
waves. A literature survey. Unpublished tech. memo No. 56-2, Pacific Naval Lab..
Esquirnault, British Columbia (1956).
82. HeILAND, E., On the stability of Couette-flow of a fluid with parabolic gravitational
stability, Vitensk. Akad. Inst. Vaer-og Kiimasforskning. Publ. no. 1, (1954).
83. HOWARD, L. N. and DRAZIN, P. G., (to be published).
84. MILES, J. W. and HOWARD, L. N., Note on a heterogeneous shear flow, J. Fluid
Mech., 90, 331-336 (1964).
86. DRAZIN, P. G., The stability of a shear Layer in an unbounded heterogeneous in-
viscid fluid, J . Fluid Mech. 4, 214-224 (1958).
86. ESCH,H. E., Stability of the parallel flow of B fluid over a slightly heavier fluid,
J. Fluid Mech., 12. 192-208 (1962).
87. HOCKING, L. M . . The instability of a non-uniform vortex sheet, .I, Fluid Mech.,
l A , 177-186 (1964).
88. HOCKING,L. M., Note on the instabilitv of a non-uniform vortex sheet. J . Fluid
Mech., 31, 333-336 (1966).
89. SYNGE,J . L., Hydrodynamical stability, Semi-centenn. Publ. Amer. Math. Sac..
S, 227-269 (1038).
HYDRODYNAMIC STABILITY OF PARALLEL FLOW OF INVISCID FLUID 89

90. BATCHELOR. G. K. and GILL,A. E., Analysis of the stability of &symmetric jets,
J . Fluid Mech. 14. 629-551 (1962).
91. F~AYLBIGH,J. W. S., On the question of the stability of the flow of fluids, Phil.
Mag., 84. 59-70 ( 1 8 9 2 ) ; also “Scientific Papers”, vol. 111, Cambridge Univ.
Press, London and New York, pp. 676-584, 1902.
92. PRETSCH, J . , Uber die Stabilitat einer LaminarstrBmung in einem geraden Rohr
mit kreisf6rmigem Querschnitt, Z . angew. Math. Mech., el, 204-217 (1941).
The Calculation of Free Oscillations of a Liquid in a Motionless
Container
BY N. N. MOISEEV AND A. A. PETROV

Computing Center 01 the U S S H Academy of Sciences, Moscow, U S S R

Page
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
11. The Variational Method Applied t o Free Oscillations of a Liquid . . . . . . 93
1. Basic Results of General Theory; the Ritz method . . . . . . . . . . . 93
2. On the Choice of a System of Coordinate Functions. . . . . . . . . . . 96
3. An Approximate Method of Calculation of Free Oscillations . . . . . . . 102
111. Application of the Ritz method to the Calculation of Natural Oscillations of a
Liquid in Containers of Various Shapes . . . . . . . . . . . . . . . . . 106
1. Free Oscillations in a n Inclined Circular Cylinder . . . . . . . . . . . 106
2. Free Oscillations in a Conical Container . . . . . . . . . . . . . . . . 111
3. Free Oscillations in a Spherical Container . . . . . . . . . . . . . . . 118
4. Free Oscillations in a Container formed by Two Horizontal Coaxial
Cylinders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5. Free Oscillations in a Cylindrical Container with Horizontal Axis . . . . . 132
6. Free Oscillations in Circular Cylindrical Containers Whose Bottom and Cover
are Spherical Caps . . . . . . . . . . . . . . . . . . . . . . . . . 138
7. Free Oscillations in a Container of Toroidal Shape . . . . . . . . . . . 143
Appendix. The Escalator Method for Solving the Equation I 3 = AQx . . . 148
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

I. INTRODUCTION

In the preceding volume of the “Advances in Applied Mechanics” the

basic ideas of the theory of oscillations of a liquid and the liquid-contain-
ing body [I] have been discussed. In creating this theory the variational
principles have been used in a systematic way. The choice of this method
has not been made fortuitously. The use of variational principles not only
makes it easier to investigate the general properties of the motion (this has
already been discussed in [l]) but it turns out to be very convenient for the
development of numerical methods of actual solution.
The present survey gives numerical methods and results of the calcula-
tion of free oscillations of a liquid in a motionless container. This is the
simplest problem which appears in the investigation of the general dynamics
of a body containing liquid masses with a free surface. Despite the sim-
plicity of the mathematical formulation it is not a t all easy to obtain accurate
numerical results in this problem of free oscillations of a heavy liquid.
91
92 N. N. MOISEEV AND A. A. PETROV

Only in the simplest cases one can manage to solve it analytically by the
Fourier method. These are the generally known oscillation problems of
a liquid in a container of cylindrical shape with a flat bottom (parallelepiped,
circular cylinder, cylinder with a circular ring or its sector as the base, and
elliptical cylinder). Apart from this, there are some papers which, in one
way or another manage to give solutions of the problem of liquid os-
cillations in a container of a certain special shape. These papers include
for example [2] and [3].
However, a number of engineering tasks require the skillful construction
of the spectrum of natural frequencies for containers of an almost com-
pletely arbitrary shape. Therefore a method of solution of these prob-
lems is required which can claim a good degree of universality in its
application. One of the possible methods satisfying this requirement is,
in the opinion of authors, the Ritz method. I t is based on the application
of Hamilton’s principle to the motion of an ideal incompressible liquid.
In paper [l] the applicability of the Ritz method has been proved for a
large class of problems.
The present paper can be considered as a continuation of this work.
Here, on the basis of the general results given in [l], we systematically
describe the application of the Ritz method to calculations of free oscilla-
tions of a liquid in containers of various shapes.
The present survey consists of two sections. In Section 1.1 we have
considered it necessary to give the basic results of the general theory. This
simplifies further reading of the material. In Sections 1.2 and 1.3 some
general thoughts of efficient realization of the Ritz scheme are given.
In the second section the solutions of practical problems of free oscil-
lations of liquid in containers of various shapes and the results of the cal-
culations are presented.
Finally, in the Appendix one of the methods of numerical solution of
an algebraic problem is discussed which arises as a result of the application
of the Ritz method.
A significant part of the results has been obtained by the authors of the
survey.
In addition, results are given which have been obtained by other authors
who solved problems by the variational method. However, here we have
used only the results of their calculations. As for the method, we have
adopted a uniform procedure for solving all the problems.
This survey considers only small oscillations of the liquid under the
action of mass forces. Many questions which by now have become actual
and to which much work has been devoted, are beyond the scope of this
paper. These are first and foremost the problems of liquid dynamics in
force fields of low intensity where surface tension begins to play a decisive
pa1 t, the dynamics of the liquid-containing body under these conditions,
and, finally, the dynamics of a body containing a viscous liquid.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 93

11. THE VARIATIONAL

METHODAPPLIEDTO FREE OF A LIQUID
OSCILLATIONS

1. Basic Results of the General Theory; the Ritz Method

1. An inviscid incompressible liquid fills part of a motionless container.
The liquid is subjected to gravity whose acceleration is denoted by g. Further
denote by T the region (and its volume) occupied by the liquid in equilib-
rium position, let the free surface of the liquid (and its area) in the equilib-
rium position be S, and the wetted surface of the container walls (and its

P i

/-

x
FIG.1.

area) be C. Introduce the coordinate system Oxyz with the xy-plane parallel
to the free surface of the liquid; the z-axis is opposite in direction to
the vector g in Fig. 1.
As has been shown in [l], in the problems considered the liquid motion
can be considered as irrotational without loss of generality. Denote the
velocity potential by y(x,y,z,t) and the elevation of the free surface above
the plane S by c(x,y,t). If one assumes that the motion of the liquid is small,
then the function q~ is a solution of the following boundary problem [l]:
(1.1) Av=O in the region t,

_
av -- 0 on the surface C,
(1.2) an

on the surface S;

here n is the outer normal to the surface bounding the region 1.

94 N. N. MOISEEV AND A. A. PETROV

2. One calls free (or natural) oscillations the solution of the problem
(1.1)-(1.4) of the form
gJ(x,y,z,t)= @(x,y,z)cos d.

S(x,y,t)= Z(x,y)sin d,
where (I is the natural frequency. From the conditions (1.1)-(1.4) it follows
that the function 0 is a solution of the following eigenvalue problem:
(1.6) A@=O in the region 7,

-a@an
=o on the surface Z,

a@
~ = i Z Q , on the surface S.
(1.7) an

The eigenvalue A of the problem is connected with the natural frequency

of the oscillations (I by the relationship

This problem of free oscillations permits an equivalent variational

formulation [l] : to find in the class of continuous functions, having contin-
uous first derivatives, the minimum of the functional

(1.9) ! I
F(@) = (I'@)'dz - 1. W d S .
s

3. One may introduce the Neumann operator which establishes a cor-

respondence between the values of the normal derivative of the function g~
on the boundary of the region T and the values of the harmonic function cp
in the region T :

The functional (1.9) can now be rewritten in the form

(1.10) F ( 2 ) = A(H2,Z) - (2.2).
where (u,v) denotes the scalar product:
 FREE LIQUID OSCILLATIONS. MOTIONLESS CONTAINER 95

The operator H is defined on the set of the functions 2 orthogonal to

1 and is bounded, completely continuous, selfadjoined, and positive, [l].
The following theorems follow from this:

(1) In the motion of a liquid in the neighborhood of the position of

equilibrium natural oscillations exist, if the volume is finite.

(2) The natural frequencies u are positive, have finite multiplicity,

form a discrete spectrum, and
ha,= 00.
*+(o

(3) The eigenfunctions 2, of the operator H which describe the principal

.. is complete
forms of liquid oscillations are such that the system l,Zl,Zz,.
in the sense of convergence in the mean.

(4) The solution of the variational problem (1.10)can be obtained by

the Ritz method.

4. The formal procedure of the Ritz method consists in the following.

A complete set of the functions {x.} belonging to the region of definition of
the operator H is constructed in the region t. The solution of the variational
problem (1.9)is sought in the form of the sum
N
(1.11)

Substitution of this sum in the functional (1.9)converts it into the

homogeneous quadratic function N of the variables al,az,.. . ,aN

where

(1.12)

The conditions for an extremum of this function, aF/aa, = 0, give us

the set of linear homogeneous algebraic equations with respect to the un-
known a,,,
N N

(1.13) 2 p,am
m-1
-1 2 q,,,,,a,,,= 0,
m-1
12 .
= 1,2,. ..N.
96 N. N. MOISEEV AND A. A. PETROV

The system (1.13) has a non-trivial solution if

(1.14) det +,I - iIq,,,,,l = 0.

This condition approximately determines N eigenvalues of the problem

(15)-(1.7), and because of (1.8) also N natural frequencies of free oscillations.
The corresponding solutions of the system (1.13) approximately deter-
mine N eigenfunctions of this problem which describe N basic forms of
liquid oscillations.
Here
lim L")= A, = an,
lim @p)
N+m N+4)

where the eigenvalues and eigenfunctions of the problem (1.6)-(1.7) are

denoted by A,, and @,, respectively.

2 . On the Choice of a System of Coordinate Fulzctions

It was mentioned above that for the practical realization of the Ritz
scheme in the region t it is required that a complete set of the functions
{x,,} should be constructed. The rate of convergence of the approximate
solutions to the problem solution depends on the successful construction of
the set of functions {x"}. Consequently, the design of the basis is the main
problem in the use of the Ritz method.
Apparently it is impossible to point out any general method for selecting
the functions x,, which will provide satisfactory convergence in all cases.
However, some general ideas can be set forth which are useful to keep in
mind in solving practical problems.

(1) If the functions x,, are harmonic and satisfy some of the boundary
conditions of the problem, one obviously will expect that the Ritz procedure
will rapidly converge. And one of the widely applicable ways of choosing the
set x,, consists in the construction of a set of linearly independent solutions
of Laplace's equation (1.6) in the region T . which satisfy the conditions
(1.6) as accurately as possible.
In Sections 11.1-11.4 solutions of a number of problems are given
which are based on the use of this idea.
In particular, if the problem (1.6)-(1.7) permits partial separation of
variables, i.e., if the solution of the problem is, let us say, of the form

(1.16) @(x,rd= 4WJ)Wl

then the minimum of the functional (1.9) should be sought among the func-
tions of the form (1.16).
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 97

To fix the ideas, assume that a container (the region t) has the shape
of a vertical cylinder whose height is h. Then, by substituting the expression
(1.16) into the functional (1.9) and by integrating with respect to z we
obtain the functional

(1.16)
s s
where

(1.17)

In the case under discussion

U(%) =
+
cosh o ( z h)
cosh o h

By substituting this expression in the equation (1.17) we find the connec-

tion between the values of il and m :

A=otanhoh.

Thus, for the case of a right cylinder with a flat bottom, we have a
variational problem for the functional (1.16), which is much simpler than
the initial one.

(2) The eigenvalues of the problem (1.6)-(1.7) are extremal values of

certain functionals. This brings to mind that the eigenvalues are not very
sensitive to the choice of the functions x,,.
The boundary conditions (1.6), (1.7) relate to the kind of natural bound-
ary conditions (i.e., they follow from Hamilton's principle), Therefore,
for the convergence of the Ritz method the functions XI,need not satisfy
these boundary conditions.
With this is mind one more possibility of choosing the coordinate func-
tions becomes apparent (41. As mentioned in the introduction, the Fourier
method makes it possible to solve completely the problem of free oscillations
of liquid inside cylinders of a special kind. Denote by toone of the regions
for which we have an effective solution of the problem (1.6)-(1.7), and let
the bounding surfaces be So and Zo. Denote the eigenfunctions by #,, and
98 N. N. MOISEEV AND A. A. PETROV

the eigenvalues by p,,,. Put the container, for which the solution of the
problem of free oscillations is required, wholly inside the region toof minimal
dimensions in such a way that the planes of areas S and So coincide (Fig. 2).

SO

I
r I

I
'\= //
\

20

FIG.2.

Clearly, from the regions tomentioned above there should be chosen one
whose shape is closest to the shape of the container, but here, certainly,
much arbitrariness remains. The set of the functions ($I~} can be taken
as coordinate functions by setting xrn= +,,,.
Denote the difference of volumes r0 and t by A t and the difference of
areas So and S by A S . The formulae (1.12)for the coefficientsof the system
(1.13) can now be rewritten in the form

$urn = j
Trn
v+n v+rndt - 1
AT
v+n v+rndt,

qnn, = [ +*+mas - [ +n+rnds*

J J
S AS

From Green's formula and the boundary values of the functions$-,I on

the surfaces So and Zoit follows that

j v+nv+rndt = pnj + n + ~ s -
7a sa

This implies, because of the equivalence of the indices rc and m, that the
functions +,, are mutually orthogonal over the surface So. If we consider
that

5
sb
= 1,
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 99

it is not difficult to show that

Furthermore, from the same Green's formula and from the boundary
values of the functions $, it follows that

Hence, the coefficients of the system (1.13) can be represented in the form
convenient for caIculations

(1.18)

(1.19)

where 6, is Kronecker's symbol.

From the obvious relations p,, = &,, and q,,, = q, it follows that

(1.20)

If the shape of a container is such that the free surfaces So and S co-
incide, then AS = 0 (cf. Fig. 2) and the equations (l.lS), (1.19) simplify to

qnm = Sum.

Correspondingly, the system (1.13) and equation (1.14) are simplified.

In Sections 11.6-11.7 solutions of several problems by the use of this
method of construction of coordinate functions is given.
We now give expressions for the eigenfunctions #, and the eigenvalues
,urn for several regions to.
100 N. N. MOISEEV AND A. A. PETROV

1. Parallelepiped. Denote by I the length, by 2d the width, and by H

the height of the parallelepiped.
Introduce the coordinate system OXYZ as shown in Fig. 3 and turn to
dimensionless variables x,y , z by the formulae

x = xl, Y =y d , z = ad.

FIG. 3.

The system of the functions $2)

is divided into two subsystems

where

and

where
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 101

2. Right circdar cylider. Denote the radius of the cylinder by R and the
height by H. Introduce the coordinate system as shown in Fig. 4, using
dimensionless variables

and introduce the cylindrical coordinates r,O,z by the formulae

x=rcosO. y=rsinO, Z=Z, rB=x8+ya.

I I li
FIG.4.

The eigenvalues of the problem have double multiplicity : to each eigenvalue

correspond two eigenfunctions

where Jn(E) are Bessel functions of the first kind, KZ)is the mth root of
the equation J,'(K) = 0, and

The case n =0 is the exception; here the eigenvalues have multiplicity one.
3. Right cylinder with an anmdar base. Denote the radius of the inner
cylinder by R, and the radius of the outer cylinder by cR where c > 1 is a
102 N. N. MOISEEV AND A. A. PETROV

dimensionless coefficient. The coordinate system is shown in Fig. 6. Intro-

duce dimensionless variables as in the previous case.

FIG.5 .

As in the case of the circular cylinder, to each eigenvalue

pJ") = KJ*) tanh K,,,"%

correspond two eigenfunctions

(1.24)

where
P,(r) = N,,'(K,,,("))J,,(K,,$"Y)
- J.'(K,,,(,,))N,,(K,,,(")Y),

J,,(E) and N,,(E) are Ressel functions of the first and second kind, K,,,('" is the
mth root of the equation
J:(K) .N,,'(CK)- N / ( K ) ' J:(CK) = 0,
and

m,n = 0,1,2,. . ..
The exception is again the case n = 0 with eigenvalues of multiplicity one.

3. A n Afq5roximate Method of Calcdation of Free Oscillations*

Assume now that the region z, for which the problem of free oscillation
of a liquid is to be solved, is close to the region to(we assume here that the
planes of the free surfaces S and So coincide) for which the effective solution
* The main content of this section is discussed in [a].
 F R E E LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 103

of a similar problem can be constructed. This means that for towe know
the system of the eigenvalues pmand the eigenfunctions #m. The closeness
of the regions t and to(S and So respectively) permits an efficient use of
perturbation theory.
In solving the problem for the region t by the Ritz method we shall
take the functions +m as a system of coordinate functions. It is assumed in
this case that the functions (Im can be continued outside the region toin
such a way that the system of these functions remains complete also in the
region t (unless z, completely encloses t).
Let us.now find the solution in the form
N
= 2
n=1
an+nq

This will lead us to the system (1.13) whose coefficients in this case will be
of the form

1 1
70 AT

qnm = +n#mdS + +n+mds.

s AS

But it has been shown in the previous section that for functions normalized
so that

the relations

hold.
Furthermore, since the

(1.25)

(1.26)

where E << 1 and the numbers l,,,,,and q,,,,, are of the order of unity.
104 N. N. MOISEEV AND A. A. PETROV

According to this the coefficients of the system (1.13) can be represented

in the form
(1.27) Pnm = PAtn + Etnm 1

(1.28) qmtn = 8nm +q m

and the system (1.13) itself can be transformed into the form
N
(1.29) (11s - A)an =-E 2 (turn - A q n m ) a m ,
tn=l
=
~t 1s2,. . - JV.
We shall find the solution of this problem in the form of the series
m

(1.30)

m
(1.31)

By substituting the series into system (1.29) and by equating the coefficients
of the same powers of 8 we obtain the following systems for the determina-
tion of the coefficients of the decomposition (1.30) and (1.31):
(1.32) (Pn - A('))U,(~) = 0, n = 1,2,. . ., N ,

The solution of the system (1.32) describes the oscillations of a liquid

in the region toand has the form

(1.34) = pe, = Bpn, t 1,2,. . . , N .

p,~=
Here the first index of the coefficients ag indicates the number of the
eigenfunction,
We shall seek that solution of the system (1.29) which a t E -,0 becomes
the solution (1.34) for p = s. By taking into account (1.34) the system
(1.33) can be rewritten as

(1.36) = ES, - psqssa

The coefficient a:) has remained indeterminate. We shall find it later from the
conditions of normalization.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 106

If we restrict ourselves to the leading terms in the decompositions (1.30),

(1.31), the final result, with the help of formulae (1.26) and (1.26), can be

1
obtained in the form

v$sv$ndz - ys j+
$s$ssds

+c
N
@, = (1 + d ) ) * S
n=1
AT
Ps
AS

- Pn
- *us

U#S

I , = ps + [ (P$s)ndt- pSJ[ 1cIsZdt.

J
A7 AS

These formulae can be simplified by using Green’s formula and the boundary
values of the function $,. Indeed, if

it turns out that

5
n#s

I , = ys - *s $dS.
z

Let us finally, determine the coefficient a$:’ from the normalization condition

I
s
asSdS = 1.

By substituting here the expression for eSand by omitting terms of second

order, we obtain the equation

2ealJ’ +AS I (G~V.S

= 0,

from which one can determine the unknown coefficient.

Thus the solution of the problem in first approximation has the form

(1.37)
106 N. N. MOISEEV AND A. A. PETROV

The formulae (1.37), (1.38) show that for an approximate calculation

of the natural frequencies and principal modes of free oscillations in containers
whose shape is close to some prototype, for which the solution of a similar
problem is known, it is sufficient to calculate a certain set of surface inte-
grals. The application of these formulae eliminates the solution of the very
complicated algebraic problem (1.13).
In Sections 11.6 and 11.7 the results of calculations of free oscillations
in some containers by the formulae (1.37) and (1.38) are given. These
results are then compared with the precise solution of the problem ob-
tained by the use of an electronic computer. The comparison shows that
the formulae (1.37), (1.38) give good estimates (which practically require
no improvement) for the natural frequencies and principal modes of the
free oscillations of a liquid.

111.APPLICATION
OF THE RITZMETHOD TO THE CALCULATIONOF NATURAL
OF A LIQUIDIN CONTAINERSOF VARIOUSSHAPES
OSCILLATIONS

1. Free Oscillations in an Inclined Circdar Cylinder [ 6]

Consider the problem of oscillations of a liquid in a circular cylinder whose

axis forms the angle p with the direction of g. Denote the cylinder radius

FIG.6,

by R and the average depth of the liquid by H. Connect the coordinate

system OXYZ with the free surface of the liquid (Fig. 6).
Introduce the dimensionless variables x,y,z using

Rx = X cosp - Zsin p, Ry = Y , Rz = X sin + Z cosp.

 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 107

In the new coordinate system the axis of the cylinder will appear vertical
and the free surface of the liquid will be inclined at the angle ,8 with respect
to the plane xy (Fig. 7). Introduce the cylindrical coordinates

r =Vx$+ y2, e = arctan-,YX z = z.

FIG.7.

In these coordinates the functional (1.9) can be represented in the form

b 0
I/ 5
2n

-h
1 r(r,e)

(2.1) F(G) = de rdr (vG)adz- il

0 0

Here z(r,B) = x tan p = r cos 0 tan p is the equation of the free surfaces
h = H / R is the dimensionless average liquid depth and the number,
1 = aaR/g are the dimensionless eigenvalues of the problem.
Let us take as a set of coordinate functions those functions which des-
cribe the oscillations of the liquid in the vertical cylinder of depth h [see
(1.23) in Section 1.21:

1
+.1((0 = N,c", J,,(K,(")Y) sin 120
+
cosh K,(")(Z 12)
s,n = 1,2,. . .,
cosh ~,(")h '

These are harmonic functions and satisfy all the boundary conditions of
our problem except the condition at the free surface.
The solution of the variational problem (2.1) will be found in the form
N N
#= 2 aju)UJu)sin n6 + b , ( W j u ) cos ne,
s,n s,n
108 N. N. MOISEEV AND A. A. PETROV

where

Substitution of this sum in the functional (2.1) transforms it into a func-

tion of the 2N variables UP)and b,c"). From the extremum conditions it
follows that the coefficientsa?) and b,(")must satisfy the system of equations
(1.13) which in our case is of the form

The coefficients of this systems are of the form

p$"' 55 I
2n

= d8 rdr
0 0
1 rcosBtanD

-h
(VUJ") * VULm)sin n8 sin m8

nm

J J
0 0

'%n !
I v1 + tanaPU,c")(r,z(r,B))V,rm)(r,z(r,B))
cos n8 cos me&,
0 0
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 109

nm
- -U p )U,(m)cos ne sin me
1%

In our case

It is easy to notice that the functions U p ) .Uim) and VU,c")-VU,(")

are even in 8. The functions sin (no) sin (mO)and cos (no)cos (me) are
also even in 8, and the functions sin (no)cos (me)are odd in 8. From
this it follows that

+ nm sin nt3 sin me

-UJn)U,(m)
78

n 1

n 1
110 N. N. MOISEEV AND A. A. PETROV

Correspondingly, the system of equations separates into two independent

systems
N N

N N

It is not difficult to show that for vertical cylinder axis (p = 0 ) ji$"") = fifi'"),
qLM) = &"') and the solutions of the systems coincide. This corresponds
to the fact that the eigenvalues have multiplicity two, and to each of them
belong two eigenfunctions and I,&(") (except for the case n = 0). If the

" t 3 5 7 9
N

cylinder is inclined ( p # 0 ) then the coefficients of the systems do not

coincide and frequency splitting takes place : different natural frequencies
correspond to the forms
N N
@= 2 u P ) U , ( sin
~ ) la0 and &, = b,(n)Us(n)cos 4.
s,n s,*

Here are some results of the numerical calculations. The systems (2.2),
(2.3) have been solved by the escalator method. The number of equations
in the systems N was taken equal to 3,6,7,9. The dependence of the value
on the approximation number is shown in Fig. 8 for two values of the
angle f l : p = 30' (dotted curve) and fl = 80" (solid curve).
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 111

These graphs illustrate the rate of convergence of the method. The

dependence of the first eigenvalues on the angle is shown in Fig. 9 and in
Table 1. In all calculations it was assumed that 12 = 3.

1.0

10 20 30 40 50 60

Po
FIG.9.

TABLE1

8"

0 1.841 1.841 3.054 3.064

5 1.816 1.837 - -
10 1.746 1.826 - -
20 1.509 1.790 2.937 2.960
30 1.184 1.709 2.674 2.734
45 0.729 1.566 2.137 2.401
60 0.390 1.393 1.41 1 1.982

2. Free Oscillations in a Conical Container*

Let the lateral surface of the container be a circular cone with the semi-
angle 8, and the bottom of the container a spherical cap of radius R.
Introduce the Cartesian coordinates O X Y Z with origin at the vertex of the
cone and with the axis Z directed downwards along the cone axis (the direc-
tion of the axis Z therefore coincides with the direction of g). We shall measure
the depth of the liquid in the container through the distance H from the
vertex of the cone to the free surface S (Fig. 10).

* In the papzr by L. V . Dokuchaev [7] the displacement potential Q is introduced.

Here we shall use the velocity potential 9. As it follows from [ 11 the functions q and Q
represent the solution of one and the same boundary (and, consequently, variational)
problem if free oscillations are considered.
112 N. N. MOISEEV AND A. A. PETROV

Take dimensionless variables

and introduce spherical coordinates r,6,q according to

x = r cosq sin 6, y = r sin7 sin 8, z = r cos 8.

FIG.10.

Remember that the problem of free oscillations in the given region reduces
to the following eigenvalue-boundary problem :
(2.4) A@ = 0 in the region r,

(2.6)
a@ = 0
- for r = 1,
a@ - 0
-- for 6 = 8,,
aY a8

(2.6)
a@ = - A@ for z = k.

where il = a2R/g the dimensionless eigenvalue of the problem, and h = H / R

the dimensionless depth of the liquid.
The solution of this boundary problem is equivalent to the variational
problem of minimization of the functional

(2.7) F(@) = I(V@)'dr - [email protected].

7 S

We shall find the solution of this problem in a class of harmonic functions

satisfying conditions (2.6).
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 113

By means of Green’s formula, the functional (2.7) can be transformed

into

From the form of the boundary values (2.5), (2.6) one notices that the problem
admits the separation of variables @ = U(r,tl)V(q). Since the solution
must be periodical in q it is obvious that
Co6
I/= nq.
sin

I
i
FIG. 11.

Accordingly we shall find the solution of the variational problem in the

class of functions of the form
cos
@ = U(r,f3) nq.
(2.9) sin
By substitution of this expression into (2.8) we are led to the variational
problem for the functional

where p is the value of the radius vector of a point of the free surface in polar
coordinates p,q on the free surface.
As can be seen from Fig. 11, the equation of the free surface has the form
Y cos 8 = h. On the free surface

p = h tan 6, dp = -de,
cosa 8
114 N. N. MOISEEV AND A. A. PETROV

and
au -
_
az
--au cos 8 - -
ar
aU sin
- ;;si;e)- .
0 = cos 6 - - -
a8 r
(y
By combining all these results the functional (2.10) can be reduced to the
form

.(
8,
au a ~ s i
-- n ~ ]
h ,o)
F(U)=
n
a8 Is r-h/care a sine de
c d e
(2.11)

sin 8
0

(In this expression the factor n2h has been omitted.)

Thus, the problem of free oscillations of a liquid in the cone has been
reduced to the variational problem (2.11) in the class of the functions U
satisfying conditions (2.4). (2.6), (2.9). Hence it appears that each eigenvalue
of the problem has double multiplicity, one eigenfunction containing the
factor cosrrq and the other sinnq. The only exception is again the case
n = 0.
It follows from conditions (2.4), (2.6), and (2.9) that the function U must
satisfy the equation
aau au + -
+ -2 - aau 1
1 - au na
u=o
c ra aea + F cote- ae - ___
(2.12) -
ca rasinae
and boundary values

(2.13)
au
-= 0 for Y =1 and
au
-= 0 for 8=eW
ar ae
Thus, for the solution of the problem (2.11) by the Ritz method a system
of coordinate functions complete in the region 0 8 0, and satisfying < <
conditions (2.12), (2.13) must be constructed.
To do this, let us add to the conditions (2.12) and (2.13)

(2.14)
au
-= - oU for r = h,
ar
and consider the auxiliary eigenvalue problem (2.12)-(2.14) in the region
To (Fig. 11).
This problem can be easily solved by separation of variables, setting
U = a(r)p(O). By substitution of this expression in the conditions (2.12)-
(2.14) we obtain the following boundary problems for the functions a and B
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 116

(2.16) ye
d2a - 2y-da
~ - pa = 0 h< r < 1,
dY2 a7

da da
(2.16) --
-0 for r=l, - -- - w a for Y =h,
dY dr

(2.18) -dP= o for e =e,.

ae
The solution of (2.16), (2.16) is of the form

(2.19) p = Y(Y + 1)s

+ l)Y2’ + ”f’ + l)r2’+’ +
+ l)Y”+’ =):(
(Y Y (Y Y
+
f
(2.20) a=
N,(Y (Y l)h2V+1+ Y’

V(Y + 1)(1 - h2’+1)

= h [ ( Y + 1)h2”+’+
(2.21)
(‘ Y] ‘

The number Y is not yet determined.

Let us make use of the relationship (2.19) and introduce the variable
E = cos 8. Then from the conditions (2.17), (2.18) we obtain the known
boundary problem

_
dP -
-0 for 8 = 8,.
dE
Besides, for 8 = 0 the function p should be finite.
The solution of this problem is represented by the associated spherical
functions

Indices v, in this expression are chosen in such a way that for each integer n
110 N. N. MOISEEV AND A. A. PETROV

Thus the solutions of the boundary problem (2.12)-(2.14) are of the

form

vs(vs + 1)(1- he"s+l)

w -
a - 12 [ ( v s + 1)h + +%I
2vs 1

The set of functions {Us)

is complete in the region To bounded by the curve
yo and the legs of the angle 0 = 0, (Fig. 11). The region T bounded by the
curve y and by the legs of the same angle is wholly included in the region
To. Consequently, the set of the functions (2.22) is complete in the region
T and can be taken as the set of coordinate functions.*
We shall find the solution of the variational problem (2.11) in the form

(2.23) u= c asUs
N

s= 1

where U , is determined by the formula (2.22). The coefficients of the set

of equations (1.13) in this case have the form
8.

L. V. Dokuchaev [7] has calculated the first natural frequency of the

single-node oscillations (n = 1). The results of his calculations are given
in Fig. 12 where the dependence of 2, on the angle 0, is shown.

0:
FIG.12.

* If the angle Oo is small, so that cos B0 = 1 - O ( d ) where a is the wave amplitude,

then the formulae (2.22) give the solution of the problem of free oscillations of liquid
in a cone.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 117

The point in this figure represent the resuIts of measurement of this

frequency which were taken from [8]. Figure 13 shows the dependence of
the eigenvalue A, on the number of approximation N for 6, = 11'. Here
the calculations by (2.22) correspond to N = 0. As seen, the Ritz method
in this case also gives good results.

N
FIG.13.

Similarly, it is possible to solve the problem of oscillations of liquid in

the circular cone shown in Fig. 14. In this case the coordinate system OXYZ
is also connected with the vertex of a cone but the axis 2 is directed vertically

9X FIG.
O 14.
Y

upwards. A characteristic length R is shown in Fig. 14. Due to this choice

the dimensionless depth of the liquid h = cos 0,. It is possible to show in
quite the same fashion that the problem of free oscillations of liquid in this
case reduces to a variational problem for the functionals (2.11) in the class
of the functions U satisfying the condition (2.12) and the second condition
(2.13).
For the construction of a system of coordinate functions one should
add the condition
au -
_
k
- OU for r = 1.
The solution of the boundary problem so formulated is of the form
(2.26) u, = rVs~:,(cos.el, = rs.
118 N. N. MOISEEV AND A. A. PETROV

The solution of the variational problem (2.11) in this case is again sought
for in the form given in (2.23) where the functions U s are calculated by
(2.26). The minimum conditions lead to the system (1.13) whose coefficients
are calculated by formulae similar to (2.24).
The results of calculations of the first natural frequency 1, of the single-
node oscillations (n = 1) are represented in Fig. 16, where the dependence
1.8

0.6

0.2

0:
FIG. 15.

of 4 on the angle 8 is given. Also, the experimental results taken from [8]
are represented by dots.
If one recalls the solution of the oscillation problem in the inclined cylinder
given in the previous section, one will agree that the oscillation problem
in inclined cones must be solved in a similar way. The set of functions
determined by (2.22) and (2.26) must be taken as coordinate functions.

3. Free Oscillations in a Spherical Container+

In the problem of oscillations of liquid in a sphere of radius R, connect
with the sphere center the coordinate system O X Y Z , the axis Z oriented
opposite to the vector g. We shall measure the depth of the liquid H as

FIG. 18.

shown in Fig. 16 and introduce dimensionless variables x = X / R , y =

Y / R , z = Z / R and spherical coordinates r,O,q by
In both [9] and [7] the displacement potential is introduced. Here we shall make
use of the velocity potential (see footnote on p. 111).
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 119

x = rsin 8 cosg, y = rsin 8 sinq, z = r cos 8.

The problem of liquid oscillations reduces to the eigenvalue-boundary

problem
(2.26) A@=O in region t,

aQ,
(2.27) --- 0 for r = 1,
ar
(2.28)

where il = . d R / g is the dimensionless eigenvalue of the problem and

A = H / R the dimensionless depth of the liquid.
The solution of this boundary problem is equivalent to the variational
problem for the functional

(2.29)

We shall find the minimum of (2.29) in the class of harmonic functions.

By Green's formula, the functional can then be transformed into the form

F(@)=
stc5: @-&-A
5
s
@ads

or

(2.30)

The equation of the surface S is Y cos 8 = h - 1. I t is easy to show that

on the surface S
M-
_ a@
--cos~---=cose
a @ sin 8 a a sin8 @ cos BAS@
az ih a0 r ar aeh--1
and
120 N. N. MOISEEV AND A, A. PETROV

On the surface

where B0 = arccos (h - 1).

By taking into account all these formulae we can write the fuiictional
(2.30) as
2n 0.

In this formula

e1=B0 for h-l<0 02=n for h-l<O,

8,=0 for h-l>O f3,=8, for h-l>O;
eSmeans that in the expression for @ one should take Y = h/cos 8, and
means that Y = 1.
Since the solution must be periodical in r], this and the form of the func-
tional (2.31) implies that the solution of the extremal problem can be found
in the form
cos
(2.32) @ = U(r,8) nr] n =0,1,. . ..
sin
The substitution of this expression into the functional (2.31) finally
leads to a reduction of the problem of free oscillations in the sphere to a
variational problem for the functional

4
(2.33)

in a class of functions satisfying the conditions (2.26). I t follows from here

that the eigenvalues of the problem have double multiplicity, one eigen-
 F R E E LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 121

function containing the multiplicator cos nq and the other sin nq. The
exception is the case 4t = 0 , where the eigenvalues are simple. The axi-
symmetrical forms of liquid oscillations correspond to this eigenvalue.
The solution of the variational problem (2.33) is constructed by the Ritz
met hod.

FIG.17.

It is possible to take as coordinate functions the set of particular solutions

of the Laplace equation

Um= vmPrnn(cos
6) m = 1,2,. . .

where Prn"(cos0) is the associated Legendre function of nth order, and to seek
the minimizing function U in the form
N
(2.34 u = C a,,,pPrn*(cose).
m = l

As it was shown in Section 1.1 we are led to a system of homogeneous

algebraic equations (1.13)whose coefficients in the given case have the form
122 N. N. MOISEEV AND A. A. PETROV

The solution of this system can be obtained by an electronic computer.

Let us give the results of the computation for the single-node oscillations
(fi = 1). Figure 17 shows the dependence of the first eigenvalue Al on the
approximation number N for Iz = 0.20 (solid line) and for h = 1.6 (dotted
line). These graphs illustrate the rate of convergence of the Ritz method.

FIG. 18.

The dependence of the first eigenvalue Al on the liquid depth h is given in

Fig. 18. The results of the experimental determination of A, are shown by

FIG. 19.

dots in the Figure. Fig. 19 shows the shape of the principal singIe-node
oscillation for h = 1.6.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 123

4. Free Oscillations in a Container Formed by Two Horizontal Coaxial

Cylinders [ 101
Let the volume t be bounded by two horizontal coaxial cylinders with
radii R,, R,= R, + 6 and length I and introduce the coordinate system
O X Y Z , the axis X directed along the axis of the cylinders and the plane
YZ coincident with one of the end faces of the container. Direct the axis Z

FIG. 20.

vertically upwards (i,e. opposite to the vector g). The liquid depth H will
be measured from the plane X Y . Thus, H > 0 if the container is filled more
than half, otherwise H < 0 (see Fig. 20).
Introduce cylindrical coordinates R,6,X
X = X, Y = Rsin6, Z =- Rcos I?

and the dimensionless variables E,s,x by the formulae

E = r - rI. s = r,O, x =Xp,

where r = R/S, O = S/80, Go = arccos(- h/rJ for h > 0, Go = arccos

+
(- h / ( r , 1)) for h < 0.
In these variables the problem of oscillations of the liquid in the volume t
reduces to the following boundary problem
(2.35) A E , ~ ,=
~ @0 in the region t,

(2.36) for t=O and [ = 1,

a@
(2.37) -=0 for x = O and x = l ,
ax

(2.38)
a@
-= A@ for z = h,
az
124 N. N. MOISEEV AND A. A. PETROV

where 1 = a2d/g is the dimensionless eigenvalue of the problem and

h = H/d the dimensionless depth of the liquid.
Let us restrict ourselves t o the investigation of those oscillations that
are symmetric with respect to the plane y = O.* This problem can be
solved for the region t* (half the region t for which y >, 0) by introducing
an additional boundary condition

a@ -
_ -0 for s = 0.
(2.39)
as

As has been shown previously, the boundary problem of the kind (2.35)-
(2.39) is equivalent to the variational problem of minimization of the
functional (1.9). In the case under consideration this functional will have
the form

1 1

where

rl arccos [ - h/(r,+ 41
~ ( 6=) -
$0

is the equation of the free surface.

From the boundary conditions (2.37) and the form of the functional
(2.40) it is seen that the minimum of the functional F(@) should be found
among functions of the form

(2.41) @ = CJ(t,s) cos nnx, n = 1,2,. .. .

* This restriction eliminates the consideration of liquid oscillations similar to those
in communicating vessels. Such a problem requires special considerations. Note that
here, perhaps, with sufficient practical accuracy, the methods of computing liquid
oscillations in balancing reservoirs can be applied.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 125

By substituting the expression (2.41) into the functional (2.40) we come to

this result: the problem of free longitudinal oscillations in the region t*
reduces to a variational problem for the functional

To each value of n in the functional (2.42) corresponds a mode of longi-

tudinal oscillations. For example, n = 1 corresponds to one half-wave
along the length of the container, n = 2 to two half-waves, etc.

FIG.21.

We shall again solve the problem (2.42) by the Ritz method. For the
realization of the Ritz method one should construct a complete set of co-
ordinate functions {x,,}.
For this purpose consider the following auxiliary boundary problem
for the region do:

(2.45) -ax= o for s = 0,

as

(2.46) ax
-- for s = rl.
as - ox
126 N. N. MOISEEV AND A. A. PETROV

o is the eigenvalue of this problem. The region A , is shown in Fig. 21.

The solution of this problem can be found in the form x = a(E)P(s). I t
follows from the conditions (2.43)-(2.46) that the functions a ( ( ) and P ( s )
should be the solutions of the boundary problems

(2.47) ~-
dab
ass k2P = 0. dP
-= 0 for s = 0,
dP
-= o@ for s = Y,;
as as

(2.48)

(2.49) --
da - 0 for t=O and f = 1.
dE
The solution of the boundary problem (2.47) is of the form

cosh ks
(2.60) p=- cosh kr, ’ w = k tanh kr,,

where k is to be determined.
By transforming the variables, (2.48) can be reduced to Bessel’s equation
whose solution will be Bessel functions of imaginary order and argument.
However, since we need the set of coordinate functions for a numerical
solution of the variational problem it is quite sufficient for us to find an
approximate but simpler and more convenient solution of the problem
(2.48), (2.49).
For such a solution we will use the following ideas. Note that for rl-+oo
the boundary problem (2.48), (2.49) takes the form

g = O for E=O and(=l,

da

and for small 8 has the solution

(2.61) av= cos p V t ,

where
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 127

From the latter we find that

(2.62) _
kv -
-
vz + 1/(. ). + (q)l
1

I
I----
2
\ ”
, v = 0,1,2,.. .
80 2
The physical meaning of the solution (2.51), (2.62) is quite clear. For
large values of the inner cylinder radius r, the wall curvature is small. The
entire motion is taking place in a layer of the order of a wavelength, i.e.,
of the order of 6, and the motion in the limit for r, 00 is similar to that
-+

in the space between vertical walls of infinite length. This motion is des-
cribed by a solution of the kind (2.51), (2.52).
Assume temporarily that the container is such that Y, is large and 811
is small, i.e., the length and the radius of the inner cylinder are larger than
the “width” 8.
It is physically clear that in this case the wavelength along the cylinder
axis will be large compared with the wavelength in the transverse direction.
Therefore, for not too large values of n, the number (r1/8,,)kcan be con-
sidered large compared with the number nndll. The equation (2.48) can then
be considered as an equation with the large parameter y = (rl/80)k which
makes it possible to apply asymptotic integration methods.
In [ l l ] the equation

finite for y-.

is considered, where y is the large parameter, and the function b(y,t) is
00 and 0 < <
6 Q. Paper [ll] shows that with accuracy
up to terms of order l/y the solution of this equation can be written in the
form
E E

where C, and C2 are arbitrary constants and yl(E) and y,($) linearly indepen-
dent solutions of the equation

describing the oscillatory motion.

By applying this formula to the construction of the solution of (2.48)
we obtain
a = A cos [to(€)] + B sin [to(€)],
128 N. N. MOISEEV AND A. A. PETROV

where

From the boundary conditions it follows that B =0 and

.. .

(2.64)

Note that for rl -* do and small S/1 the formulae (2.53), (2.64) imply

These expressions precisely coincide with (2.61), (2.62).

Thus, with accuracy up to terms of the order 8,/b, the solution of the
boundary problem (2.43)-(2.46) is

(2.66)

where the functions wv(E) and the numbers k, are determined by (2.63)
and (2.64).
For h = 0 the expressions (2.56) give an approximate solution of liquid
oscillations in the region T,which is the more precise the greater the ratio 116.
The set of functions xv is complete in the region A , bounded by the
circumferential arcs and the radius 6 = 8,. The region A bounded by the
line z = h is completely immersed in the region A,, hence the set {xv}defined
by (2.65) is also complete in the region d and can be taken as coordinate-
function system in solving the variational problem (2.42)by the Ritz method,
setting
N
(2.66) u = v2- 0 avx..
 F R E E LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 129

The procedure described in Chapter I leads us to the set of homogeneous

algebraic equations (1.13) whose coefficientsare calculated from the formulae

This set was used for the calculations carried out on a computer. As a result
the dependence of the eigenvalues on the parameters rl, It, and 118 has
been obtained and is shown in Figs. 22-24. In all computations n = 1 was
assumed.

h 0.0

h:-1.67

-= 5.0

FIG. 22.

The dependence of the coefficients of the decomposition (2.66), av, and

of the eigenvalues & on the approximation number is shown by the graphs
of Fig. 26 and by Table 2 for different values of the problem parameters.
130 N. N. MOISEEV AND A. A. PETROV

These results indicate that for lhJ<Y, the solution converges rapidly. As
it was to be expected the convergence becomes worse if lhl> r,.

2.2
2.0
1.8
1.6

1.4
xt
1.2

1.0
0.8
0.6
0.4

0.2

1.0 2.0 3.0 4.0

I

FIG.25.

50

40

30
A,

20

10

5 10 15 :

-
I
S
FIG.24.
 F R E E L I Q U I D OSCILLATIONS, MOTIONLESS CONTAINER 131

TABLE2

I, = 0.67 11s = 6.0 h = 0.0 I, = 1.6 11s = 6.0 h= - 0.3

N = l N = 3 N = 6 N-1 N = 3 N = 6

a1 1.oooO -1.2793 -1.2794 1.oooO 0.9374 0.9378

a, - -0.0256 0.0254 - 0.0457 0.0472
- 0.0037 -0.0029 - 0.0039 0.0088
a4 - - 0.0032 - - 0.0084
- - - 0.0048 - - 0.0047
- -
@b
a, - - 0.0104 0.0036

Table 3 presents the matrix composed of eigenvectors of the

system (2.27) for rl = 1.5; h = - 0.3; and 1/13 = 5.0. The matrix is

N
FIG.25.

characterized by the following feature: the largest value among the

coefficients a, occurs for that index which coincides with the index of
the eigenfunction.
132 N. N. MOISEEV AND A. A. PETROV

TABLE3

3 4 5 6

1 0.9378 0.2881 0.0502 0.0655 0.0228 0.0324

2 0.0472 1.7687 0.6487 0.1826 0.1618 0.0472
3 0.0088 0.1748 2.2497 1.0929 0.3585 0.2432
4 0.0084 0.0551 0.3682 2.8795 1.7666 0.6259
6 0.0047 0.0642 0.1297 0.5788 3.6824 2.4367
6 0.0035 0.0098 0.0942 0.0774 0.7724 3.8159

Y = coefficient index in the sum (2.50); n = eigenfunction index.

5. Free Oscillatiorts i n a Cylirtdrical Container with Horizontal Axis [ 4 ]

For oscillations in a circular cylinder, whose axis is perpendicular to the

direction of the vector g, denote the base radius by R, and the cylinder
length by 1 and introduce the Cartesian coordinates O X Y Z , the X axis

FIG. 26.

coinciding with the cylinder axis, the plane YZ being one of the end faces,
the axis Z directed opposite t o the vector g. Again we shall measure the
liquid depth H from the plane XY as indicated in Fig. 26.
Here our dimensionless variables are

where d = R , if H 2 0 and d = v R 2 - H 2 if H < 0. In these variables

the problem of free oscillations reduces to the following boundary problem:

(2.67) (+) aw
a
w+4,,@=0 in the region t.
 F R E E L I Q U I D OSCILLATIONS, MOTIONLESS CONTAINER 133

(2.58) -a@
=o
ax
for x = O and x = l ,

(2.59) -a@
=o
ar
for r = ro (r2= z2 + y2),
(2.60)
a@
-= A@ for z = h.
az

Here A = a2d/g are the eigenvalues of the problem, h = H / d the dimension-

less liquid depth, and ro = R/d is the dimensionless cylinder radius.
The boundary problem (2.57)-(2.60) is equivalent to the variational
problem of determining the minimum of the functional

(2.61) F(@)= I(. ).

7
- AI@'dS
S

Using the Ritz method again, we construct the set of coordinate functions
by the procedure described in Section 1.2. For this purpose immerse the
container in a parallelepiped with length 1, height R H , and width 2d. +
For the set of coordinate functions we shall take that set which describes
the natural oscillations of the liquid in the parallelepiped. Again, it was shown
in Section 1.2 that these functions have the form

1
= N,c", cos mny
cosh K,,,(")(z ro)+
+("),,
cosh ~,,,(")(h + ro)'OS nnxJ

1 2m + 1 cosh R,,,(")(z + h)
&(") = N,o sin ~

2 nycosh &,(*)(A + ro)cos n m ,

&(") = R,,,(")tanh~,,,(")h,(&("))a = (T
2m+ 1
n)'+ (T)
nnd
, m,n =0,1,2,. . ..

(2.62)
We shall determine the solution of the variational problem (2.62) in the
form
N N
(2.63) @= 2 Urn(")+,,,(*) + 2 b,,,(")I?,,,("),
m.* m,n

where +(",) and $,,,(") are given by (2.62).

134 N. N . MOISEEV AND A. 4.PETROV

As was shown in Chapter I, we obtain in the result the set of equations

(1.13) whose coefficients must be calculated by the formulae (1.18), (1.19).
In the present case, (1.13) will have the form
N N
2
=0
&,s
$:i)@kCs) + k,s2
=0
&y)bk(') - 1

where

82)= 1 if m = n, k = s and 13:;) = 0 in all other cases. Under 2, ob-

viously, only the lateral surface of the cylinder should be understood here.
Introduce the polar coordinates Y and 8 in the plane x = 0 by z = --rcos8,
y = Y sin 8. It is not difficult to show that, for example,

where h = - yo cos 8, and in the integrand z = - yo cos 8, y = yo sin 8.

Correspondingly,

where z = h in the integrand.

Consider now the structure of the expressions (2.66) in detail. Each
function with upper index n has the factor cos nnx. Therefore, in the
expressions (2.68) the integration with respect to x from 0 to 1 gives a result
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 136

differing from zero only in the case n = s. This means that the set (2.64)
decomposes into N independent sets which correspond to n = 0 , 1 , . . . , N .
The solution of the set corresponding to n = no describes oscillations
of the form @ = U(y,z)cos nonx which comprise no half-waves along the
axis x . Therefore we shall write from now on only one upper index 12 for all
the coefficients of the set (2.64).
It also follows without difficulties that the functions t,hk(") (at,h,,,(")/an)
and $?)(&(")/an) are even and the functions $k(")(a+,,,(')/a?8) are odd, both
with respect to 8.
Correspondingly, the functions +m(")+k("), $,,,(")$k(") are even, and the
functions t,h,,,(")$p)
are odd with respect to 8. It follows from here that
cly = 0 and PLi = 0; thus for each n the set (2.64) decomposes into two
independent sets
N N
(2.66) 2 ptiak(*) -A 2 &ak(nj = 0, m =0,1,. .. ,
k=O k=O

N N

whose coefficients are determined by the formulae

[ K , , , W-
- K,,,(") cos (mmocos e) sinhcosh ~ (cos
~ O)]
+ yo)
~,,,(")(h
cos e

cosh [K,,,(")Y~(1 - cos 8 ) ]

- mn sin (mnrocos 8 )
+
cosh ~,,,(")(h r0)

sin (2mn sin 0,)

for k =m

=I
4mn
+
sin [(m- k ) n sin €lo] sin [(m k)nsin O0]
for k # m,
2(m - k ) n -k 2(m k ) n +
136 N. N. MOISEEV AND A. A. PETROV

2k +1
2 ny sin ~ 2 ZYdY
sin 8.

+
sin [(2m 1)nsindo]
+ for k= m

=l
2(2m 1)n

sin [(m- k)nsin O,] sin [(m+ k + 1)nsin O,]

2(m - k)n
-
2(m + k + 1)n for k # m.

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8

FIG.27.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 137

TABLE4

n=O h = 0.2 n = O li = - 0.6

-\. 1 3 1 3

approx. accur. approx. accur. approx. accur. approx. accur.

1 1.0200 1.0184 -0.0600 -0.0444 1.0000 0.9809 0.0498 0.0480

2 0.0134 0.0138 0.0944 0.0741 -0.1110 -0.1632 -0.2269 -0.2907
3 0.0101 0.0106 1.0200 1.0130 -0.0498 -0.0849 1.0000 0.9185
4 -0.0075 -0.0080 0.0859 0.0899 0.0254 0.0482 -0.2636 -0.2221
5 -0.0056 -0.0061 0.0406 0.0430 0.0155 0.0321 -0.1183 -0.1195
6 0.0044 0.0049 -0.0255 -0.0281 -0.0106 -0.0231 0.0760 0.0770

v = eigenfunction number, m = number of the coefficient b,(") of the series (2.63).

-
1
R

FIG. 28.
138 N. N. MOISEEV AND A. A . PETROV

The solutions of the set (2.66) describe the modes of oscillations that
are symmetric with respect to the plane y = 0 ;those of the set (2.67) describe
the modes asymmetric with respect to the same plane.
These sets were solved on an electronic computer. The results of the
computations are shown in Figs. 27 and 28. Fig. 27 shows the graphs of the
dependence of the dimensionless natural frequencies A = aaR/g on the liquid
depth for n = 0 (solid curves 1) and for n = 1, l / R = 1 (solid curves 2).
The approximate calculation method of the natural oscillations is des-
cribed in Section 1.3. The results of the calculations of eigenvalues by
formula (1.37) are shown in Fig. 27 (dotted curves). The typical values of
the coefficients of the series (2.63), obtained by the escalator method on a
computer and by formula (1.41) for various values of the variables, are
given in Table 4. From the comparison of all these results one can see that
the approximate method can be successfully applied in estimating the
natural oscillations of liquid in containers.
The dependence of natural frequencies on the relative length 1/R of the
container for n = 1 and h = - 0.8 (solid curves) and h = 0.2 (dotted
curves) is shown in Fig. 28.

6. Free Oscillations i n Circzclar Cylindrical Containers whose Bottom and

Cover are Spherical Caps [4]

Let the axis of the cylinder be parallel to the vector g and introduce
the coordinate system O X Y Z the plane X Y of which coincides with the plane
of the free surface of the liquid, the axis Z being opposite to the vector g.

FIG. 29.

Denote the cylinder radius by R, the liquid depth by H,the cover radius
by H,, and the free-surface radius by L (see Fig. 29). The dimensionless
variables are
x=-
X y = -Y z
R’ R’ R
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 139

Cylindrical coordinates r,f3,z are introduced by

x = icos 8. y = r sin 8. z = z.

The principal modes and dimensionless frequencies of free oscillations

in these containers coincide with the eigenfunctions and eigenvalues of the
following boundary problem :
(2.68) A@=O in the region t,
a4j
(2.69) --=A@ for z = 0,
az

a@ -
_ -0 for
(2.70) Y = 1,
a?-
(2.71) -a@
=o on the surface 2.
an
Here, the wetted part of the bottom and cover surfaces of the cylinder is
denoted by 27, and the dimensionless natural frequency by 1 = a2R/g.
The boundary problem (2.68)-(2.71) has the equivalent formulation
which requires the determination of the minimum of the functional

(2.72)
I
F(@) = (V@)%*- 1 @ V S .
1 S
I
Using again the Ritz method, let us use the method described in Sec-
tion 1.2 for the construction of the coordinate functions. Circumscribe
our cylinder with a flat-bottomed circular cylinder with radius R and
height H (Fig. 29). Take as coordinate functions the set which describes
the modes of oscillations in that cylinder. In Section 1.2 this set was
given as

= K,,,(*) tanh K,,,(*&,

,urn(*) (Nrn(n))'= f (1 - A) Jn2(Krn'*'),

(2.73)
where Js(~,,,(n)~)are Bessel functions of the first kind, the numbers K,,,(") are
the roots of the equation J,,'(K) = 0, and It = H / R the dimensionless depth
of the liquid.
140 N. N. MOISEEV AND A . A. PETROV

Similarly to what was done in Sections 11.2 and 11.3 one can show that
the spectrum of the problem under consideration has double multiplicity
(except for the case n = 0) so that two eigenfunctions, one with the factor
cos no, the other sin nf3 as factor, correspond to each eigenvalue. Thus the
solution of the variational problem (2.72) can be found for instance in the
form

As was shown in Section I, we finally reduce the problem to the solution

of the set of homogeneous algebraic equations (1.13) whose coefficients
must be calculated by the formulae (1.18)-(1.19). For example, if a cylinder
has a convex upper cover and the liquid level is inside the upper spherical
surface, then the coefficients of the set (1.13) are of the form

0 1

where

1 = L/R, ro = R,/R.
If the liquid level is inside the lower spherical surface, then the region
under discussion should be put inside the cylinder of height H and of radius
k = L = //2RoH - H2. In the transition formula to the dimensionless
variables and also in the expressions for the function $2) and the coefficients
pz and q2, the value R should be replaced by 8.
After determination of the eigenvalues, they must be multiplied by
RIIBR,,H - H 2 .
The set of equations (1.13) was solved on an electronic computer
by the escalator method. The maximum number of the terms N of the sum
(2.74) has been taken equal to seven. We do not intend to give the graphs
of the dependence of the eigenvalues A on the number N , because they are
of the same kind as the corresponding graphs given in the previous sections.
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 141

TABLE6

1 2 3 4 5 6 7

1 0.9974 -0.0681 -0.0195 -0.0094 -0.0056 -0.0036 -0.0024

2 0.0640 0.9867 -0.1079 -0.0481 -0.0249 -0.0161 -0.0099
3 0.0259 0.1263 0.9795 -0.1405 -0.0544 -0.0298 -0.0186
4 0.0147 0.0692 0.1224 0.9792 -0.1349 -0,0648 -0.0308
6 0.0096 0.0356 0.0610 0.1158 0.9808 -0.1276 -0.0528
6 0.0067 0.0241 0.0383 0.0591 0.1097 0.9828 -0.1187
7 0.0050 0.0174 0.0267 0.0380 0.0566 0.1037 0.9868

y = eigenfunction number; M = coefficient number in the sum (2.74).

11
1 I I I 2-n=i

10

6
Xi
5

0 1 2 3 4 5 6
h
FIG.30.
142 N. N. MOISEEV AND A. A. PETROV

We shall only note that, as was to be expected, the convergence of the process
became worse when the liquid level was either in the higher or in the lower

11

10

6
Xi
5

3
t

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

h

FIG. 31.

of the two spherical caps. However, even in these cases, for calculations
of the first three eigenvalues with a relative error less than 1%, it was suf-
ficient to take only seven terms in the sum (2.74). If the liquid level was
inside the straight cylindrical portion of the container, then for obtaining
the same accuracy three terms of the sum (2.74) were sufficient.
 FREE L I Q U I D OSCILLATIONS, MOTIONLESS CONTAINER 143

Table 6 gives the matrix whose columns represent the eigenvectors of

the system (1.13) at ro = 1.36; h = 0.5 and 1z = 1. As it is seen, in this
case again the diagonal elements are larger than the other elements.
The dependence of the dimensionless natural frequencies on the liquid
depth h for a cylinder with convex upper cover is shown in Fig. 30, The
solid curves correspond to the container whose total height is equal to
6.5 R, while ro = 1.33. The dotted curves characterize the container
the overall height of which is equal to 2.2 H, and ro = 1.1.
The dependence of the frequencies on the liquid depth h for concave
upper cover is shown in Fig. 31. For the calculations it was assumed that
the overall height is equal to 1.6 R and ro = 1.1.

7 . Free Oscillations in a Container of Toroidat! Shupe [4]

The liquid occupies part of a container in the shape of a torus whose

axis is parallel to the vector g. Denote the radius of the meridianal cross sec-
tion of the torus by R , and the inner radius of the central cross section,

FIG.33.

perpendicular to the torus axis, by R,, the inner and outer radii of the
free surface of the liquid by L, and L,, respectively. Introduce the coordinate
system O X Y Z whose axis Z coincides with the torus axis and is opposite
to the vector g, and the plane XY coincides with the plane of the free surface.
We shall measure the liquid depth H as it is shown in Fig. 32.
Introduce the dimensionless variables
X Y z=- z
x = - -R
,O y=q’ RO
and the cylindrical coordinates r, 8, z by the formulae
x=rcos8, y=zsin8, z=z.

In contradistinction to the containers discussed previously the torus is

a doubly-connected region. We therefore encounter here a considerably
larger number of forms of possible free oscillations than before. First there
144 N. N. MOISEEV AND A. A. PETROV

are standing waves. Their superposition gives us progressive waves which

can propagate in either direction. This motion is characterized by zero
circulation. In addition, there are different oscillations which appear under
the condition of non-zero circulation. These motions can also have a certain
practical interest. However, the authors intend to investigate only motions
of zero circulation. In order to study these motions it is sufficient for us to
consider standing waves only. Their determination reduces, as we know,
to the problem of finding the minimum of the functional

(2.76) !
F(@)= (P@)'dt - iZ @'dS,
s I
to which we again shall apply the Ritz method. Let us construct the system of
coordinate functions by the procedure described in Section 1.2. Circumscribe
the region t with annular cylinders bounded by cylindrical surfaces of radii
R, and cR, = R, + 2 R , and by planes Z = 0 and Z = - H. As coordinate
functions we shall take the set of functions which represent the natural
oscillations in the region between two coaxial circular cylinders of height H.
The solution of this problem is given in Section 1.2 and has the form

(2.76) pJn) = K,,,(") tanhK,(")k, m = 1,2,. .., n = 0,1,2,. . .,

P,(r) = N,'(KJ"))Jn(Km(n)r) - J;(Knr("))Nn[Km(")r),
where ],,(K,(%) and N,,(K~(")Y)
are Bessel functions, ,A") is the root of the
equation
]1II(K)Nd(CK) - J,)(CK)N1II(4 = 0,

it is the liquid dimensionless depth, and c is the dimensionless outer radius

of the central cross section perpendicular to the torus axis.
The normalizing factor N,(*)can be written as

It is not difficult to show that the spectrum of this problem is twice degenerate
for rt # 0, i.e., to each eigenvalue correspond two functions one of which
includes the factor cos no and the other sin d3. Thus the solution of the
variational problem (2.75) can be sought for in the form
 FREE LIQUID OSCILLATIONS,MOTIONLESS CONTAINER 146

As a result we reduce the problem to the solution of the set of homo-

geneous algebraic equations (1.13) the coefficients of which should be cal-

FIG.33. Key: Solid line ___ : a = 0.6; dashed line - - -: a = 0.4; open dots
represent experimental points when o! = 0.6 ; solid dots ( 0 )represent experimental
(0)
points when a = 0.4.

culated by (1.18) and (1.19). If in the case under discussion It > a, where
a = R/R,, then
146 N. N. MOISEEV AND A. A. PETROV

[$,(n)t,~)], = Ndr,
0 1,

where the subscripts 1, 2 mean that Y should be set equal to

respectively. The dimensionless radii of the free surface are denoted by

1, and I,.
TABLE6

n = l n = 2

h -\Y 1 2 1 2

1 1.406 -0.122 1.240 -0.186

0.760
2 0.061 0.891 0.166 0.878
1 1.160 -0.08 1 1.120 -0.134
0.67'1
2 0.020 1.096 0.074 1.084
1 1.066 -0.044 1.066 -0.073
0.681
2 0.008 1.088 0.028 1.084
1 1.013 -0.016 1.013 -0.021
0.491
2 0.004 1.024 0.009 1.021
1 1.000 - 0.007 1.000 -0.010
1.ooo
0.401
2 0.006 0.008 1.000
1 1.000 -0.008 1.000 - 0.016
1.ooo
0.288
2 0.006 0.016 1.000
1 1.ooo -0.007 1.000 -0.018
0.179
2 0.006 1.000 0.017 0.998
1 1.000 -0.007 1.003 - 0.009
0.064
2 0.006 1.000 0.022 1.000
1 1.000 0.000 1.000 0.000
0.016
2 0.000 1.000 0.000 1.000

Y = eigenfunction number; m = coefficient number in the sum (2.77).

 FREE LI QUID OSCILLATIONS, MOTIONLESS CONTAINER 147

If the liquid fills less than half of the torus then it is more convenient
to describe the region by cylindrical surfaces of radii R, = R, + R -
V2RH - H2 and c" R, = R, + +
R V2RH - H2. In this case 1, = 1 and
1, = c". In the transition to dimensionless coordinates in the expressions
for the functions $(,") and the coefficients py
and q t i the values R, and c
should be substituted for a,, and E. The eigenvalues obtained from the
solution of the set (1.13) should finally be multiplied then by Rol Po.

0.6

0.5

0.4

'4 0.3

0.2

0.1

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
h

FIG.34

The system (1.13) has been solved by the escalator method on an elec-
tronic computer. It turned out that a t any values of the problem parameters
it was enough to take eight terms in the sum (2.77) in order to calculate
the first two natural frequencies with a relative error less than 1 per cent.
Figure 33 shows the dependence of the first two dimensionless natural
frequencies li = ai2R/g on the liquid depth h in two containers, the first
with a = 0.6 (solid curves) and the second with a = 0.4 (dotted curves).
This figure also contains the results of experimental measurements of
these frequencies for the container with a=0.6 (open circles) and for the
container with a = 0.4 (full circles).
In Table 6 the first two coefficients of the sum (2.77) for the case a = 0.4
are given.
The first natural frequency was also computed by the approximate
formulae (1.37) and (1.38). Figure 34 shows the dependence of A, on It,
computed by the escalator method on a computer (solid curve) and by the
approximate formula (1.37) (dotted curve). In this case also the ap-
proximate formula gives satisfactory results.
148 N. N. MOISEEV AND A. A. PETROV

APPENDIX

The Escalator Method for Solving the Equation P x = AQx

The application of the Ritz method reduces the problem of liquid os-
cillations in a container to the solution of the linear homogeneous equations
(1.13). If we introduce the matrices

p= ~pm\l:m=l# Q = IlqnmlIL1
and the vector x(a,,.. .,a,) then the system (1.13) can be replaced by the
matrix equation
(A.1) Px = AQx.
Since the problem is self-adjoint and kinetic and potential energies are
positive definite, the matrices P and Q are symmetrical and positive definite.
In the process of designing a scheme of calculation for the natural fre-
quencies and modes of free oscillations there have been tested various methods
to construct the solutions of the system (A.1). A. A. Petrov, Yu. V.
Pukhnachev and Yu. P. Popov have developed a special form of the escalator
method which makes it possible to improve sequentially the unknown
values. For the simplest case the matrix Q is unitary and the roots
of the equation (A.l) are simple this method has been discussed in various
textbooks (see, for example, [12]).
The modification of the escalator method for the case where Q is an
arbitrary, symmetrical, positive definite matrix is discussed below.*
The idea of the escalator method is as follows. Consider the equation
(A4 PRX = A Q R X .
The matrices P R and QR are composed of the R first elements in the R
first rows of the matrices P and Q. Assume that the solution of this equa-
tion is known. Denote its eigenvalues by &(R) and the eigenvectors corre-
sponding to these eigenvalues by x , ( ~ ) .
Further, compose the matrix P R + l by adding to the matrix P R the R
+
first elements of row and column number R 1 and the element $ R + l , R =a
located on the diagonal. Let us call this procedure “bordering of the matrix
PR.” Denote by p the row (column) of R elements which we have added
to the matrix. Similarly, border the matrix Q R by the row (column) q and
the element p located on the diagonal.
Consider the equation

(A.3) p R + l x = AQR+,l_x*

This method is discussed in [a].

 F R E E L I Q U I D OSCILLATIONS, MOTIONLESS CONTAINER 149

Denote by l(R+l) one of the eigenvalues of this equation. Denote further

the R-dimensional vector by x ( ~ + I ) which, together with the element y,
.
forms a corresponding eigenvector ( x ~ ( ~ .+ ~.,)x ~, ( ~ + ' ) , Yof) the equation
(A.3).
Here it is always assumed that (Qx,,xs) = drS where d,, is Kronecker's
symbol.
Similarly to the procedure in [12] one can derive the escalator equation

with
A , c R ) ( A ) = (p - Aq,xP),

and the formulae for the calculation of the eigenvectors

The formulae (A.4)-(A.6) make it possible to find a solution of system

(A.3) if the solution of the system (A.2) is known. For R = 2 the equation
(A.2) is equivalent to a quadratic equation.
Sometimes it is more convenient to use the formulae (A.4)-(A.6) in the
form

Proceeding from the minimax properties of the eigenvalues (see, for in-
stance, Chapter I, $ 4 of [13]) it is easy to show that the eigenvalues of the
equations (A.2) and (A.3) are intertwined as follows:

(A.lO) 3y(R+1) < < &(R+l) < &(R) < . . . < 1 ~ \<( ~Ak",i".
)
160 N. N. MOISEEV A N D A. A. PETROV

Consider some specific cases which may occur in the calculations:

(a) Assume that the value [A,(R)(A$R))\
for Y = v, is small. Take from
the formula (A.7) the term of the last sum with the index r,, denote by
FRfl(A)the remaining part of the escalator function ER+l(A), and set

I t is evident that the smaller the value (y1, the closer to A!? lies one of the
zeros of the escalator function A(R + l). The approximate relationship

2;'- p + u = - Y
can be improved:

Here the prime means differentiation with respect to I.

However, if IyI is less than the chosen accuracy of the calculation, one
can set Z R + l )= A!:). Here, within accuracy of the order [A!?(AIf')la, the
following equalities hold :

(b) If the equation (A.2)has a root of multiplicity Y , &(R) = &(R) =

...= A,.(R), then the equation (A.3) will at least have a root of multiplicity
v - 1, i.e.,
&(R+1) = jZg(R+l) = . .. = j l * ( R t 1 ) ;

this follows from the inequalities (A.lO). The inequalities &(R+l) <
&(R)
<
and k(R) 2:;') show that the multiplicity of this root may reach v
and, finally, Y + 1 if the given inequalities are replaced by equalities. Thus
in the transition from (A.2) to (A.3) the root multiplicity may either de-
crease by unity, remain unchanged, or increase by unity. N o other
possibilities exist.
Gmsider in more details the case when the root multiplicity decreases
by unity. To the eigenvalue I(R+l) of multiplicity Y - 1 we should put
in correspondence P - 1 eigenvectors. Suppose that all AI'R)(AI'R)) # 0,
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 161

.
Y = 1,2,. ..v. We determine x,('+l) in the form of a linear combination
of the vectors xJR)with y = 0, s = 1,2,. , .,r + 1:

xp+1)= x p + & 1 x p J + . , . + LdT1, Y = 1,2,. ..,Y - 1.

The coefficients t,s are determined from the condition of orthogonality of
the vector x , ( ~ + ~ ) to the previously constructed x A R + l ) (where m< I)
with respect to the matrix Q R + i and from the condition

(p - A1(R)Q,X,(R+1)) = 0.
These conditions give us the set of equations

(QR+lX,(R+",X1(R+l)) =1 + tlltrl = 0,

( Q ~ + i d ~ + ' ) , ~ g ' ~ +=
~ ) 1) + t&i + t&,2 = 0,
.......................................
(p - A p q , x , c R + ' ) ) = A J R ) ( A + R ) ) + t,lA2(R)(Al(R)) + . . . + t,,A;yl(Ap) = 0.

Let now some of the AJR)(AJR)) be equal to zero. Then the corresponding
X,(~+I) is assumed to be equal to x i R ) , and the rest is constructed by the
procedure discussed above. But applying this procedure in the case where
all A J R ) ( A i R ) ) are equal to zero, we would be able to find for an (v - 1)-
fold eigenvalue v instead of v - 1 linearly independent eigenvectors, which
is impossible. However, one can prove the following statement : The vanishing
of all A1[R)(A,,,(R)), .
m = 1,2,. . ,v is achieved then and only then, when the
multiplicity of the root A;R+l) of the equation (A.3) is equal to or greater
than the multiplicity of the root &(R) of the equation (A.2).
The proof of this statement is rather complicated, and we do not present it
here. Thus, if the root multiplicity reqains, to each of the roots A,,,(R+l)=
m = 1.2,. . ..v we can put in coeppondence the vector x , ( ~ + ~ ) assuming
x J R + l ) = xJR), y = 0. In solving the escalator equation we may obtain
one more root equal to jzl(R) and'from formulae of the kind (A.S), (A.6)
the eigenvector corresponding to it. The root multiplicity, consequently,
will increase to v + 1.
Systematizing what was said ahqve one may explain the following
procedure of solving the escalator equation E R + l ( A ) = 0. We calculate
first

and on the basis of their closeness (or equality) to zero we decide on the
closeness (or coincidence, respectively) of zeroes of the escalator function and
162 N. N. MOISEEV AND A. A. PETROV

+
its poles, Partition, further, the semi-infiniteinterval [0,m) into R 1intervals
.
by the points {A$R),. . , i l R ( R ) } . Assume that one of the points at which
1yrJ is small (or equal to zero) lies between points at which (y,1 is, at least,
of the order of unity. Between these two points there are two eigenvalues
of the equation (A.3). One of them is close to ilf) and has already been
taken into account. If Iy,,J = 0, then, in order to find the other root we take the
whole interval [A!:!- l , A ! ~ ~ l ] and, simultaneously, from the last sum of the
formula (A.7), which determines the escalator function, we omit the fraction
with the index Y,. In case of coincidence of the root determined with z?
we notice an increase of the root multiplicity. If lyr,l # 0 then, for the
determination of the second root out of the two intervals [A!:i,,#:)] and
[A!P’,il!:il], we take either one or the other depending on whether the value
FRfl@)) is positive or negative. Similar reasoning can be camed out
in the case when between the points at which 1y.l is of the order of unity
there are some points at which lyrl is small or equal to zero.
We use the same procedure if the root multiplicity Y> 1, examining
the expression

The calculation of a zero in the finite interval makes no difficulties.

I t is possible to show that in the above-mentioned choice of intervals the
escalator function tends to - w (respectively +
w), when the argument
tends to the left (or, respectively, to the right) end of each interval. One
may also show that ER+l(0)< 0.
Hence the following algorithm of search is convenient: We divide the
interval into two halves and take for further search either the left or the
right half, depending on whether the escalator function is positive or negative
at this point. After narrowing the interval to a value smaller than the
preset accuracy, we stop the search. Considering the interval [AdR),co)
we do the following. All but one of the zeroes of ER+i(il) are already
known to us. Represent it as a polynomial in il of power R 1 +
and use the fact that the product of aII polynomial roots is equal to the
last term divided by the coefficient of the leading term. Hence
 FREE LIQUID OSCILLATIONS, MOTIONLESS CONTAINER 163

CONCLUSION

The work here presented to the reader is a special illustration of the

first paper in which the basic features of the Ritz method were explained
(see Advances i n Applied Mechanics, Vol. 8 ) . There the universality of the
Ritz method for solving oscillation problems of liquids and liquid-containing
bodies was claimed. A look at the problems considered here confirms to a
considerable extent this point of view. Apart from this, these -problems
are interesting in themselves, since we have chosen such shapes of containers
as are applied in various branches of engineering. The dimensionless
form of the results given here can serve as reference material, but the principal
purpose of the authors has been to show by means of practical examples
how various computational difficulties can be overcome; for, in view of the
great abundance of container shapes used in engineering, it seems hopeless
to put together a comprehensive reference book.

References

(Titles of Russian publications are translated)

1. MOISEEV. N. N., Introduction to the theory of oscillations of liquid-containing

bodies, Advances in A#di ed Mechanics 8, 233-289 (1964).
2. TROECH, B. A,, Free oscillations of a fluid in a container, in “Boundary Problems
in Differential Equations”, Univ. of Wisconsin Press, Madison, pp. 279-299 (1960).
3. BUDIANSKY, B., Sloshing of liquids in circular canals and spherical tanks, Journ.
Aevo-Space S c i . 9 7 , 181-173 (1960).
4. PBTROV,A. A., POPOV, Yu. P.. PUKHNACHEV, YU.V.. Calculation of free oscilla-
tions of a liquid in immovable containers by a variational method, Z h V M i
M F 4, 880-896 (1964).
5. PETROV,A. A., Approximate method of calculation of free oscillations of a liquid
in containers of arbitrary shape, and the Zhukovskii potentials for these con-
tainers, Z h Y M i M F 8, 958-964 (1963).
6. BORISOVA, E. P., Free vibrations of a liquid in an inclined cylinder, in “Variational
Methods for Problems of Oscillation of a Liquid and of a Liquid-Containing Body.”
Vychisl. Tsentr A N USSR, Moscow 1962, pp. 203-213.
7. DOKUCHAEV, L. V., About the boundary-value problem of liquid oscillations in
conical cavities, P M M SR, 161-156 (1964).
8. MIKISHEV,G. N., and DOROSHKIN, N. Ya., Experimental investigation of the free
vibrations of a liquid in a container, Zsv. A N USSR, Sect. of eng. sc., mech., and
mech. engineering, 4. 48-53 (1961).
9. BOGORYAD, I. B., About the solution by a variational method of the problem of
oscillations of a liquid filling a cavity partially, PMM $36, 1122-1127 (1982).
10. PETROV,A. A., Oscillations of a liquid in a container, whose shape is an annular
cylinder with horizontal generators, Zh V M i MF 1, 741-764 (1961).
11. MOISEEV, N. N., On the theory of asymptotic representation of integrals of linear
differential equations with a large parameter, Uch. sa#iski Rostovskogo Uni-
versiteta 8, 131-134 (1965).
164 N. N. MOISEEV AND A. A. PETROV

12. FADDEEV,D. K., and FADDEEVA, V. N., “Computational Methods of Linear Algebra.”
Fizmatgiz, 1960. (An English translation is available published by Freeman,
San Francisco, 1963).
13. COURANT,R., and HILBERT, D., “Methods of Mathematical Physics.” Interscience,
New York. 1963.

Abbreviations:

PMM = Prikladnaya matematika i mekhanika

AN = Akademia nauk
Z h V M i M F = Zhurn. vychislitel’noi matematiki i matematicheskoi fiziki
On Nonlinear Vibrations of Systems with Many Degrees of Freedom

BY R. M . ROSENBERG

.
Department of Mechanical Engineering. Division of Applied Mechanics. University of
California. Berkeley Califwnia

Page
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
I1 . Admissible Systems and Motions . . . . . . . . . . . . . . . . . . . . 168
1. The General System . . . . . . . . . . . . . . . . . . . . . . . . 158
2. Admissible Systems . . . . . . . . . . . . . . . . . . . . . . . . 159
3. Admissible Motions . . . . . . . . . . . . . . . . . . . . . . . . . 162
I11 .
The Trajectories in Configuration Space . . . . . . . . . . . . . . . . 163
1 . Transformation and Trajectories . . . . . . . . . . . . . . . . . . . 163
2. The Restricted Principle of Least Action . . . . . . . . . . . . . . . 166
IV . General Description of the Geometrical Method . . . . . . . . . . . . 167
V. Trajectories of Admissible Motion . . . . . . . . . . . . . . . . . . 169
1. General Properties . . . . . . . . . . . . . . . . . . . . . . . . . 169
2 . Admissible Trajectories of Autonomous Systems . . . . . . . . . . . . 170
VI . Normal-Mode Vibrations of Nonlinear Systems . . . . . . . . . . . . 173
1. A New Definition of Normal Modes . . . . . . . . . . . . . . . . . 173
2. Straight Modal Lines . . . . . . . . . . . . . . . . . . . . . . . . 173
3. Interpretation of Modal Lines . . . . . . . . . . . . . . . . . . . . 176
VII . Properties of Motions with a Rest Point . . . . . . . . . . . . . . . 176
1. The Transversals (or P-curves) . . . . . . . . . . . . . . . . . . . 176
.
2 The 2'-curves . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
VIII . Special Autonomous Systems . . . . . . . . . . . . . . . . . . . . 187
1. Smooth Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 187
2. Uniform Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 188
3. Sequential Anchored Systems . . . . . . . . . . . . . . . . . . . . 189
4. Homogeneous Systems . . . . . . . . . . . . . . . . . . . . . . . 190
6. Symmetric Systems . . . . . . . . . . . . . . . . . . . . . . . . 190
IX . Theateb-Functions . . . . . . . . . . . . . . . . . . . . . . . . . 191
1. Their Origin and Importance . . . . . . . . . . . . . . . . . . . . 191
2. The Inversions . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
3. The Amp-Functions . . . . . . . . . . . . . . . . . . . . . . . . 196
4. The Sam- and Cam-Functions . . . . . . . . . . . . . . . . . . . . 197
X. Nonsimilar Normal-Mode Vibrations . . . . . . . . . . . . . . . . . 199
1. The Perturbation Potential . . . . . . . . . . . . . . . . . . . . . 199
2 . The Curved Modal Line . . . . . . . . . . . . . . . . . . . . . . . 200
3. Integrable Cases . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4 . GeneralProperties . . . . . . . . . . . . . . . . . . . . . . . . . 202
XI . Exact Solutions to Steady-State Forced Vibrations . . . . . . . . . . . 203
1. Introductory Remarks . . . . . . . . . . . . . . . . . . . . . . . 203
2. Homogeneous Systems . . . . . . . . . . . . . . . . . . . . . . . 204
3. The Response Curves . . . . . . . . . . . . . . . . . . . . . . . . 210

166
166 R. M. ROSENBERG

XII. Steady-State Forced Vibrations by Approximate Methods . ....... 214

1. Description of the Method. ..................... 214
XIII. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2. Stability of Normal-Mode Vibrations . . . . . . . . . . . . . . . . .220
3. Homogeneous Systems . . . . . . . . . . . . . . . . . . . . . . . 221
4. Symmetric System with Two Degrees of Freedom . . . ........ 225
5. Nonsimilar Normal-Mode Vibrations . . . . . . . . . . . . . . . . . 230
8. Forced Vibrations . . . . . . . . . . . . . . .
. . . ....... 230
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

I. INTRODUCTION

The study of the vibrations of nonlinear systems with many degrees of

freedom is concerned with the search for some or all periodic solutions of
systems of nonlinear differential equations, and to deduce as many properties
of these solutions as the state of the applicable mathematical knowledge
permits. Unfortunately, this body of knowledge is limited and not unified;
in consequence, many and vaned disciplines within mathematics are com-
monly used to deduce partial results. In a general way, these results fall
into two broad categories: those which apply to systems that are “weakly
nonlinear,”* and those which apply when the systems are “strongly non-
linear.”
The results in the first category contain a good deal of detailed informa-
tion, and they resemble in many ways those familiar from linear theory.
Those in the latter category usually contain fewer details, being more con-
cerned with general questions of existence, uniqueness, boundedness and
stability of solutions.
The task of presenting a meaningful survey of the methods and results
concerning the vibrations of nonlinear multi-degree-of-freedom systems is
too ambitious for this relatively brief contribution. Moreover, several survey
articles and books, foremost among them the recent, admirable work by
Minorski [l] have served this purpose. Here, it is intended to display cer-
tain geometrical methods, and to summarize the results stemming from their
application.
The only systems considered are those whose mechanical models can be
constructed of masses and “massless” springs. One, several, or all masses
may also be acted on by time-dependent, periodic “exciting forces.” The
nonlinearities of these systems are those arising from “nonlinear springs.”
Within these restrictions, the system may be linear, “weakly” or “strongly”
nonlinear, or “nonlinearizable.” In the geometrical methods used here,
“weak” nonlinearity is of no advantage, and results for strongly nonlinear

All terms in quotation marks are defined precisely in the text.

 NONLINEAR VIBRATIONS 167

or nonlinearizable systems are frequently as easy to deduce as those of

almost linear systems.
It is plausible that general solutions (i.e. the class of all possible motions)
cannot be deduced for such nonlinear systems. However, if the class of
desired solutions is suitably restricted, one can frequently gain a great deal
of detailed knowledge regarding them, no matter how strongly nonlinear
the system. However, to be meaningful for the physical scientist, this
restricted class of desired solutions must be such as to explain and/or predict
the incidence of physically important phenomena. In linear systems, these
physically interesting solutions are the so-called “natural free vibrations,”
and the steady-state forced vibrations. It is precisely these types of motion
that are treated here for nonlinear systems.
The methods used to deduce these motions are not conventional (in the
sense of conventional present-day research in nonlinear vibrations), but they
are classical in the sense of theoretical mechanics that originated with
Hamilton, Jacobi and their contemporaries and that was described and ex-
tended relatively recently in a beautiful paper by Synge [2]. These methods
are especially useful in the search for “free vibrations in normal modes”
when the systems are strongly nonlinear. However, the concept of “normal
modes” and of eigenvalues is well defined only in linear systems. In fact, the
demonstration of the existence of eigenvalues has its roots in the theory of
quadratic forms, and its application to vibration problems requires that the
potential energy be a quadratic form. But, it is the very essence of non-
linearity, as defined here, that the potential energy is not a quadratic form.
Hence, the question of existence of normal-mode vibrations cannot be decided
in the conventional way of finding eigenvalues, and of attaching to each an
eigenfunction. Therefore, new definitions and methods are required when the
problems are nonlinear.
The usefulness of the concept of normal-mode vibration is, of course,
greatly impaired in nonlinear problems because the principle of superposition
fails. Nevertheless, the normal-mode vibration retains, even in nonlinear
systems, a position of central importance because, as will be shown, resonance
occurs in the neighborhood of normal-mode vibrations, whether the system is
linear or nonlinear. Hence, when the amplitude of the forcing function is
small, the solution of the normal-mode vibration problem furnishes the
“zero’th order” solution of the problem of steady-state forced vibration, as
demonstrated by Rauscher [3], for instance. As in the Duffing problem [a],
it is also necessary in problems with many degrees of freedom to examine the
stability of the solutions; this is done in the last section.
Although examples are given, the question of application has not been
treated in a general way. Suffice it to say that in our model a displacement
of any one mass has “an influence in depth”; i.e., its effect is not only felt
by the neighboring masses, but by all of them. Hence, it is likely that the
158 R. M. ROSENBERG

systems discussed here are applicable to crystal and molecule vibrations.

It is also quite possible that one can discuss, by the methods described here,
the mechanism of pulsating stars [S]. At any rate, these methods have been
examined in detail [S] by an astronomer having this purpose in mind.
No attempt has been made to cite in the bibliography all recent litera-
ture dealing with vibrations of nonlinear systems having many degrees of
freedom; it is, in fact, restricted to the material which was used in this
contribution,

AND MOTIONS
11. ADMISSIBLESYSTEMS

1. The General System

The general system considered here may be modelled on a great variety of

mechanical, electrical, or other devices, and its configuration may be deter-
mined by many different types of physical measurements. One such model

FIG.1. Mechanical model of spring-mass system.

is Figure 1. It consists of a chain of n + ..

1 mass points P,,(i = 0,. ,lr)
< <
and of s springs, n s n(lr 1)/2.+
<
The mass of P, is m,.(O< m, a).Any m, = a is a fixed point in
inertial space. We require that at least one of the mass points have infinitely
large mass, e.g. m, = 00.
The term “spring” is used to describe a massless one-dimensional device
that changes its length under the action of a force. In physical terms, a
spring is assumed to shorten under a compressive force of given magnitude
 NONLINEAR VIBRATIONS 159

precisely by the same amount as it lengthens under a tensile force of the

same magnitude, and it is capable of storing, but not of creating or destroying
energy. These conditions are fulfilled when
(2.1) o i j( w ) = oij(lwl),
and when

where O,(w) is the negative of the potential energy V, stored in the spring
Sij terminating on the ith and jth mass points, w is the length change of
Si, and Fii is the force which has produced the length change. Evidently,
combining (2.1) and (2.2), one has
(2.3) F-(m)
$1 = - F;i(- 20).

Each mass point is connected to one, several, or all other mass points
+
by a spring; hence, the maximum number of springs is n(n 1)/2. If one
extremity of a spring terminates on an infinitely large mass point it is called
an anchor spring; all others are coq5ling springs. I t is not essential that all
+
n(n 1)/2 springs be present in the system. However, it is required that,
in the absence of additional constraints, no mass point can be displaced
without giving rise to an elastic force acting on every mass point. For in-
stance if
(2.4) u*=o, ( i = O ,... I - l ) ,
)

the rth mass would be decoupled from all its neighbors on the left, and the
+
system of n 1 mass points would be separated into two spring-mass
systems with no elastic coupling between them. Such a separation into
several mutually uncoupled systems is excluded. In consequence the mini-
mum number of springs is n.
Each mass point has a single degree of translational freedom ui in the
direction of the chain. In view of mo = 00, one has uo E 0; hence, the
system cannot accelerate in the rigid body mode.
Each mass point may be acted upon by a time-dependent force, called
an exciting force and denoted by fi(t). Exciting forces act in the direction
of the chain of mass points. It is assumed that the spring forces Fii and the
exciting forces f, are the only forces acting on the mass points.
2. Admissible Systems
In order to fit into the framework of vibrations, linear or nonlinear,
systems must have certain properties whose physical description is :
(i) in the absence of exciting forces, the system must possess a single
equilibrium configuration, and it can execute “free vibrations” about that
equilibrium configuration ;
160 R. M. ROSENBERG

(ii) in the presence of exciting forces, the system can execute so-called
“steadystate forced vibrations.”
Some of these conditions can be insured by endowing the potential
function of the entire system with certain properties. This potential
function is, evidently,
n-1 I(

where uj - ui is the length change in the spring Sij.

We shall assume that, on any domain where it exists, 0 is smooth and
its first partial derivatives with respect to the ui are, a t least, piecewise
smooth. The second partial derivatives of 0 are thus assumed to exist
everywhere except possibly at isolated points of discontinuity of the a0/aui.
These assumptions are consistent with the physical properties of springs
in that they provide for the possibility of spring forces “with corners”;
yet they assure the existence of unique solutions under specified initial
conditions.
In addition, we require that the potential fulzction 0 be negative definite, or

O(0,.. .,O) = 0,

O(u,,. . . ,un)< 0 when the ul,. . , ,u, do not vanish simultaneously,

and the first partial derivatives of 0 vanish at the origin only, or

a .
-O(0,. .,O) = . .. = -a O(0,. . *,O) = 0.
8% aun

Finally, as a consequence of (2.1) and (2.6), 0 is symmetric with respect to

the origin of the zc,, ..
.,un-s#ace, or

Equations (2.6) and (2.7) insure that there exists a unique equilibrium
configuration, and the origin of the ui has been so chosen that every ui
vanishes in the equilibrium configuration. Equation (2.8) is a direct con-
sequence of the symmetry properties of the individual springs, described
in (2.3).
A potential function 0 which satisfies all of the above properties will
be called admissible. Moreover, when 0 is admissible, the azctonomous system
i s admissible.
 NONLINEAR VIBRATIONS 161

Exciting forces yi(t) are required to be a t least piecewise continuous,

and bounded for all t . In addition, they must satisfy

(2.10)

where T is a constant. Equation (2.9) says that the exciting forces are
periodic and, if there is more than one such force, they all have the same
period T.
Exciting forces /<(t)which satisfy the above conditions are called ad-
missible. Moreover, when both, 0 and the f,(t) are admissible, the nonauto-
nomous system is said to be admissible.
Among admissible systems, autonomous or not, we distinguish between
those that are linearizable, nonlinearizable, and weakly, or strongly non-
linear. In our system, these properties are associated with the spring forces;
hence, it may be best to define them in terms of these forces; i.e., in terms
of the partial derivatives of 0.
Because of the admissibility conditions on 0, these derivatives may
be written in the form
,. n
(2.11)

where the functions Pi vanish at the origin and are of degree higher than 1
and, because of (2.8),
(2.12)
-
aij = aii.

Let the determinant of the cZii be defined by

Then, a system is said to be linearizable if d # 0, and nonlinearizable if

d = 0. A system is said to be weakly nonlinear if

(2.14) Pi(u1,.. .,u,,) = E&u~,. . .,u,,), (i = 1,. . .,n)

where every leil << Id1 where d is defined in (2.13) and every Qi is bounded.
If (2.13) is not satisfied, the system is said to be strongly nonlinear.
162 R. M. ROSENBERG

3. Admissible Motions
From the definition of admissible systems, it is evident that these may
be strongly nonlinear, or even nonlinearizable. In consequence, it is not
likely that one can find the general solutions of the equations of motion or,
the class of all motions of which the system is capable. However, it is
possible to deduce certain types of motion and, in particular, the motions
which are the most important from the point of view of technical application.
These are the motions which, in linear systems, are called the “natural free
vibratiqns in normal modes” if the system is autommo%s, and the “steady-
state forced vibrations” if the system is nmutonomous. The latter include
the phenomena of “resonance” and of the tuned dynamic vibration absorber.
Both, the free vibrations in normal modes, and the steady-state forced
vibrations are so-called “vibrations-in-unison”. An admissible system is
said to execute a vibration-in-unison if its motion satisfies all of the fol-
lowing properties [7] :
(i) There exists a T = corzstant such that
(2.15) ui(t) = z ~ i ( b + T), (i = 1,. . .,%).
In words, the motions of all masses are equiperiodic.
(ii) If t, is any instant of time, there exists a single to in

tr < t < tr + TI2

such that
(2.16) u&) = 0, (i = 1,. . .,rz).
In words, during any time interval of a half period, the system passes
precisely once through its equilibrium configuration.
(iii) There exists a single 4 # to in
t, < t < 4 + T/2
such that
(2.17) ti&) = 0.

In words, during any time interval of a half period, the velocities of all
masses vanish precisely once or, all masses attain their maximum displace-
ment (in absolute value) precisely once during any interval of a half
period.
(iv) Let r (fixed) be any one of the i = 1,. . . ,n. Then every %&) and
ur(t) may be written, for a.ll t, in the form
(2.18)
 NONLINEAR VIBRATIONS 163

in which t is a parameter. It is now required that every ~~(21,) be single-

valued. In other words, the displacement of any one mass at any instant
of time determines uniquely that of every other mass a t the same instant
of time.
These are well-known properties of normal-mode vibrations, and of
steady-state forced vibrations, of the linear system. We shall show that
these same types of motion also exist in admissible nonlinear systems. Any
motion of an admissible system which satisfies (2.15) to (2.18) is said to
be an admissible motion.
Among admissible motions, it is necessary to distinguish between those
which are “similar” and those which are not. An admissible motion is said
to be similar [8] if
(2.19) =
uj(t)/zli(t) cii = const.

This terminology corresponds to common usage because, when (2.19) is

satisfied by an admissible motion, the wave-shapes of the time-displacement
curves are all geometrically similar. When (2.19) is not satisfied by an
admissible motion, that motion is said to be nonsimilar.
The normal-mode motions of linear systems, and their steady-state
forced vibrations under simple harmonic excitation are of the form
(2.20) aci=Qicosol. (i= l,...,n)
where the Qi are constants. Hence, these motions satisfy (2.19); i.e., they
are similar. We shall show that the linear system is not the only one having
similar vibrations-in-unison. In fact, there exist strongly nonlinear, and
nonlinearizable systems whose vibrations-in-unison are also similar.

111. THE TRAJECTORIES

IN CONFIGURATIONSPACE

1. Transformation and Trajectories

The equations of motion of the physical system are

a
(34
.
mjii, = -0(z+, . .,us) f#),
a@, + (i = 1,. , . ,n)

where dots denote differentiation with respect to time. It is now convenient

to introduce the transformations
(3-2) xi = F ui. (i = I , . ...n).
Then the system (3.1) transforms into the system

(3-3) H: Ei = U,(x,,. ..,zs)+ fj(t), (i = 1,. ..

$8)
164 R. M. ROSENBERG

where

and the subscript xi denotes partial differentiation with respect to xi.

But (3.3) may be regarded as the equations of motion of a mass point
having unit mass that moves in the n-dimensional space of Cartesian co-
ordinates xi,(i = 1,. . .,n), called the configuration space. This unit mass
point is being accelerated by two forces: One is the conservative force whose
components in the xi-directions are the UZi,and the other is a time-dependent
force whose components in the xi-directions are the fi(t). The unit mass
point and the forces acting on it constitute what may be called the pseudo-
system [9].
When moving in the configuration space, the unit mass point traces out
a trajectory. Here, we derive the equations of this trajectory. The solutions
of (3.3) are of the form

(3.4) xi = xi@), (i = 1,. . ,,n).

If one eliminates the parameter t between them, one obtains the equation
of the trajectory in the form

(3.6) xi = x i ( x l ) , (i= 2,. . .,n)

where x, = x1 has been arbitrarily chosen as the independent variable in
accordance with (2.18). One finds from (3.6) by direct differentiation
(3.6) i j= X i t i . , , Xj = XilZl + Xj)’il8
where primes denote differentiation with respect to xl. Moreover, if one
regards the velocities as functions of the displacement, the equations of
motion H become

Now, in view of (2.7),every xi takes on its extreme value Xi at t = t,. More-

over, in view of the continuity and differentiability of strictly* Newtonian

* A system is said to be “strictly Newtonian” if the forces, and hence the

accelerations, are bounded for every t. Thus, infinite forces of bounded impulse
and discontinuities in velocities do not arise in them.
 NONLINEAR VIBRATIONS 166

systems, there aIways exists some domain [ x j , X j ] ,(i = 1,. . .,n),of the
configuration space in which every xi@) has an inverse

(3.8) t = ti( q) I

Hence, summing the equations (3.7) over N, and integrating them once over
the domain [xi,Xi] one finds
"
(3.9)
1 "
-2
2 i-1
=
i<* . .JJ + 12 + i=
u(x~, 21Fi(xjJX<)
where h is a constant of integration. and

(3.10)
i
F ~ ( x < , X=~ )fi(tj(xi))dxi.
xi
It is evident that (3.9) is an equation of the energy balance of the system,
and the quantities Fi(xi,XI),(i = 1,. . .,rc) represent the work done on the
unit mass p i n t by the force components fi(t) in a displacement [xi,Xi].
When these forces vanish identically, (3.9) reduces to the energy integral
of the autonomous system. The idea of utilizing the inversions (3.8) is also
employed in Rauscher's method [3].
The substitution of the first of (3.6) in (3.9) results in an equation in
xla which may be explicitly solved for that quantity. That equation together
with the system H and the second of (3.6) is sufficient to eliminate the
parameter t. Hence, the resulting equation is that of the trajectory of the
unit mass point. It turns out to be
r 1

(3.11)
where we have omitted the self-evident subscripts on the 6. The corresponding
equation for the autonomous system is

(3.12)
The equation (3.12), but restricted to two degrees of freedom, was
probably first given by Kauderer [lo]. An equation, equivalent to (3.12)
has been given earlier [ll];in it, the arc length s was used as a parameter.
166 R. M. ROSENBERG

The elimination of s which has resulted in the compact forms (3.11) and
(3.12) was first given by Kinney [12].
The systems of equations (3.11) and (3.12) are denoted as the systems M
because they are derivable from a principle akin to the principle of Mau-
pertuis, given below. Similarly, the notation H in (3.3) may recall Hamilton's
principle.

2. The Restricted Principle of Least Action

It can be readily shown that the system M of the autonomous system;

i.e., (3.12) constitutes the Euler-Lagrange equations of the variational
principle
(3.13) 6A=O
where the action

(3.14)
'I

and

(3.16)

is the square of the line element in the configuration space. Hence, equa-
tions (3.12) follow directly from the principle of Maupertuis.
Now, it is well known that the principle of Maupertuis is only applicable
to scleronomous conservative systems. However, if one defines an action

$1

one can show that (3.11) is immediately obtained from the variational
principle
(3.17) SA=O
and from (3.10). Thus one has the [13].
Restricted Princi~lcof Least Action: The natural motion of any system
.
with equations of motion (3.3), 0% a%y interval [xi,Xi], (i = 1,. .#n)in which
the imersions (3.8) exist, i s such that the action A, defined in (3.16), i s
a mircimzlm.
 NONLINEAR VIBRATIONS 167

While it has only been shown here that the action is stationary, the
usual proof that the stationary value is minimum also succeeds here.
The above principle is restricted to the domain [xi,Xi] where the in-
versions (3.8) exist. In problems of dynamics in general, this domain is
not known u priori. However, because of (2.16), (2.17) and (2.18), that
domain is, for admissible motions, sufficiently large to contain at least the
closed intervals lying between the maximum amplitudes and the equilibrium
position, and the corresponding time interval is sufficiently large to cover
at least one quarter period. Since the systems M are used directly in the
methods to be presented, the restricted principle of least action is of con-
siderable usefulness in problems where admissible motions of admissible
systems are examined.

IV. GENERALDESCRIPTION METHOD

OF THE GEOMETRICAL

The problem in hand has now found a dynamical representation by means

of the system H of equations of motion (3.3), and a geometrical one by
means of the system M of equations of the trajectories (3.11) and (3.12).
It was shown by Darboux [la], for instance, that the correspondence between
these for corresponding initial conditions is one-to-one. The essentials of
the methods to be used here consist in utilizing the system M to deduce the
trajectories and, subsequently, to utilize these trajectories in order to deduce
the behavior of the system in the the-displacement domain. This is called
“the geometrical method”, a term borrowed from Synge’s paper [2].
It is interesting to note that the equations M of the trajectory are
intrinsicalIy nonlinear, even when the springs are linear. In fact, there
exist no admissible systems, autonomous or not, for which the equations of
the trajectory become linear. The assumption of dimear springs implies
merely that all Uxi are of the form
U

(44 u, = 2
i- 1
uijxj, (a = 1,. . ..n)
where the aii are either constants or functions oft. Under it, equations (3.3)be-
come linear while (3.11) and (3.12) remain nonlinear. The intrinsic nonlinearity
of the system M will perhaps be better appreciated if on recalls that its
integrals are Lissajous curves. These may have many points of self-crossing
and may even be compact on some domain of the configuration space. One
may conclude that admissible systems with linear springs are more easily
examined by means of the system H of equations of motion than by M.
The fortunate circumstance that the superposition theorem holds in this
case can, then, always be utilized to find the general motions and, if one
wants to work very hard, their trajectories in the configuration space [XI.
168 R. M. ROSENBERG

While linearity of the physical system has no essential simplifying ef-

fects on the system M , nonlinearity has no fearful, complicating effects
on MI either. Hence, if these equations can be used to deduce solutions
of the physicaZZy linear problem, they can frequently be used with equal
ease to deduce those of physically nonlinear problems [MI.
Now, it is quite possible that a system of very difficult, nonlinear dif-
ferential equations have certain simple integrals. In fact, we shall demon-
strate that the trajectories of admissible motions are such simple integrals.
They are of the form
(44 xi = x&J, (i = 2,. . .,%), (9 = 1,2,. . .)
where the index p denotes the mode of vibration. Say, the trajectory
corresponding to the mth mode of admissible motion is given by the functions
(4.3) xi = xim(X1), (i = 2,. . .$2)
that satisfy the system M, and which are supposedly known. Then, their
substitution in the first equation of the system H,for instance, results in

(4.4)
.. = UX,(Xl*%n(&
x1 * * * %&l)) + fl(&

an equation in x1 only.
If the system is autonomous, fl(t) 3 0, and (4.4) is of the form
(4.6) 21 = g(x1).
Then, for initial conditions (for instance)
(44 x1(0) = x,, ZJO) = 0 ,
and with the definition

(4.7)

the integral of (4.4) becomes

and the period of the motion is

0

(4.9) T m = 4 112 [G(v,XJ - G ( X ~ , X ~ ) ] } - W V .

XI
 NONLINEAR VIBRATIONS 169

If the system is nonautonomous, and the amplitude of the forcing function

is small; i.e.,
(4.10) fl(4 = &!At), (El small,
the steady-state forced vibration problem can be shown to become a pertur-
bation problem on that of free vibrations in admissible motion. It is evident,
then, that the central task in either problem (autonomous, or forced)
consists in determining the trajectories of admissible motions of autonomous,
admissible systems.

V. TRAJECTORIES MOTION
OF ADMISSIBLE

1. General Properties

We shall assume that, after formal reduction to a first-order system,

the right-hand side of (3.3) satisfy a Lipschitz condition. Trajectories then
exist and all are continuous. Because of (2.16), every admissible trajectory
passes through the origin of the configuration space. In view of (2.17), the
equation (3.9) reduces at t = t, to

This equation defines a surface in configuration space called the L-surface,

or the bounding surface. Then, it follows from (2.16) that every admissible
trajectory intercepts the L-surface. The equation (2.18) states that the xj(x,)
are single-valued for every i and any r. Hence, for instance, x k ( x I )is single-
valued, and the same is true for its inverse x I ( x k ) . But this requires that
each be monotonic. We denote this property by saying that every admissible
trajectory is strictly monotonic. Finally, since trajectories satisfy the Euler-
Lagrange equations of (3.17), existence of real trajectories requires that
the quantity under the square root of (3.16) be non-negative. But, i t is
seen from (5.1), that this quantity vanishes on the L-surface. Then, it
follows from the properties of U and the F , that no admissible trajectory
can pierce the L-surface. These properties are summarized in
Theorem V-I Every continuous, strictly monotonic curve, defined by
solzltions of (3.11) or (3.12), which passes through the origin of the configuration
space and which terminates on the L-surface (5.1) is that of a vibration-in-
unison as defined in (2.15) to (2.18).
From the preceding remarks it follows that the unit mass point of the
pseudo-system oscillates back and forth along this trajectory, and the
period of this oscillation is the same as that of the vibration-in-unison of
the physical system.
170 R. M. ROSENBERG

If the vibrations-in-unison are similar as defined in (2.19), the correspon-

ding trajectories are defined by
(6.2) xi(xl) = c i l x l r (i = 2,. . .,n)
where the cil are constants. In that case, (3.11) reduces to

(5.3) xj' +
[ux, f i ( t ( 4 ) I = uxj + f j ( t ( x j ) ) s (I' = 2,. - 1%)

because

But, (6.3) may be recast in the form

It follows that the differential equations of similar vibrations-in-unison

satisfy the system (6.3) or (6.4).
All properties of admissible trajectories, given above, hold whether the
system .is autonomous or not. Additional properties of admissible trajec-
tones can be deduced when the system is autonomous.

2. Admissible Trajectories of Azltonomozcs Systems

The equations of autonomous systems are obtained from those above

by putting
(6.6) f i ( t ) = Fi(Zi,Xi) 0, (i = 1,. . .,?z),
In that case, the energy integral

(6.6) T(R1,.. .'Xu) - U(X1,...,xu) = h

is seen from (3.9) to exist where

T(21,.. .,A?,,)= - 2
1
Xi'
2 i=l

is the kinetic energy of the pseudo-system, and h 0 is the energy level

of a given motion. The system M of the autonomous case is given in (3.12),
and the corresponding action integral is (3.14).
For this case, the bounding surface is defined by

(6.8) U(Xl,. . .,X") + h = 0,

 NONLINEAR VIBRATIONS 171

and this is also the maximum equipotential surface because (6.8) coincides
with (6.6) when the kinetic energy vanishes. In order to call attention to
the fact that (5.8) defines the bounding surface when the system is auto-
nomous, it will be denoted as the La-surface.
From (2.5), (2.6), (2.7), and (5.8) it follows that:
(i) the La-surface is a closed, smooth surface surrounding the origin
of the configuration space, and star-shaped with respect to it [9].
From (2.8) one sees that
(ii) the La-surface is symmetric with respect to the origin of the configura-
tion space.
From the admissibility condition (2.6) and (6.8) it is evident that
+
(iii) the quantity U h > 0 a t the origin of the configuration space.
This quantity does not change sign in the open, finite domain D, bounded
by the La-surface, it vanishes on L,, and it is negative (at least near the
La-surface) in the infinite open domain bounded by the La-surface.
These properties lead immediately to
Theorem V-I1 : Every trajectory, admissible or not, defined by solutions
x I . ( x I ) ,(i = 2 , . . .,n) of (3.12), lies i n the closed, finite domain D, of the con-
figuration space, bounded by the La-surface defined i n (5.8).
It is not necessary that all trajectories actually attain the La-surJace.
However, if they do, this occurs necessarily at an instant t, when all a,(t,)
vanish. Trajectories which attain the L,-surface will be called trajectories
of motion with a rest point [19]. Not all motions with a rest point need be
admissible because the corresponding trajectories need not pass through
the origin. However, all admissible motions are motions with a rest point.
Next we state a property, first observed by Mawhin [6]. I t is
Theorem V-I11 : If 8 i s any trajectory, admissible or not, of an admissible,
autonomous system, the trajectory 8 ,symmetric to 8 with respect to the origiB,
is also a trajectory.
The proof follows immediately from (2.8) and its consequence

Of importance to admissible trajectories is [S] the

Corollary V-IV: Every trajectory, admissible or not, (of an admissible,
autonomous system) which passes through the origin is symmetric with respect
to it.
172 R. M. ROSENBERG

Finally, we have

Theorem V-V: If a trajectory, admissible or not, of a n admissible, auto-

it intercepts thut surface orthogonally.
nomous system attains the L,,-surface.
To prove it we observe that the system M of autonomous systems, given in
(3.12), reduces on the La-surface to

(6.10) xjrUx,= U x j , ( j = 2 , . . .,n),

and this may be cast in the form

(6.11)

But equations (6.11) are precisely the conditions for orthogonal intersec-
tion.
Special additional properties hold for similar vibrations-in-unison of
autonomous systems. They are given in [9].

Theorem V-VI : Every trajectory of a similar uibration-in-unison of a n

admissible autonomous system intersects every equipotential surface ortho-
gonally, and [9]

Theorem V-VII: If there exists a straight line which intersects every equi-
potential szcrface of an admissible, autonomous system orthogonally, it i s a
trajectory.

To prove these theorems, we observe that equipotential surfaces are

defined by

(6.12) . +
U(X1,. .,xn) h* = 0. (O< h* < h)
and, in view of the properties of admissible potential functions, the equi-
potential surfaces are simple, closed, non-intersecting surfaces surrounding
the origin, and compact on D,.
It follows from (6.2) that, in the case of similar motion,

(6.13) xl’(xl) E 0, (i= 2,. , . ,n).

Hence, the system (3.12) reduces throughout D, to (6.10) or (6.11). Now,
+
the total differential of every equipotential surface d(U h*) is independent
of h* which proves the first of the theorems. It is clear that the second theo-
rem also holds because (6.11) is merely one way of writing the system M
when equations (5.13) are satisfied.
 NONLINEAR VIBRATIONS 173

VI. NORMAL-MODE
VIBRATIONS SYSTEMS
OF NONLINEAR

1 . A Neu Definition of Normal Modes

It was observed in connection with the definition of admissible motions
that the normal mode vibrations of linear, autonomous systems are in fact
vibrations-in-unison. They are usually deduced from the dynamical system
by means of eigenvalue theory, and in the proof of their existence the fact
is utilized that the potential function U is a quadratic form [17]. Now,
it is the very essence of nonlinearity (as defined here) that the potential
function is not a quadratic form : hence, the application of conventional
methods for deducing normal mode vibrations is difficult to generalize to
nonlinear systems.
We shall show that the normal mode vibrations of linear systems may
be defined in terms of admissible trajectories, and that these are deducible
from the system M ,rather than H. In fact, the normal mode vibrations of
autonomous systems are its only vibrations-in-unison.
Furthermore, it will be demonstrated that nonlinear admissible systems
also possess such vibrations-in-unison. The existence proof is given in the
appendix. These observations permit a new definition of normal mode
vibrations which is wide enough to include the well-known normal-mode
vibrations of linear systems as a special case.

Defhition I ; Vibrations-in-unison of an admissible, aertolsomous system

are the normal-mode vibrations of that system.

We shall also use

Defirtition I1 : Trajectories in the configuration space corresponding to

normal-mode vibrations of admissible, autonomous systems are called modal
lines.
Then, one sees from Theorem V-I that modal lines are strictly monotonic
curves passing through the origin, and terminating on the La-surface.

2. Straight Modal Lines

In (2.20), it was noted that the vibrations-in-unison of the linear. auto-

nomous system are similar or, that the modal lines of that system are
straight. Since we wish to deduce these vibrations by the geometrical
method, we derive first the conditions, necessary and sufficient for the
existence of straight modal lines, and we demonstrate subsequently that
the linear, autonomous, admissible system satisfies these conditions. The
derivation of the n.a.s. conditions is conveniently done by transforming
174 R. M. ROSENBERG

from Cartesian to spherical coordinates in n-space. These transformations

are of the form [ll]

(6.1) xi = rf& .. ..
(i = 1,. ,n)
and, specifically, they are [9]

(6.2) xi = Y sin en+l--i n

n--i

j- 1
cos ej, e,, = n/2, (i= 1,. ..,n).

When the equation of the L,surface is transformed into spherical coordinates,

it is of the form
(6.3) O(r,el, . . + h =0
Then, one has [9] the
Theorem VI-I : The conditions, necessary and szcfficient for the existence
of at least one mode of similar normal-mode vibration of an admissible, azcto-
nomow system are that every partial derivative of 0 with respect to the Oi be
ofthe fovm

(6.4) 0, = si,(r,el,.. .,en-,) - @,,(el,.. . (i = 1,. . .,n - 1),

and that the eqwations
(6.6) 8ie(81,. .. = 0, (i = 1,. ..,n - 1)
have at least one system of real roots &*.
To sketch the proof, we observe that, in spherical coordinates, the equi-
potential surfaces are given by

(6.6) O(r,e,,. ..,e,,-l) + h* = 0, o< h* < h.

Now,radius vectors r to points on equipotential surfaces having stationary
distances from the origin with respect to neighboring points on these same
surfaces satisfy
(6-7) dr = 0,
and these radius vectors intersect the equipotential surfaces orthogonally.
From (6.3)

or in view of (6.7)
(6.9) 0, = 0, (i= 1,. . .,n - 1).
 NONLINEAR VIBRATIONS 176

The locus of these points of stationary distance on the equipotential surfaces

is a straight line if, and only if, the system (6.9) has roots independent of
Y which requires that the Oei be of the form (6.4).
Theorem VI-I is useful not only because it may be used readily to test
whether straight modallinescanexist. but the direction-cosines of the straight
modal lines, if they exist, are furnished by the roots of (6.6). Hence, the
theorem provides also a method for computing the straight modal lines.
It can be shown that the new Definition I of normal mode vibrations
yields only the normal modes, and all of the normal modes of admissible,
autonomous, linear systems [S].

3. Interpretation of Modal Lines [ll]

The modal lines have a very simple, geometrical significance which will
be illustrated here by means of an admissible, autonomous system having
the three degrees of freedom
(6.10) XI = x, xg = y , xg=z.

i
Y

X
- y=ct ; x

FIG. 2. Modal line of linear system with three degrees of freedom.

Suppose the system is linear. Then, the equipotential surfaces are given
by the concentric three-dimensional ellipsoids (with center at the origin)
(6.11) t(x2 + y2 + %a! = h*, o< h* < h,
and the vibration in the pth normal mode is given by
(6.12) x = Xp cos apt, JJ= Yp cos opt, z = 2, cos apt
where X,, Y,,Z p are constants.
It follows from these equations that the pth normal mode satisfies

(6.13) ylx E YpIXp = c$, alx =Z,lX, = c?#.

176 R. M. ROSENBERG

Then, the modal line in the pth mode is the intersection of the planes

(6.14) P
y = cy+x, z = l$X.

Evidently, this intersection is a straight line passing through the origin.

From the known geometric properties of ellipsoids and from Theorem V-VII
it follows that there exist three such modal lines, and they are the principal

(a1 (b)
FIG. 3. Modal line of nonlinear system with three degrees of freedom.

axes of the ellipsoid. The construction of a modal line, and its location
inside the bounding ellipsoid is shown in Figures 2(a) and (b).
If the problem is nonlinear, but the normal mode vibrations are similar,
equations (6.14) still apply, but, in that case, the bounding ovaloid is no
longer an ellipsoid.
If the problem is nonlinear] and the normal-mode vibrations are non-
similar, equations (6.14) are replaced by equations of the form

(6.16) y = qJ(X), = $(4

where y and $ are nonlinear, monotonic functions of x . Each is a cylinder

in xyz-space containing the origin, as shown in Figure 3(a). Their intersection
is a modal line, and this modal line inside the bounding ovaloid is shown
in Figure 3(b).

VII. PROPERTIES
OF MOTIONS WITH A REST POINT[19, 241

1. The Transversals (or P-cwves)

Motions with a rest point have been defined as those in which there
exists a time t, at which the velocity of all mass points of the physical system
vanishes. If the system is autonomous, the trajectories corresponding to
motions with a rest point are those which attain the Lasurface, and ad-
 NONLINEAR VIBRATIONS 177

missible motions belong to this class. The equipotential surfaces of these

systems are given by (5.12), and the transversals are defined in terms of
these equipotential surfaces. Transversals, or P-curves, are the lines ortho-
gonal to the equipotential surfaces. From their definition, one has imme-
diately
..
Theorem VII-I : At a point (xl,. ,xn)of the configuration space where
the slope of a transversal is well-defined, it coincides with the direction of the
force which acts, at that point, on the w i t mass point of any admissible, auto-
mmous pseudo-system.
Hence, the P-curves have an important dynamical significance, and i t
is of interest to determine their properties. One observes that, when n > 1.
the transversals constitute a family of curves which pass through the origin
of the n-space, which are compact on D,, and which are nonintersecting,
except a t the origin. It would be interesting to determine their geometrical
properties in the entire domain 0,.However, this is a difficult task when
the number of degrees of freedom exceeds two; therefore, our attention will
be restricted to the case n = 2.
Consider an admissible, autonomous system having the degrees of freedom
(7.1) u,= u, u2= v .

We admit linearizable and certain nonlinearizable systems by writing

a
-O(U,V) = - auk - 6 V k + P(z4.V).
au
(7.2)
-o
a ( ~=, -~ ) tUk - avk + Q(.,.)
av
where k is a positive, odd integer, and P and Q are of degree higher than k.
The determinant

(7.3)

With these definitions, the system is linearizable if k = 1, and nonlinear-

izable if k > 1.
The equations of motion of the physical system are

.. -
m2v =
a O(u,v),
av
178 R. M. ROSENBERG

those of the pseudo-system are

(74

and the equation of the trajectories is

(7.6) 2 [u(X,y)+ h]Y" + (1 f y'*) Ly'UJW') - Uy(%y)I= 0

where priqes denote differentiation with respect to x.
The system possesses the energy integral

(7.7) 7-(2,$)- V ( x , y )= h.

The bounding surface (here a curve) is the locus of the points P(X,Y) for
which the kinetic energy vanishes when the energy level of the motion is k,
and the equation of the La-curve is

(7.8) + h = 0.
U(X,Y)
The equipotential curves are the functions

(7.9) Y = &)
which satisfy the equation
(7.10) +
U(x,qJ) k* = 0, OC h' <h
and they will be called E-crcrves. By differentiating (7.10), one finds that
E-curves satisfy the differential equation

(7.11)

It follows that the transversals, or P-curves, are the functions

(7.12) Y =e(4
which satisfy the differential equation

(7.13)

We observe that (7.13) is singular only at the origin, and the equation
of the trajectories (7.6) is singular only on the La-curve. Consequently, we
define as regular p o d s of the xy-plane all points, except the origin, which
lie in the open, finite domain D, bounded by the La-curve.
 NONLINEAR VIBRATIONS 179

To begin with, we prove

Theorem VII-I1 : The origin i s always a node for the P-curves of admissible,
autonomous systems having two degrees of freedom.

Suppose the system is linearizable [19]. or k = 1 in (7.2). Then, in the

neighborhood of the ongin,

(7.14)
u, = XU,(O,O) + y ~ x y ( 0 , O+) . -
* I

u, = X V , y ( O , O ) + yU,,(O,O) + . . .*

The substitution of these in the equation (7.11) of the E-curves results in

(7.16) *=
dx
- xu, - 'pu,
XU, + pU,, i-
+ . ..
. ..
where the partial derivatives are evaluated at the origin. But, the integrals
of (7,11), and hence (7.16), are known; they are E-curves and they form a
continuum of simple, closed, nonintersecting curves surrounding the origin.
Then, it follows that the origin is a center for the E-curves and this requires,
by the Poincart! theory of singular points [l], that

(7.16) u,,u, - uz,> 0.

The substitution of (7.14) in the equation (7.13) of the P-curves results
in

(7.17)

Application of the PoincarC theory to this equation has as immediate con-

sequences

(7.18)

These are two of the three conditions which are necessary for the origin to
be a node. But, in view of (7.16).

(7.19) U,Uee - Ute > 0.

This is the third necessary condition, and (7.18) and (7.19) are also sufficient.
Hence, Theorem VII-I1 is satisfied in the lilrearizable case.
180 R. M. ROSENBERG

To prove the theorem for nonlinearizable systems [21], we note that,

in the neighborhood of the origin, the potential function of the pseudo-
system is

(7.20)

where a,b,c > 0 are constants.

The derivatives of this function (with y replaced by 0) are now sub-
stituted in the equation of the P-curves (7.13), and the transformations

(7.21)

are introduced in the resulting equation. Then, one finds, instead of (7.17)
in the neighborhood of the origin

(7.22) _
dr] - - cvk + b(E - q ) k + . . .
at - - atk - b ( t - q)b + .. . *

The singularity at the origin of this equation may be discussed by the method
of Argemi and Sideriadks [22, 231. These authors consider the equation

(7.23)

where X and Y are homogeneous functions in x and y of the same degree

k. They introduce a transformation
(7.24) y =d ( x )
under which the functions go over into
(7.26) x = a+/(& Y = x*g(A)
and the differential equation (7.23) becomes

(7.26)

where k(A) = g(1) - it/(& Now, this last equation has simple singular
points on the it-axis at the zeros Ri of k(it), and their discussion is accessible
to the Poincarh theory. Expansion of the numerator and denominator
of the right-hand side of (7.26) near the singular points gives
 NONLINEAR VIBRATIONS 181

_ - h’(li)l + . . .
dl -
(7.27)
dx f(&)x + ,. ’
,

The authors show that only saddles and nodes can occur, and one has
a node if f(li)h’(&)> 0,
(7.28)
a saddle if f(A,)h’(l$)< 0.
The quantity t(&)h’(&)does not vanish, in general, because the zeros of
/ ( A ) do not coincide with those of h(1) when X # Y,and h’(1) does not
vanish at the zeros of h(1). Moreover, when the 1, are ordered, according

(a 1 (b)
FIG.4. Saddles and nodes in xl-plane and mapping on configuration space.

to their magnitudes (in ascending or descending order), nodes and saddles

occur alternately. Clearly, the inverse of the transformation (7.24) maps
all singular points of (7.27) into the origin of the xy-plane, and the zeros
li of h(1) are the slopes of the trajectories at the origin. This proves The-
orem VII-I1 for the case of nonlinearizable systems.
It may be interesting to follow the construction in greater detail, as
.
done in Figure 4. Let A,,&,, . . . give the nodes in the xl-plane, and &,,A,, . .
the saddles. Next, construct the straight lines that pass through the origin
.
with slopes &,(i = 1,2,. .) and assume that 4 > 0; i.e., the line having
this slope lies in the first and third quadrants. Next consider the trajectories
lying in the sector enclosed by the lines of slopes 1, and 3.8 (starting in the
182 R. M. ROSENBERG

first quadrant and proceeding in a counterclockwise direction). This sector

contains a continuum of trajectories that pass with slope 4 through the
origin, and they tend to lines parallel to the line of slope 4. The latter is
a degenerate trajectory of this class. The trajectories in the next sector
between the lines of slopes & a n d & pass through the origin with slope &,
and tend to lines parallel to the line of slope 4, that line being also a degener-
ate trajectory of this class. This process is continued until the xy-plane
is filled in the neighborhood of the origin with trajectories. IndFigure 4(a),
the trajectories and singular points of the xl-plane are shown, and in Figure
4(b), the corresponding picture in the xy-plane is shown.
Suppose, as a special example, that k = 3 in (7.20) and (7.22). Then,
[24] a simple computation shows that

(7.29)
/(A) = - b[(2 - 8) + (1 - 4 s 1 v
h(A) = - b[A4 - aAs + P A - 13
where
C
a=2--
b'
(7.30)

It is now necessary to determine first the zeros & of h(l) and, next, ,the
sign of the product fh' at these zeros. Writing h(A) = 0 in the form
(7.31) A4 - 1 = (als - 8)l
one sees that the left-hand side is even in A and has two real zeros at 1 = f 1,
and the right-hand side is odd and has one or three real zeros. Hence, these
two functions have always at least two real intersections, or h(A) has always
at least two real zeros. One may assume without loss of generality*
that c >, a. Now, it follows from the second of (7.29) and from the assump-
tions on a, b, and c, that
h ( w ) = h(- 00) = - 00,
h(0) = b,
(7.32)
h(1) = - b(- a + 8) = - (C - a ) < 0.
h(- 1) = - h(1) = c - a 2 0.

* If, on the contrary G < a, we simply denote the displacement of mII by u and
that of m1 by u which restores the assumed relation.
 NONLINEAR VIBRATIONS 183

Hence, h(A) looks as shown in Figure 6 (where no attention is to be paid to

the location of the maximum). It follows from (7.32)and Figure 6 that

hc. 6. The curve h(A).

one root, say 4, is positive and lies in the interval O < A,< 1. and & is
negative. Moreover, the slopes are

(7.33)

Next, we examine f(A). It follows from the first of (7.29) that

(7.34)

and that /(A) has only a single real zero a t A = 1 - v2 - B. To evaluate

the /(Ai), it is convenient to return to the physical constants, so that

Since b > 0, a/b > 0, and 0 < 4 < 1, and since 4 < 0, one has from (7.36)
f(4)< 01
(7.36)
f(U< 0.
Combining (7.33) and (7.36),

(7.37)
184 R. M. ROSENBERG

Hence, the xl-plane has, at least, two singular points on the A-axis. One
lies on the positive branch between 0 and 1, and it is a node. The other
lies on the negative branch, and it is a saddle.
We shall assume that Al,z are the only zergs* of h ( l ) . Under this assump-
tion we have now demonstrated
Theorem VII-11: When k = 1.3 and h ( l ) has only two zeros, all trans-
versals of admissible, autonomous systems pass through the origin of the xy-
plane with common slope OOp.
To determine the properties of P-curves not near the origin, we determine
the locus of their inflection points, if any. By differentiating (7.13), one has

This locus will be called the F-curve:

It follows from (7.39) that F-curves pass through, and are symmetric with
respect to, the origin.
Next we determine the locus of points such that the tangents to P-
curves at these points pass through the origin. Evidently, these are the
points for which

(7.40)

Hence, the locus of these points, called the G-curve, is given by

(7.41) G(x,Y)= Y U , - X U , = 0.
G-curves also pass through the origin and are symmetric with respect to it.
Obviously, F and G-curves have the slope 0; at the origin.
Now, transversals are seen from (7.13) to be themselves symmetric with
respect to the origin. Hence, the field of transversalslooks as shown in Figure 6.
In that diagram, P,(Xo,Yo) is the point where the straight line, passing
with slope 0; through the origin, intersects the La-curve. P1(X,,Y,) is that
where the C-curve intersect the La-curve, and P,(X,,Yp) is that where the
F-curve intersects the La-curve.
We show now, that a t least one F-curve and G-curve always exist and
intersect the La-curve. Since every equipotential curve has, in the first
quadrant for instance, always at least one point of stationary distance

It can happen that k(L) has four real roots when R = 3. This has been shown
to be an exceptional case and is not treated here. [MI.
 NONLINEAR VIBRATIONS 186

from the origin, that quadrant contains at least one G-curve. Consider
now the transversal issuing from P I . I t points initially towards the origin.
If Po and P I do not coincide, this particular transversal must have an
inflection point prior to its amval at the origin with slope 6;. In that
case, the transversal issuing from Po points initially above the origin, and

F I G . 6 . The directed transversals in nonlinear system with two degree of freedom.

it must also have a point of inflection before arriving with slope 6; at the
origin. Hence, in general, there exist transversals with points of inflection,
and there exists at least one F-curve in the first quadrant. The transversals
will be endowed with a sense of direction by considering the points P ( X , Y )
which compose the Ldcurve as their points of issue. These directed trans-
versals in Figure 6 are supplied with arrow heads pointing toward the origin.
In this way, the transversals give not only the direction, but also the sense,
of the forces acting on the unit mass point of the pseudo-system.
2. The T-curves [19]
The trajectories corresponding to motions (at energy level h) of admissible,
autonomous systems fall into two general classes: those which attain the
La-surface, and those which do not. By definition, T-curves are those integral
curves of (3.12) which do intercept the La-surface. Hence, the T-curves
constitute the class of all trajectories which correspond to motion with a
186 R. M. ROSENBERG

rest-point of admissible, autonomous systems. Note that T-curves need

not be admissible in the sense of Section V.
We shall establish here certain general properties of T-curves of systems
having two degrees of freedom. All T-curves intercept the L,-curve, and
all P-curves intersect it. The P-curve which issues from the same point
P(X,Y) from which a T-curve issues will be called “the transversal associated
with that T-curve,” or simply the associated P-came.
Some of the properties of T-curves are easily established by means of
the equation of the trajectories (7.6), written in the form

(7.42)

it results immediately from a substitution of (7.13) in (7.6).

Because of the definition of P-curves and T-curves and Theorem V-V,
one has:
Properly 1 : Every T-curve is tangent to its associated P-curve on the
L@-curve.
Less obvious is:
Property 2: The curvature K ~ ( X , Yof) a T-curve at P(X,Y) has the
same sign, but is less in magnitude than the curvature K ~ ( X , Yof) its asso-
ciated P-curve.
To prove it, one observes that the second derivative y“ in (7.6) is in-
determinate on the La-curve. Evaluating it by 1’Hospital’s rule

(7.43)
1 Uxy
y”(X,Y) = -
3 1
But, by Property 1 and (7.13),
+ y‘(Uyy -
ux
UXZ

1-
- Y’UZY)
x.*

(74 y y x , ~=
) e y x , y ) = u,(x,Y)/u,(x,Y).
Consequently,

and, comparing (7.38) and (7.45)

(7.46) KT(X,Y)= iKp(X,Y),
which proves Property 2.
Next, we have:
Property 3: A T-curve which is Not a straight line cannot coincide
everywhere with its associated P-curve.
The proof is evident from (7.42). From that equation, one can also
deduce immediately:
 NONLINEAR VIBRATIONS 187

Property 4: If any trajectory (and, hence, any T-curve) is tangent to

a P-curve at any regular point, the trajectory has zero curvature at that
point.
This property of trajectories was earlier noted by Kauderer [lo].
Property 6 : At any regular point, every curved trajectory presents
its convex side to the incoming, directed P-curves.
This property is a consequence of Theorem VII-I. It merely states that
the trajectory of the unit mass point is being turned, by the forces acting
on it, in such a direction as to yield to these forces.
Prq5erty 6: Every T-curve that intersects its associated P-curve at a
regular point Q(x,y), must have passed through a point of zero curvature
on the arc between P(X,Y) and the point of intersection Q(x,y).This prop-
erty follows from the observation that the P-curves constitute a regular
field, and the T-curves are smooth. Hence, there exists a point on the arc
lying between P(X,Y) and Q(x,y) where the T-curve is tangent to some
P-curve. By Property 4, this is a point of zero curvature of the T-curve.
Finally, we have :
Property 7: Every trajectory (and, hence, every T-curve) either passes
through the origin of the xy-plane or, else it must have at least one point
of tangency with at least one E-curve (see the appendix).
To prove it, we observe that the E-curves constitute a regular field of
simple, closed curves surrounding the origin. Moreover, trajectories are
smooth, and by Theorem V-11, they are confined to the closed domain 0..
It follows from these observations that Property 7 is true.
It is clear that one may consider the origin as the (degenerate) equi-
potential line of zero energy. In that case, Property 7 might be interpreted
as stating that every T-curve is tangent at least once to an E-curve, and
T-curves which are tangent to the E-curve of zero energy are necessarily
simple trajectories and, perhaps, modal lines.
It seems probable, that the above properties of T-curves could also be
established for systems whose number of degrees of freedom exceeds two.

VIII. SPECIALAUTONOMOUS
SYSTEMS

1. Smooth Systems
It is evident from the .equations of motion (3.3) of the pseudo-system
that all differences between given, autonomous systems consist in differences
between their potential functions.
An autonomous system is said to be smooth if its potential function is
of the form
188 R. M. ROSENBERG

where the a$) are constants. Then, the spring force of a spring Sij between
the masses mi and mi is

where

(8.3) w=u.- "j

is the length-change. Hence, all derivatives of every spring force with

respect to the length-change exist, which is the reason for the terminol-
ogy "smooth". While all spring forces of smooth systems are assumed to be
finite polynomials, it is not required that these polynomials be all of the
same degree. Thus, while the restriction to smooth systems may appear
severe from the mathematical point.of view, it is mild from the standpoint
of physics.

2. Uniform Systems [9]

A system is said to be uniform if its potential function is of the form [9]

One sees from (8.4) that uniform systems are those in which all springs are
equal, and all masses are equal (in which case one may, without loss of
generality, put the masses equal to unity).
For uniform systems, one can readily demonstrate an interesting and
useful property. Let us denote as the potential function of the associated,
linear system that whose potential function is [9]

where the values of a(l) are the same in (8.4) and (8.6). Then, one has
Theorem VIII-I : The modal lines of an autonomous, uniform system are
straight. Their directions coincide with those of the associated linear system.
One proof of this theorem is due to Mawhin [6] who shows that the modal
lines of (8.4) are straight, and that their direction is independent of m.
The importance of this theorem is evident. The modal lines, when U = 0,
are the principal axes of the n-dimensional ellipsoids

(8.6) O + h* = 0, (0< A* < h)

 NONLINEAR VIBRATIONS 189

and the directions of these axes are readily found by means of linear eigen-
value theory. The time-historyof the motions can, then, be found subsequently
by simple quadratures.
The geometrical interpretation of this result is that the equations

(8.7) U + h+ = 0, (0< h* < h)

are ovaloids which are “distorted ellipsoids.” However, these distortions
are such that the ovaloids have the same symmetries with respect to the
modal lines as the ellipsoids of (8.6).

3. Sequential Anchored Systems

With Haughton [27]. we call a system seqwntial if it consists of a chain
of n finite masses in which each mass is coupled by a “smooth” spring (as
above) to only its neighbors. It is said to be anchored if the first and nth
masses are also coupled to a fixed point by “smooth” springs. This is in
fact, the model that is usually employed in the theory of linear vibrations
of multi-degree-of-freedom systems. The potential function of anchored,
sequential systems is of the form

+
where i = i 1.
Haughton [27], using methods similar to those of Mawhin [S]proves the
following remarkable
Theorem VIII-I1 : If an admissible. anchored, seqrcential system has straight
modal lines, their directims are the same as those of the associated, linear
system of potential function

(8.9)
If, in an anchored, sequential system, all masses are equal and all springs
are equal it is said to be anchored, sequential and d i m . In that case,
one can readily show that straight modal lines do, indeed, exist; the prob-
lem of normal mode vibrations can, then, be completely solved by con-
sidering a linear system and by performing one additional quadrature.
190 R. M. ROSENBERG

4. Homogeneous Systems [28]

An admissible, autonomous system is said to be himogmcoscs if its po-

tential function is of the form
n u
(8.10)

where k is a real number in O C k < do. I t derives its name from the fact
that the potential function is homogeneous in the xi of degree k 1.+
It is interesting to forni the expression for the spring forces. These
forces are given by

This equation shows that the spring forces are odd functions, and proportional
to the kth power, of the length-changes. Clearly, when one puts k = 1,
one recovers the linear system.
Physically, neither masses nor springs need be equal in homogeneous
systems. Instead, the springs are nonlinear “in the same way”; i.e., all
obey the same simple power law.
It is not difficult to construct a physical system for which k = 3 [21].
Consider an arrangement of masses and springs as in Figure 1. However, in
the present case, the translations ui are not in the direction of the chain,
but t t o d to it. Moreover, every spring is assumed l i w and its free
length is exactly equal to the distance between the masses interconnected
by that spring when the system is in the equilibrium configuration. Clearly,
the spring forces are always restoring, or 0 is negative, definite; moreover,
the spring forces reverse sign when the displacements do, or 0 is symmetric
with respect to the origin. However, infinitesimal displacements I(~ which
are small of the first order produce length-changes in the springs which
are small of the second order. Hence, for sufficiently small displacements,
k = 3.
By applying the criterion (6.3) one can readily prove [S]
Theorem VIII-111: The modal lines of admissible, autolumtous, homo-
g m u s systems are straight. Their direction i s defined by the sysiems of roots
of the transcedental eqwations (6.4).

6. Symmetric Systems

With Mawhin [6] we call a system symmetric if its potential function is

of the form
 NONLINEAR VIBRATIONS 191

The physical properties of symmetric systems are the following: All masses
are equal (hence, assumed equal to unity, without loss of generality), all
spring forces are smooth (in the above sense), all anchor springs are equal,
and the spring forces in all coupling springs are polynomials of thesame degree.
This is a generalization of the “symmetric two-degree-of-freedom system”
1291 which was called “symmetric” because the system had literally physical
symmetry about its centerline. (It should be noted that Mawhin denotes
the spring between m, and m, as So, that between m, and mi as S j , and
that between m, and m,, as S,.)
If a symmetric system has only two degrees of freedom, one can readily
show [29] that straight modal lines exist, and they have the same inclination
as those of the associated linear system.

IX. THE ATEB-FUNCTIONS

[20]

1. Their Origin and Importance

As the name indicates, the ateb-functions are inversions of certain beta
functions, much in the same way as elliptic functions are inversions of
elliptic integrals. Similar to elliptic functions which satisfy the autonomous
Duffing equation, the ateb-functions solve the problem of normal-mode
vibrations sf homogeneous systems. In common with elliptic functions,
the ateb-functions depend on a parameter n as well as an argument. I t is
interesting to note that Gauss [30] was the first to invert a beta-
function for the case n = 2. His problem arose in the rectification of the
lemniscate, and his lemniscate functions were, in fact, the first functions
to be defined by the inversion of an integral. Later, Legendre showed [30]
that the period of the lemniscate functions was expressible in terms of
gamma functions-an observation that has been repeated independently
by many authors in recent times [31, 32, 10, 331.
We have shown that the modal lines of homogeneous systems are straight.
Hence, they are expressible in the form
(9.1) xi = ciPxl, (i = 2,. . .,n; p = 1 , . . .,m)
where m is the number of modal Lines. If one substitutes (9.1) into the
first of the equations of motion
192 R. M. ROSENBERG

of a homogeneous system, one finds

(9.3) 2, + C9x11x1~)-1 = 0, ( p = 1,. * .,m)

where

(9.4)

pj" = mil-.
We shall show that (9.3) has periodic solutions of period Tp,and it follows
from (9.1) that, then all x&), (i = 2,. , ,,n)are also periodic of the same
period. Hence, the resulting motion is a normal mode vibration.
We shall integrate (9.3) for the Cases I and 11 where, for

Case I and 11: %l =0 when x, = XI> 0,

(9.5) Case I: x,= 0 when t = 0,

Case 11: xl = X I when t-0.

The conditions for Case I1 are ilzitial conditions because x, and are speci-
fied at the same instant of time. However, in Case I the velocity and dis-
placement are prescribed at different instants of time. Hence, in Case I,
it must be shown a posteriori that, if x,(O) = 0, then there exists an %,(O)
such that %,(to) = X,, (X,,t, > 0).
First, we change (9.3) into the canonical form by means of the trans-
formations [20]
(9.6) z = (c&8)l/2Xl"-1t; x, = t X 1 ; 1( = (k + 1)/2
and, because of the definition of n and the bounds on k,

(9.7) n=l when k = l ; 1 / 2 < ~ < 00.

Under (9.6), the equation (9.3) goes over into

and equations (Q.6)become for

Case I and 11: c=O when E= 1,
(9.9) case I: €=O when v=O.
Case 11: €= 1 when r=0.
 NONLINEAR VIBRATIONS 193

A first integral of (9.8) satisfying the first of (9.9) is

(9.10) E' = f (1 - I[p)'le

and a second integral becomes, for Case I

c
(9.11) 7 = kj(1 - IYlen)-w,
0

and for Case I1

< <
with 0 [ 1. The sign ambiguities in (9.11) and (9.12) are readily re-
solved with the result that the +
sign must be chosen in (9.11) and the
- sign in (9.12) [W].
The question whether there exists a real to>0 such that E ( T ~ )= 1
is now easily answered by observing that (9.11) exists when E = 1. In
fact, as observed by many authors 131, 32, 10, 331 under the change of
variable
(9.13) r& --s

the integral (9.11) with upper limit 6=1 becomes

is the completebeta function B(P.q) of parametersp = 1/2%,

where B(1/(2%),+)
q = 4. Similarly,

(9.16)

where B&q) is the incomplete beta function+ of argument [* and par-

ameters p = 1 / 2 ~ q, =+.
t We use here Pearson's notation [84].
194 R. M. ROSENBERG

2. The Inversions
From the above results it follows that the solutions of (9.8) are, for
case I

(9.16) z=
1 ’
(- )
1
Bp 2n’T ’
1

and for Case I1

(9.17) t= &[B 1 1
-

with
(9.18) [* = €11”.
But, if € ( t is
) periodic, the solutions (9.16) and (9.17) cannot be single-
valued on an interval exceeding one-half period. Hence, it is desirable to
invert these solutions. The definitions used in the inversions, and the trans-
formations are summarized in the table below. Cumbersome sign distinction
are avoided if “n is regarded to behave like an odd integer.” This phrase
+
means that negative quantities, raised to the power fi p , where 9 is either
zero or an even integer, remain negative, and negative quantities, raised
+
to the power n q where q is an odd integer, become positive. The
absolute values of these powers are given by
(9.19) ,*+P = l,P+P

with a similar expression for powers n + q.

Case I Case I1
 NONLINEAR VIBRATIONS 196

0 < t1& < 1;

It is seen that the inversions lead to four ateb-functions: Those called
a m p n ~ are
~ , ~inversions of C19(~1,2)r
where GIm2are beta functions, and
those called Sam (nul) and cam (nuz) are respectively the inversions of
Flr(tl,z)which are also beta functions.
A certain number of properties of ateb-functions have been determined
[20] and are summarized below.

3. Thc Am9-Functions

The amp-functions are odd, or

(9.20) amp - n ~ amp

~= - , nu1.2.
~

When n = 1,
(9.21) amp %I+ = '%,a.
Particular values are given by

amp nu* = 4 2 ,
(9.22)
amp0 = 0,
where

(9.23)

They satisfy the relation (also satisfied by the elliptic am-function)

where fi is a positive odd integer.

Their first derivatives are
d
.- (amp mI)= N sam* - (nu1),
clrc,
(9.26)
d
-(amp nt4z)= n cam" - (n94).
dua
196 R. M. ROSENBERG

Special values of these derivatives are

du,
=1 when It-1, for u,=O
=oo when n < 1,)

for %=y+.

(9.26)

I
d
-(ampn%) = 0, for %= 0,
du,
d
-(ampn%)=O
dust
=1 for %=%+
=do when n t 1.
,
(9.27)

“amp”-functions look as shown in Figure 7(a) and @).

amp nul amp nu2

I I

FIG.7. The amp-functions.

 NONLINEAR VIBRATIONS 197

4 . The Sam- and Cam-Functions

The sam- and cam-functions are periodic of the same period; i.e.,

(9.28)
+
sam (ml)= sam (nu1 4nu*),

cam (nus)= cam (1221, + 4nu*),

where M* is defined in (9.23).
One is odd, the other even, or

sam (-a%) = - sam (n%),

(9.29)
cam (- nup) = cam (nHn).
When n = 1, they become circular functions, and when n = 2, they become
elliptic functions, or
sam (ul) = sin ul,

cam (113 = cos %,

sam (2%) = sn (2-'/*.y),

cam ( 2 ~ =
~ cn
) (2-1/2,~a).

If ulo and U s o are those values of u1 and ugfor which amp nulo = amp nuaO,

(9.31) +
Sam2*(nulo) cam& (nuso)= 1,
thus generalizing a well-known identity of trigonometric functions. The

I
derivatives of these functions are

d
-
dU,O
[sam (nulo)]= cam" (nuno),
(9.32)
d
K~[cam (nuno)]= - sam" (nulo),

I
-
d4
[sam (aul)]= - n sam%- 1 ( n y ) ,
(9.33)
da
7[cam (*US)] = - n cam*"-' (w).
The sam- and cam-functions look as shown in Figure 8 and Figure 9 for
a variety of values of n in 9 n < <00.
198 R. M. ROSENBERG

FIG. 8. The sam-function.

For n = 1, the curves are the trigonometric sine and cosine curve,
respectively. For n = 4, they are parabolas, and for B = 00, they consist
of segments of straight lines.

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 16 1.8 2.0

u2

FIG. B. The cam-function.

It is evident that the solutions of (9.8)are for Case I and 11, respectively,
.f = Sam (m),
(9.34)
.f = cam (n7),
as one can readily verify by direct substitutions, and by making use of (9.33).
 NONLINEAR VIBRATIONS 199

X. NONSIMILAR VIBRATIONS[8, 181

NORMAL-MODE

1. The Pertwrbation Potential

The trajectories of nonsimilar normal-mode vibrations of autonomous,
admissible systems are strictly homogeneous curves passing through the
origin of the configuration space and intercepting the La-surface, and they
are integral curves of the system M defined in (3.12).
At present. there exists no hope of finding the integrals of M in the
general case, considering the nonlinear character of the system (3.12).
However, as shown above, there exist many strongly nonlinear systems
whose modal lines are straight. Systems "neighboring on these" may have
curved modal lines; hence, their normal-mode vibrations may be nonsimilar..
Here, we shall show how such modal lines may be found by perturb-
ation methods. The results, reported here, are due to Kuo [MI.They are
generalizations of the results found earlier in the case of two degrees of
freedom [S]. Let
(10.1) Uo= Uo(xl, . . ., x , , ; q , . . ..m,,;al,. . .,at)
be the admissible potential function of an autonomous system having at
least one straight modal line, defined by
(10.2) xi* = cix,, (i = 2,.. ,,n).
In (lO,l), the mi, (i = 1,. . . , f i ) are the n masses and the u j , ( j = 1,...,2)
are parameters defining the properties of the springs.
We shall examine systems whose potential function neighbors on U,,;
hence, we call Uo the potential function of the parent system.
We shall say, "a system neighbors on the parent system" if its poten-
tial function is of the form
(10.3) 0 = u, + ewe, 1.1 << p o l ,
where 0 is admissible; consequently, wo must also satisfy the admissibility
conditions of potential functions. We call 0 the potential function of the
perturbed system, and w,, the perturbation potential. It is evident from
(10.1)that the mi and uf are the only parameters subject to perturbations.
Therefore, the most general form of the perturbation potential is

(10.4)

where the masses and spring parameters of the perturbed system are

(10.6)
*i = mi + &a& (i = 1,.. .,?$),
Gj = uj + && .
( j = 1,. . J ) .
While some of the ai and Pi may be zero, all are fixed, once chosen.
200 R. M. ROSENBERG

2. The Curved Modal Line

Kuo [8] demonstrates

Theorem X-I: If the parent system
(10.6) zi = avolaxi, ..
(i = 1,. ,n)

possesses, at some energy level h, a straight modal line, then the fierturbed system

(10.7) Yi = a 0 / a x , , (i = 1,. . .,%)

moving at the same energy level h, possesses a wnique modal line in an &-neigh-
borhood of that of the parent system.
The proof of this theorem is lengthy and will not be reproduced here.
In view of this theorem, the modal line of the perturbed system may be
written as
(10.8) x&) = cixl + E&~(X,) + e*&(xl) + . . ., (i = 2 , . . .,n)
and this is valid everywhere except, possibly, on the La-surface where the
system M is singular.
Then, substituting (10.8) in (3.12); i.e., in the system MI and expanding
in powers of E , it turns out that must satisfy the system of linear, inhomo-
geneous, second-order equations

(10.9)
where

(10.10)

and a sirn..ar notation is used for oo* and for the partia. derivatives of
these quantities. Moreover, it turns out that hi must satisfy a system
of equations in which the left-hand side is identical to (10.9) while the right-
hand side becomes a function the tlj. Consequently, one can find &, if one
can solve (10.9), and we need concern ourselves only with (10.9); i.e., with
first-order perturbations.
 NONLINEAR VIBRATIONS 201

3. Integrable Cases

There exist certain cases in which the solutions of (10.9) is reducible

to quadratures [8, 181. Where this cannot be done, (10.9) must be integrated
by numerical methods.
One of the integrable cases is that of weak coupling where the forces,
exerted by the coupling springs under finite displacements, are of order E ;
i.e.,

(10.11)

Under the assumption of weak coupling, the system (10.9) reduces to

where the subscript 1 on the tj1has been deleted.

Defining the function
(10.13) uo* + h = G(Xi),
it can be shown that the system (10.12) becomes the self-adjoint system

(10.14) 2(G[i')' - (G't;)' = - C&(Z~), (i = 2 , . . .,n)

where

(10.16) =5
a@,) h= 1
ch2 5
XI

0
(cj awe*
ax,
- w)
axj
dx,, (i = 2,. . .,%).

The general solution of (10.14) is

where A j and Bi are constants of integration.

Inasmuch as the modal line of the perturbed system must pass through
the origin, one can readily show that Bd = 0. (i = 2,. ..,n).However, the
only other condition on modal lines-that they intercept the &-surface-
does not fix the values of the Ad. Nevertheless, it has been shown for systems
m2 R. M. ROSENBERG

with two degrees of freedom [8], and by Kuo [18] for those with any finite
number of degrees of freedom, that the Ai = 0, (i = 2,. . .,n)as well, when
coupling is weak. The arguments leading to this result are lengthy and
not simple, even in systems with only two degrees of freedom, because the
system of equations M is singular on the La-surface. Therefore, we shall
not reproduce them here. However, the result leads directly to
Theorem X-11: The slope at the origin of the modal line of the Perturbed
system i s the same as that of its weakly coupled parent system, on which it
neighbors.
It follows that the modal line of the perturbed system is given, to first
order in E , by
(10.17) xj(xl) = cix, + e&(xl), (i = 2,. ..,n)
where

(10.18)

and the. 0,are defined in (10.16).

I t is evident that the modal line of the perturbed system is curved in
the case considered here because the modal line of the perturbed system
has, at the origin, the same slope as has the straight modal line of the parent
system, but everywhere else, these modal lines do not coincide. Hence,
the corresponding normal-mode vibration is nonsimilar.
A number of examples, involving perturbations of mass .and/or spring
parameters have been computed [8, 181. In one of these [8], results of
first-order perturbation theory have been compared with the actual behavior
of the physical system. Even though, in that particular case, IeI is much
larger than contemplated in perturbation theory, the agreement is excellent.

4. General Properties

It was observed in Theorem V-VI that every modal line, straight or

curved, intercepts the L,-surface orthogonally. One can also show readily
by means of (3.12)that a curved modal line intersects no equipotential
surface orthogonally, other than the La-surface. Moreover, Theorem V-V
states that straight modal lines intersect every equipotential surface ortho-
gonally. From these observations it follows that a straight modal line
is invariant under changes of energy level h, but a curved modal line is
not. In other words, if the modal line in a given mode of motion is curved,
a new modal line must be computed for each value of h.
 NONLINEAR VIBRATIONS 203

Consider now (4.4) with f,(t) 0; i.e., the equation determining the
period of a normal-mode vibration. It is well known that the period T,
given in (4.9), is independent of X, if (4.4)is linear, but it does depend
on X, if (4.4) is nonlinear. Summarizing these results leads to an interesting
gradation of normal-mode vibrations. In it, we replace the words “modal
line” by “mode.”
(i) In linear systems, the normal-mode vibrations are always similar;
the normal-mode coordinates decouple the equations of motion for arbitrary
motions; mode and period of the normal-mode vibration are independent
of the energy level of the motion. Linear combinations of solutions are also
s o htions .
(ii) If the system is nonlinear, the normal-mode coordinates decouple
the equations of motion for that mode only. If the normal-mode vibration
is similar, the mode is independent of the energy level, but the period is
not. Linear combinations of solutions are no longer solutions.
(iii) If the system is nonlinear and the normal-mode vibration is non-
similar, mode as well as period depend on the energy level; otherwise, the
conclusions are the same as in (ii) above.

XI. EXACTSOLUTIONSTO STEADY-STATE VIBRATIONS[36]

FORCED
1. Introductory Remarks
Steady-state forced vibrations can only arise in nonautonomous systems,
They are periodic motions of certain physical systems that are acted upon
by time-dependent, equiperiodic forces. We employ
Definition 111: Steady-state forced vibrations of an admissible, nonautono-
mous system are vibrations-in-mison having as their least period that of the
exciting forces.
In all that follows, we shall assume (as is usually done in the linear
problem) that only one of the masses, say m,,is subjected to an admissible
forcing function. Hence, the system H is

(11.1)

and the system M is

where primes denote differentiation with respect to x,.

204 R. M. ROSENBERG

Among the systems represented by (11.1) and (11.2) there exists an

entire class whose steady-state forced vibrations can be determined without
approximation. It is clear that the linear system, for instance, belongs to
that class. We shall show that exact solutions of the steady-state forced
vibration problem can be found for the entire class of homogeneous systems
under suitable excitation; among them is the linear system under simple
harmonic excitation.
The technically most interesting information regarding steady-state
forced vibrations is contained in the so-called “resonance curves” well known
from linear theory. These are the families of curves in the frequency-
amplitude planes which relate, for prescribed amplitudes of the forcing
functions, the amplitudes Xi of the motions in the various degrees of freedom
to the frequency o of the exciting force. In the steady-state vibrations
.
of nonlinear systems these response curves Xd(o), (i = 1.. . ,w) play a
similar role. We shall show that these curves are integrals of equations
of the form [35]

(11.3)

and that the Xiw-planes contain certain singular points. The number,
location, and character of the singular points of (11.3) can then be used to
determine the geometrical character of the response curves. In particular,
one can ascertain where (if a t all) resonance occurs, and at what frequencies
(if a t all) the phenomenon of “tuned dynamic vibration absorption” takes
place.

2. Homogeneom Systems

The steady-state response of the linear system to periodic excitation

is normally discussed for the case when the excitation is simple harmonic
[36, 371. In fact, the response curves cease to have meaning under general
periodic excitation. Since the steady-state response of the linear system
to simple harmonic excitation is, again, simple harmonic, it turns out that
in that case,

(11.4) xi(t)/f(t) = const., .

(i = 1,. . ,n).

We shall show that the linear system under simple harmonic excitation
is merely a special case of the homogeneous system under the excitation
V1(t)lkwhere
(11.6) f i ( t ) = P,cam (Ha),
 NONLINEAR VIBRATIONS 206

and the steady-state response of this class is of the form

(11.6) x&) = X i cam (%at), (i = 1 , . . . ,n);
hence, (11.6) and (11.6) are such that (11.4) is satisfied for all k in 1 k < w <
<
(or 1 n < 00). We shall make this demonstration only for a system having
two degrees of freedom [35], but we indicate the results of generalizing
to any number of degrees of freedom [12].
Let the degrees of freedom be
(11.7) u, = u, 21, = v,

and write the forcing function in the form

(11.8)

where
(11.9) D = (cX)~-*,
and c is a constant. Form (11.8) has a certain advantage over (11.5). In
(11.8), the quantity I plays a role similar to that of the circular frequency
of simple harmonic excitation, but it is not the frequency of excitation when
k > 1. Hence, I will be called the generalized circular frequency.
Then, the equations of motion of the physical system are

(11.10)
m,ii + aluk + a& - v ) k = P, camk (n~ / ( D A ~t )/ ,N )
m,i + a3vk- a2(u- v ) =~ 0,
those of the pseudo-system are

(11.11)
x+-=
.* ;L( Vm1
m, 7(
+=-=-=
m1 Vm1 VY_,)*=+camh(nEt),

and the equation of the trajectories is

208 R. M. ROSENBERG

( 11.12)
Simple substitution of

(11.13)
y = cx

into (11.11) and (11.12) (and making use of the formulas (9.33) for the
derivatives of cam-functions) shows that (11.13) are indeed the solutions,
and the constant c in the second of (11.13) is the same as that in (11.9).
But it follows from (11.13) that, in the case considered, x(t) and y(t)
are cam-functions of the same argument. Hence, the steady-state vibrations
are similar, or the trajectories satisfying (11.12) are straight lines.
If we define
(11.14) Y = cx.

the equations (11.11) and (11.12). with (11.13) and (11.14) substituted in
them, are three equations in the three unknowns X,Y and c, and they can
be solved explicitly for these quantities.
One finds
c = aJl*/{a,'Ik + (a, - ~ J h + W J ~ ) 1 / b } ,
(11.16) Xk = { P o m ~ / 2 [ a , + * - ?%a(~+~l/*A*)1P]k)/R,
~ / (as

Yk = { Poa,m,RI,}/R
where
 NONLINEAR VIBRATIONS 207

(11.17)

and substitution shows that

(11.18)
x1 = x,cam (n YFt)
Xj(t)/Z,(t) cj = X j / X , , (i = 2 , . . .,m)
satisfy this system as well as the equations of the trajectory which are not
reproduced here. Now, there are m quantities Xi and m - 1 quantities
cj which constitute the unknowns, and there are m equations of motion
and m - 1 equations of the trajectory. Hence, the problem is determinate
WI.
It is now advantageous, as done in the linear forced vibration problem,
to introduce non-dimensional quantities as follows:

(11.19)

y = P/[a,/m,(k+1)q
where it should be remembered that X and P are nondimensional ampli-
tudes, and y is a nondimensional generalized frequency ratio. With the
introduction of these quantities, (11.15) become

where
208 R. M. ROSENBERG

One must now find the locus of the endpoints of the trajectories, i.e.,
the L-surface, which is, here, a curve. The L-curves are found as follows:
the first of (11.20)is solved for y as a function of E. This is substituted in
the second, thus eliminating y , and resulting in an equation of the form
(11.22) xk = qE).
If one substitutes in that equation
(11.23) E= PIR,
one finds an equation in x
and p ; this is the equation of one of the loci
of end-points of the trajectories. The second locus is found by substituting
x
- for 19; then, the equations of the two loci become [36]
(11.24) x k y + aa1(8- p ) k ( p+ p:2x\) - a z l a , , p l ~ ~ ~ *p = 0.
2

From these equations one can readily deduce the following properties of
the L-curves :

(i) they intersect the y-axis at y = 0 and at 9 = f a i l ” ;

(ii) they intersect the &axis at the origin only;
(iii) in the linear system ( k = l), they are hyperbolas;
(iv) as 1x1
and (91+00,718- constants equal to the slope of the modal
lines of the autonomous system [36].
The last property proves the following
Theorem XI-I: A s the generalized freqzcency ratio y + y1,2, where yl,2
are the generalized frequency ratios of normal-mode vibrations of the autonomozcs
system, the nondimensional amplitudes IxI,IP1 -* w . Hence, the system goes
into resonance in the neighborhood of free normal-mode vibrations.
The L-curves and trajectories of steady-state vibrations have been
computed for the nonlinear problem

2
(11.26) ple = 2, aB1= 112, as2= 3, k =3
and for the associated linear problem in which all parameters have the
values given in (11.25),except that k = 1. The results in the 27-plane are
shown for these two examples in Figures lO(a) and (b), respectively.
For an understanding of the general theory, to be presented in the next
section, it is helpful to interpret these diagrams. In general, the trajec-
tories of steady-state forced vibrations are, here, segments of straight lines
 NONLINEAR VIBRATIONS 209

( b)

FIG. 10. Trajectory of steady-state forced vibration in (a) nonlinear and (b) corre-
sponding linear systems.
210 R. M. ROSENBERG

that pass through the origin and terminate on the L-curves. Their slope
is a function of the frequency y and, in fact, their slope E(y) is given in the
first of (11.20).
Assume that y, i.e., essentially the frequency of excitation, is small,
so that E is less than the slope of the modal line in the first and third quadrant
(shown dotted). As y is increased, the trajectory begins to rotate in a counter-
clockwise direction and becomes longer, tending to infinity as the slope
E(y) approaches that of the modal line. As y is further increased, the trajec-
tory continues to rotate, but it becomes now shorter until E = 00. This
occurs where G ( y ) = 0 and, at this frequency, there is bounded 9-motion,
but no 3-motion. In other words, the +motion vanishes identically because
m2 acts as a tuned dynamic vibration absorber for m,. A further increase
in y will further rotate the trajectory, and the motion becomes out-of-phase
because the slope ?< 0. At first, the trajectory will lengthen until the mo-
tion becomes, again, unbounded when y = yz. However, further increases

these amplitudes tend to zero as y -.

in y will reduce the amplitudes of the steady-state forced response, and
m.

3 . The Response Curves [36]

The differential equations, satisfied by the response curves, are found

from the second and third of (11.20); they are

(11.26)

where
(11.27) F ( y )= ~ G ’ o v w- G(YW’(Y),
and G and H are defined in (11.21).
Both of (11.26) are seen to have singular points on the y-axes a t the
zeros of H ( y ) . The first of them has additional singular points on the y-axis
at the zeros of G ( y ) .
Now, the generalized frequency ratios of normal-mode vibrations of the
autonomous system are, in general, functions of the amplitudes R and P,
and they tend to certain limits as the amplitudes tend to zero. It can be
shown [12, 131 that their limiting values coincide with the zeros of H ( y ) .
In other words, the equation

(11.28) H(Y) = 0
 NONLINEAR VIBRATIONS 211

can be shown to play the part of the so-called “frequency equation” of the
autonomous system. Hence, both equations (11.26) have singular points
at the natural frequencies of normal-mode vibrations at vanishingly small
amplitudes.
Denote the zeros of H(y) by y,,. Then expanding the first of (11.26)
in the neighborhood of (K= 0, y = y,,) one finds

(11.29)

because
(11.30) fi(yn) = - G(y*)H’(yn).
Expanding the second of (11.26), one has directly

(11.31)

The integrals of (11.29) and (11.31) are

(11.32) .pm” = const., Py,,k = const.

These are equations of hyperbolas; hence, the singular points on the

y-axes at the zeros of H ( y ) are for both (11.26) saddles.
Let the zeros of G(y) be denoted by ym. Then, expanding the first of
(11.26) in the neighborhoods of (8= 0, y = ym) one finds

(11.33)

because

(11.34) fi(yu) = kG‘(yu)H(yu).

The integrals of (11.33) are the straight lines

(11.36) Rly,,, = const.

Hence, the singular points of the first of (11.26) on the y-axis at the zeros
of G(y) are star points (nodes).
Now, it follows from the definition (11.21) that Gly) has only one real
zero at
(11.36) yu = y* = 1 + Q.2.

Hence, we have [36]

212

FIG. 11. Frequency-amplitude curves in (a) &-plane and @) ?y-plane.

 NONLINEAR VIBRATIONS 218

(a) ( b)

FIG. 12. Frequency-amplitude curves in physical %&plane (a) and l b p l e n e (b).

<
Theorem XI-11: The system (11.11)i s szcch that, for any k ifi 1 k < 00,
there exists a single nondimensional generalized fveqzrency ratio y* of the ex-
citing force swh that the mass m, acts as a tuned dynamic vibation absorber
for the mass ml that i s bcilrg excited.

Response curves in the 8 y and Py-planes for the values of the parameter
given in (11.26) are shown in Figures ll(a) and @), respectively.
It remains to map the Xy and Py-planes on the KQ and PQ-planes,
where the frequency Q of the forcing function is defined through (11.6).
It follows from (9.23) and (11.19) that this mapping is given by [%I

and by a similar formula involving P which is found by replacing 8 in

(11.37) by l’/E The results of this mapping for the example of Figure 11
is shown in Figures 12(a) and (b).
214 R. M. ROSENBERG

XII. STEADY-STATE FORCEDVIBRATIONS BY APPROXIMATEMETHODS

1. Descriptkorr of the Method [12]

The problems of steady-state forced vibrations, considered here, are those
of systems that neighbor on systems of known vibrations-in-unison. Two
cases may arise : either the vibrations-in-unison of the unperturbed (parent)
system are themselves steady-state forced vibrations of the parent system,
or the vibrations-in-unison of the parent system are normal-mode vibrations
of that system. It is clear that in the first case the amplitude of the forcing
function need not be small, but in the second case, that amplitude must be
small in absolute value because the nonautonomous system now neighbors
on the autonomous one. In either of these two cases, the problem of finding
steady-state forced vibrations can be reduced to the perturbation problem
discussed in the section on nonsimilar normal-mode vibrations. The results
reported here are, largely, due to Kinney [12].
We consider a parent system of potential function
(12.1) u = U(X,, . . ,xn)
I

for which at least one vibration-in-unison is known. Let the trajectory in

configuration space of this vibration-in-unison be given by the (known)
functions
(12.2) xj = X j * ( X , ) , (i= 2,. . .,n).
Now, the only vibrations-in-unison that have been deduced are either
straight lines, or they are curved in the sense treated in the section on non-
similar vibrations. Hence, they are always given by
(12.3) Xf*(X1) = cp, + O(a,)
where is small, and when the trajectory is straight, the terms of order
E~ are absent in (12.3).
The time-history of x1 of the known vibration-in-unison is the (known)
function
(12.4) x1 = x,*(t),
and the time-histories of the other xi are found by substituting (12.4) in
(12.3). Since the motion is a vibration-in-unison, (12.4) possesses on the
interval LO,X,] (where X, is defined by %l(Xl)= 0) the inverse
(12.6) t = t*(x,).
We shall, first, suppose that the parent system is nonautonomous, or
the unit mass of the unperturbed pseudo-system is subjected to a force
in the /,-direction given by
(12.6) fl = f(t) = f(t + T)
 NONLINEAR VIBRATIONS 215

where T is the only period of f . Hence, the circular frequency of the periodic
force is
(12.7) w = 2n/T.

We consider a neighboring, or perturbed system of potential function

(12.8) 0 = U(x,, .. .,xn) + &+(XI, * * -,%&)

where #(xl,. . .J,,) is a given perturbation potential which satisfies the

admissibility conditions, and
(12.9) &* = O(&)
everywhere where t,4 is defined. Then, the trajectories of the perturbed
system are the xi(xl),(j= 2,. . .,n)which satisfy
"

i.e., t * ( x l ) is the zeroth order approximation of t ( x l ) .

If one now uses an iteration scheme with respect to t(xl), combined
with conventional perturbation techniques for the xi(xl), the problem
of determining the steady-state forced vibrations reduces to that of (10.9),
treated in the section on nonsimilar normal-mode vibrations. The procedure
consists in replacing t ( x l ) by to(x,) in (12.10) and then using conventional
perturbation methods ; these result in a linear nonhomogeneous system
of equations which the &(xl) must satisfy. Now,the solutions of that system
are functions of to(%,),and this will be indicated by using an appropriate
superscript on the ti. Using the notation

(12.13) Q*(xJ = Q ( x 1 ~ 2 . 1 t - scnxi)

for U and t,4 and their partial derivatives, and using the definition of F(x,,X,)
given in (3.10), the equations of the tj turn out to be
216 R. M. ROSENBERG

(12.14)

But, this system of equations is of the form (lo.$), and questions of its
integration are discussed in the section on nonsimilar normal-mode vibrations,
and also by Kinney [12] for different examples.
Once the integrals #o)(x,) have been found the functions (12.11)
(12.16) xj(x1) = cjxl + E#')(x~), (i = 2,. . 1%)

are substituted in the first of the equations of motion; that equation can
then be integrated by quadratures for the conditions
3, = 0 when x1 = X,,
(12.16)
xl = XI when t=0.

This results in a function

(12.17) x, = X,(t,X,)
and its inverse
(12.18) t = tl(xl).

This latter is the first approximation of t ( x l ) (as compared to the zeroth

approximation), and it may now be substituted in (12.10), and the entire
procedure is repeated. Kinney [12] shows that it is more convenient to
introduce the transformation
(12.19) t = wt,

and he has computed a number of interesting examples, including that of

the tuned dynamic vibration absorber of weakly nonlinear two-degree-of-
freedom systems.
If the parent system is autonomous, one simply deletes the terms in
..
(12.10) involving f ( t ( x l ) ) and F(x,,X,), and one replaces y5(x1,. J,,)by
g(t(x,)) where g(t) is periodic of period T. In view of (12.9), this latter step
implies that the forcing function has small amplitude] and the result will
be that of steady-state forced vibrations in the neighborhood of the vibrations-
in-unison of the autonomous nonlinear system. It may be interesting to
interpret the results of forced steady-state vibrations neighboring on the
 NONLINEAR VIBRATIONS 21 7

vibrations-in-unison of a parent system. This interpretation will be made

in terms of trajectories in the configuration space, and of response curves
in the frequency-amplitude planes. As an example, we consider a parent
system that is linearizable within the meaning of (2.11).
Consider the n-space and an equipotential surface indicated in Figure 13
together with the modal lines; in this diagram, the modal lines are sur-
rounded by &-tubes. Then, the desired trajectories of the steady-state

FIG. 13. Trajectories of free and forced vibrations in nonlinear n-degree-of-freedom

system.

forced vibration pass through the origin and are monotone because they
are those of a vibration-in-unison; moreover, they must lie wholly in the
&-tubesbecause of (12.11). Consequently, the L-surface which is the locus
of the endpoints of these trajectories, also lies inside the &-tube.
The direction with which the trajectories pass through the origin is a
function of the frequency w of the exciting force as explained in connection
with Figure 10. If that frequency lies near that of a normal-mode vibration,
the trajectory lies near a modal line and is long. If not, the trajectory is,
generally, short, and the amplitudes of the forced vibration are of O(E).
In this way, one understands readily the occurrence of resonance.
It is also instructive to translate this information into the frequency-
amplitude planes, shown in Figure 14. The dotted lines, composed of the
w-axes and the backbone curves [29, 381, are those of every vibration-in-
unison of the autonomous system. In other words, if the autonomous system
does not movein normal-mode vibrations, it is at rest with respect to admissible
motions. (Because of Definition I, every vibration-in-unison of the auto-
nomous system is a normal-mode vibration.)
Hence, we may surround the w-axes and the backbone curves by &-tubes
as shown in the diagram, and the forced steady-state vibrations of the non-
autonomous neighboring system must be represented by response curves
lying inside these &-tubes. One such response curve, for a prescribed, small
force-amplitude is shown. It is evident, then, that the intersections of
218 R. M. ROSENBERG

w-axes and backbone curves are saddle points. In addition to these, there
may be star points as well, if the steady-state forced vibrations are similar.
If they are not, one can easily show that the only singular points in the

FIG. 14. Frequency-amplitude curves in nonlinear rP-degree-of-freedom systems.

frequency-amplitude planes occur at the intersection of backbone curves

with the w-axes, and all are saddles.

XIII. STABILITY

1. Definitions
Two definitions seem particularly useful in the examination of the
stability of vibrations-in-unison : “stability in the sense of Liapunov,”
or L-stability and “stability in the sense of Poincarb” or P-stability [39].
To define the former, let
x; = % i +. (.t ) , I
(13.1) (i = 1,. ..,n),
x;*(O) = xi, &*(O) = 0,
 NONLINEAR VIBRATIONS 219

denote the set of solutions of a vibration-in-unison, satisfying the system

H. Further, let
(13.2) xi = Zi(t), (i = 1,. . .,n),
denote any set of solutions of the system H. Then, the set x,*(t) is called
L-stable if, for every E , with O < e << X i , (i = 1,. . .,n),there exists a S(e),
<
with O < S(E) E , such that
(13.3) IZi(t) - x;*(t) I < E
whenever
(13.4) IXi(0) - S*(O)l< b(E).

To define P-stability, let

1
xi = .j*(X1),
(13.5) (i = 2, . . .,N),
Xi(&) = xi,

denote the set of solutions of a trajectory corresponding to a vibration-in-

unison, and satisfying the system M. Further, let

(13.6) xi = X i ( % , ) , (i = 2 , . . .,n)

denote any set of solutions of the system M. Then, the set xi*(xl) is called
P-stable if, for every E with 0 < E << X i , (i = 1,. . . ,n),there exists a 6(e)
<
with O < S(E) E such that
(13.7) Ixi(x1) - xi*(xI) I < E
whenever
(13.8) Ixi(x,) - Xi*(X,) I< 44.
The interpretations of these two definitions of stability are clear. Let
.
the space having coordinates (x,,. .,x,,,t) be called the event-space. Then
..
x,*(t), (i = 1,. ,N) is a curve in the event-space. If any solution xi@),
having a single point inside an &-tubeabout xi*(t) remains within it every-
where, the solution is L-stable. Similarly, xi*(xl) is a curve in the con-
figuration space. If any solution xi(xl), having a single point inside an
e-tube about xi*(xl) remains within it everywhere, the solution is P-stable.*
It is evident, therefore, that every L-stable solution is also P-stable, but
the converse need not be true. Moreover, every solution that is P-unstable,
is also L-unstable but, again, the converse need not be true.

This type of stability is frequently called orbital stability [a].

220 R. M. ROSENBERG

In both definitions of stability it is required that I E ~ be small. Consequently

it is sufficient to consider the equations of the first variation of H with
respect to xi*($) ; i.e.,

(13.9) &= jzl

I
as
tjjgjg [V(x,*(t),.. .,x,,*(t))], .
(i = 1,. .,n).

Clearly, this is a system of linear equations with periodic coefficients, and

stability in any sense requires that all characteristic exponents of (13.9)
have real negative parts. Now, the number of characteristic exponents
depends on whether the system H is an explicit function of t, or not; i.e.,
whether the system H is autonomous or not [26]. Hence, the two cases of
normal-mode vibrations and of steady-state forced vibrations must be
discussed separately.

2. Stability of Normal-Mode Vibrations

I t is simple to deduce [21]

Theorem XIII-I: If an admissible potential function U is not a quadratic
form, the normal-mode vibrations are never L-stable.
This result is an immediate consequence of the well-known fact that the
periodic solutions (if they exist) of autonomous, nonlinear second-order
equations are not isochronous. It follows from this theorem that normal-
mode vibrations of nonlinear systems are, at most, P-stable, and their
stability properties depend in general, not only on the magnitudes of the
mass and spring parameters, but on the energy level of the motion as well.
The problem of discussing the behavior of the ti in (13.9)is greatly
simplified when that system of coupled equations can be decoupled. If the
system (13.9)is written in the matrix form

(13.10) i'+B(t)E= 0
it turns out that the matrix B(t) is of the form

(13.11)

where B, and the Bi are square matrices with constant elements, and the
fi(t) are periodic scalar functions of time. Then, if a transformation matrix
T exists such that
(13.12) T-'B(t)T = d ( t )
 NONLINEAR VIBRATIONS 221

where A ( t ) is a diagonal matrix, the transformed system (13.9) is decoupled.

Hsu has shown [M] that the conditions, necessary and sufficient for the
existence of such a transformation matrix T is, that the Bi, (i = 0,. ..,m)
be all commutative and that at least one of them have distinct eigenvalues.

3. Homogeneozcs Systems
It is simple to demonstrate [7]
Theorem XIII-11: The system of variatioml equations of admissible,
homogmeous systems with respect to any normal-mode vibratioH can dways
be decwpled.
To prove it, one shows that, in homogeneous systems, (13.11) can always
be reduced to

(13.13)

where A is a constant, square matrix. Thus, Hsu’s criterion is trivially

satisfied.
The normal-mode vibrations of homogeneous systems have the further
(unusual) property that their stability does not depend on the energy level
of the motion. In this respect, they are similar to linear systems, the latter
being stable at all amplitudes. If a normal-mode vibration of a homoge-
neous system is stable (unstable) at any one amplitude, it is stable (unstable)
at all amplitudes [9]. Whether a given normal mode vibration of a given
homogeneous system is stable or not depends only on the mass and spring
parameters of the system, and on the particular normal mode examined,
but not on the energy level.
As an example, consider a homogeneous system having the degrees of
freedom zc and v . Then, the equations of motion may be written in the
form suggested by Loude [41]:

(13.14)
%ii = - b#(v) + A#(% - v )
where
(13.16) $(w) = wlwl*-l.
If the trajectory of a given normal mode is

(13.16) v = cu;
222 R. M. ROSENBERG

(13.17)

one can show that the variational equations have solutions that satisfy [41]

where B takes on one or the other of the two values

Bl = C,
(13.19)
B, = - ml/cm,.
The variational equations reduce to

(13.20) € +4 s $ ( q ) € =0

where b,,*are constants defined by

D 1 --- a
+ -((IA - c)Il - c(k--1,
m1 m1
(13.21)

It is easy to show that these equations are independent of the am-

plitude X. To do this, we note that the motion in normal modes is an
ateb-function. (The arguments which follow are the same, whether we deal
with Sam- or cam-functions; thus, only one of these need be considered.)
From (9.6) and (9.34),
x = X cam (nt),
(13.22)
T = (c/n)"2X"- ' t .

But, the independent variable in (13.20) is t, not t. Now, one has from
(13.22)

(13.23)

where E is a constant not containing X, and from the definition (13.16)

of the #-function together with (13.22),

(13.24) lp(gl) = Xb-'*(cam (nt)).

 NONLINEAR VIBRATIONS 223

Hence, (13.20) in terms of t, becomes

(13.26)

where Dlg = &/E. It follows that the two equations (13.26) do not
contain the amplitude X.
The discussion of the characteristic exponents of the equations (13.26)
is difficult because they are Hill equations in which the periodic coefficients
axe cam-functions. The detailed stability properties of such Hill equations
are not known.

-1 0 1 2 3 4

FIG.16. Strutt chart.

However, one can gain some insight into the stability of normal-mode
vibrations of homogeneous systems by replacing the cam-function in (13.25)
by a cosine-function. Proceeding in this manner is, in fact, equivalent to
expanding the periodic coefficient into Fourier series and retaining only the
first term of the expansion. This is usually done in examining the details
of stability of vibratory motions of nonlinear systems [4]. It has the con-
sequence that equations (13.26) become Mathieu equations whose stability
properties are well known [a, 421.
224 R. M. ROSENBERG

One finds, in place of (13.25), two Mathieu equations of the form

(13.26) €” + (81.2+ &1,2cos t)€= 0

and for stability of the normal-mode vibration. it is necessary, that both

these equations “indicate stability.“
The stability study is carried out by means of the Strutt chart [4].
This chart consists of a set of curves in the &-plane separating stable regions
(shaded) from unstable ones; it is shown in Figure 15. If the values P1(6,,el)
and P , ( ~ , , Edefine
~ ) two points PI and P, in the &&planewhich lie in the
interior, or on the boundary, of a stable region, the motion is stable. If
one or both of these points lie in the interior of an unstable region, the
motion is unstable. It turns out that the values of the parameters e,b in
one of the equations (13.26) is such that one of the points P ( E , ~always
)
lies on a boundary between stable and unstable regions. This is known to
lead, at most, to P-stability [4], thus verifying Theorem XIII-I. The other
point may lie anywhere in the &&plane.
The results of the computation for
(13.27) B=B1=c
are given below for that one of the two Mathieu equations for which P(e,S)
does not automatically lie on a boundary.
That equation is
(13.28) F’+ (6 + & C O S t ) 6 = 0
where [9]

(13.29)

k
B = - [a + A l l - cIR-1(1 - c)].
 NONLINEAR VIBRATIONS 226

For any given system parameters] the formulas (13.29) can be utilized to
determine the location of P(S,&). Similar results are found, when
(13.30) B=B,=-m 11cm8'

In either case, it turns out that the point P(e,S) lies on a straight line in
the &plane which passes through the origin, which lies always in the right
half-plane, and whose slope is a function of k only. When K = 1, this slope
is zero so that the point lies then on the (positive) S-axis which consists
entirely of stable points. Thus, linear systems are always stable. When
k > 1, the slope of the straight line is positive, and when O < K < 1. the
slope is negative. The location of P(8,e) on the straight line is a function
of the system parameters [28].

4. Symmetric System with Two Degrees of Freedom [9]

As a second example permitting decoupling of the variational equations,
consider the symmetric two-degree-of-freedom system (not homogeneous)
[9]. The equations of the pseudo-system are

(13.31)

and the trajectories of the normal modes are

(13.32) y = Yi*(X) = x ; y =yo*(%) = - x
where the subscripts i and o indicate an in-phase mode and an out-of-phase
mode, respectively. Using the notation

(13.33) e(x*(o*r*(w= Q*
for U and its derivatives, it is easily shown that, in symmetric systems,
one has always
(13.34)
*
uxx = uy,
*
for both, the i and o-modes.
The coupled variational equations are
.. * *
E - uxxt - u x y q = 0,
(13.36) * *
q - UXYE- U,,q = 0.
Under the transformations
tl = 'I+ 5,
(13.36)
c, = q - E,
226 R. M. ROSENBERG

the equations (13.35) become, in virtue of (13.34),

(13.311

and these are, evidently, decoupled.

As a specific example, consider the system+

(13.38) u = u ( L ) + U(N)
where

(13.39)

We now introduce the following notation, already indicated, in part, in

(13.33): when U or its partial derivatives are evaluated along a modal
line, that quantity is marked with *. When, its value along such a modal
line is computed at the values X , y ( X ) = Y, the quantity is marked with **.
If the modal line in question is that of the in-phase mode, the quantity is
supplied with an i-subscript, if it is that of the out-of-phase mode, the
quantity has an o-subscript. If no subscript is supplied the quantity is the
same in the two modes.
With this notation we define:

(13.40)

Note that the quantities arising from the linear part of the potential function
are the only ones independent of the mode.
It will be remembered from the earlier example that the nonlinear
vibrations x*(t),y*(t) will be replaced, in the vuriatiod eqaruhzs, by their
zeroth approximation; i.e., by simple harmonic functions. It is entirely
consistent with this approximation, to replace the frequencies of the normal-
mode vibrations by their Duffing approximations [4]. These are, in terms
of the quantities defined in (13.40)

t The superscripts (L)and (N)indicate the terms in U which give rise, respectively,
to linear and to nonlinear terms in the equations of motion.
 NONLINEAR VIBRATIONS 227

If one utilizes

(13.42)

the equations of the first variation become

(13.43) cY,~+ (&,\+ &\ cos ti,o)Ci,a = 0

where primes denote differentiation with respect to qO,corresponding
super- and subscripts must be used, and

or, combining them,

(13.46)

In (13.44) and (13.46), the + signs are used with the subscript 1, and the
- signs with the subscript 2.
Evidently then, (13.43) represents four equations, those for two modes,
and two equations for each mode. As in the earlier example, one of the
two equations in each mode defines points in the Strutt chart which lie always
on the boundary between stable and unstable regions of the &plane. Hence,
it is the other two equations which determine the stability; these are the
combinations (1,o) and (2,i).
In applying the foregoing formulas to the example of potential function
(13.38) and (13.39), it is convenient to introduce the nondimensional quanti-
ties
228 R. M. ROSENBERG

In terms of these quantities, the parameters become, for the i-mode,

(13.47)

and a similar result in the o-mode. In both cases, the parameters &,8depend
on E ; i.e., on the amplitude of the motion.

FIG. 16. Stability of in-phase mode of nonlinear, symmetric system with two
degrees of freedom.

The results for the i-mode are shown in Figure 16 and are as follows:
(i) The point on the Strutt chart, deciding on the stability of the i-
mode lies on a straight line passing through the point (C#, and intersecting
the positive &axis at a point that depends on oc, only. This straight line
is called a stability chracteyistic.
(ii) The portion of the stability characteristic lying in the upper half-
plane applies yhen u8 > 0, that in the lower half-plane when a, < 0.
 NONLINEAR VIBRATIONS 229

FIG. 17. Stability of out-of-phase mode of nonlinear, symmetric system with two
degrees of freedom.

The results for the o-mode are shown in Figure 17. They differ in some
respects from those for the i-mode.
(i) There exist, again, straight stability characteristics. These terminate
on a point lying on a straight line of unit slope, passing through the origin.
The point on this line. at which the stability characteristics terminate.
depends on a, only. Further, the stability characteristics intersect the
positive &axis at a point whose value depends on a, only. Thus, know-
ledge of a, and ol, determines the stability characteristic and the amplitude
of the motion determines the point on the stability characteristic. Items
(ii) and (iii), listed above for the i-mode also apply to the 0-mode. From
these results, knowledge of El, a,, a3 and the amplitude of the motion permits
the determination of its stability.
230 R. M. ROSENBERG

6. Nomimilar Normal-Mode Vibratiom [S].

We shall consider here the stability of a nonsimilar normal-mode vi-

bration of an autonomous system that neighbors on a parent system of
potential function
(13 -48) u = V(Xl,. . .,xn)
of equations of motion

(13.49) Zi = U,(xl,. . .,xn). .

(i = 1,. .,n)

and having a similar normal-mode vibration

(13.60) xi = Xi*(t), (i = 1,. . .,n).
Let the perturbed system have the admissible potential function

(13.61) 0 = V ( X ~ .,xn)
, e + E O ( X I , . . -,xn)j (14 IUl).
The equations of motion of the perturbed system are
(13.62) . + .
Y i = U,(X,,. .,xn) ~ o , ~ ( z ~.,x,)),
,. (i = 1,. ..,a),

and it has been shown in a previous section that this system has a normal-
mode vibration (in general nonsimilar) of the form
(13.63) g6(t)= xi*(t) + e&(t) + .. ., (i = 1,. ..#n),
Then, one can easily demonstrate [8].
Theorem XIII-111: The stability of the nomttal-mode Vibration (13.53)
of the Perturbed system i s the same as tlte stability of the similar n o d m o d e
vibratiolz (13.60) of the pareNt system.
For the proof, one forms the equations of the first variation of (13.62)
with respect to the solution (13.53). It turns out that these are precisely
the same equations as those of the first variation of the equations of motion
of the parent system with respect to the solutions (13.60). which proves the
theorem. Hence, if the variational equations of the parent system can be
decoupled, the stability problem of normal-mode vibrations of neighboring
systems is solved.
6. Forced Vibratiows

In general, the variational equations of the equations of motion with

respect to a vibration-in-unison cannot be decoupled, even when the tra-
jectories of the motion in configuration space are straight. In these cases,
Hsu has developed criteria for determining the stability [43, 4-41; his results
 NONLINEAR VIBRATIONS 231

are based on the application of the method of Strubel [39] in a slightly

modified form. A comprehensible summary of Hsu’s methods and results
is too lengthy to be reproduced here. However, we shall apply his method
to a particular case of steady-state forced vibrations in which the variational
equations cannot be decoupled; this example is due to Kinney [12].
We consider the nonautonomous, admissible system

21 = Ox,(%# - pxn) + f(@,

(13.64)
%j = Oxj(%l,....xn) ; 0’ = 2,. ..,%I,
where
(13.66) 0 u(+e- - - +n) + -$(%p - - -A),
and
(13.56) . . .,%”)
u = U(%,,
is a negative quadratic form. In other words, the parent system is a linear
admissible, nonautonomous system in which the mass “r, is subjected to
periodic forcing of period
(13.67) T=2 ~ 1 ~ .
We inquire whether a vibration-in-unison of the perturbed system,
given by
(13.68) xi = Xi*(& (i = 1,. ..,%)
is stable.
The equations of the first variation of (13.64) with respect to (13.58)
are

where use was made of notation (13.33) with respect to the second deriv-
atives of U and 4. Since ZJ is a quadratic form, the a W / a x i a x , are constants
for all i and k.
We Write (13.69) in the matrix form
(13.60) 2 + [B, + sB(l)]z= 0
where z is a column matrix, B, is a constant square matrix,
S
(13.81) B(1) =
8-
2 .. B,cosswt,
I$,.

and the B, are constant square matrices.

232 R. M. ROSENBERG

It will be noticed that, here, we have not made the assumption that the
solutions (13.68) are to be treated as simple harmonic for purposes of a
stability analysis. In fact, that assumption is unnecessary here because the
approach through Mathieu equations is not used.
It is now supposed that B, and the B, do not satisfy Hsu's criterion
[40] and that, in consequence, (13.69) cannot be decoupled.
We introduce the transformation
(13.62) z = TE

which is such that T is a constant matrix that diagonalizes B,. Under

it, (13.60) goes over into a matrix equation of the form

g+(n+EzD,COSSd 1
s
(13.63) l=o
s-1

where A
2 is a diagonal matrix with elements
(13.64) ~ 1 < wg' < . . . < wn2,
2

and the
(13.66) D,= [d$'], (S = 1,. . . ,S)
are constant matrices.
Then, Hsu's results show that instability occurs when, and only when,
the following inequalities are satisfied :

(13.66)

Now, the functional relation between the frequency w of the exciting func-
tion and the amplitude X, of x,*(t) is known. Hence, the inequalities (13.66)
determine the amplitudes X , (if any) for which the motion is unstable.
Kinney [12] has applied this theory to the weakly nonlinear problem
of the dynamic vibration absorber. The analysis is too detailed and tedious
to be reproduced here because of the involved form of the matrix elements
dii'. His result for that problem is:
The portions of the response cwves (in the freqwncy-am#Littde planes)
which have negative slope cmespond to unstable motion. All other portions
of these cwves correspolzd to stable motion.
 NONLINEAR VIBRATIONS 233

On the Existence of Simple Trajectories ila Admissible n-Degree-of-Freedom

systems *
In this appendix we employ
Definition A-I: An integral of the autonomous system H (3.3) or of the
system M (3.12)is said to be a simple trajectory in configwation space if
and only if:
(i) it passes through the origin;
(ii) it reaches the bounding, or maximum equipotential surface (6.8);
(iii) it has no tangency with any equipotential surface (6.12)except at
the origin and the bounding surface.
The last restriction was not imposed by Kuo [18] who introduced the
concept of the simple trajectory. But the lack of tangency with any equi-
potential surfaces U + h* = 0, 0 < h* < h is a property of modal lines.
In fact, every modal line is a simple trajectory, and every simple trajectory
which satisfies (i) to (iii) as well as
ax;
-# 0, (i = 1, . . ,n)
I

dx,
for all xl, is a modal line.
With the above definition we prove in this appendix the central
Theorem A-I: Every admissible azctonomotrs system possesses at kast
one simple trajectory.
Before presenting the proof of this theorem it may be helpful to sketch
its essential ideas; these are very similar to those used in the two-dimensional
problem [191.
We consider trajectories in configuration space which pass through (or
issue from) a point Po on the bounding surface; such trajectories are called
T-curves. We then demonstrate that every T-curve passes through an
infinity of isolated distinct points Qi(i = 1,2,.. .), corresponding to con-
figurations of stationary potential energy with respect to neighboring
points on T. The corresponding levels of potential energy are denoted by
hi, and the transit times through the Qj by t i . We show, further, that the
ti and hi are uniquely determined by the initial point P o , and both are
differentiable with respect to P o . Next, we demonstrate that there exists
at least one T = T*, issuing from a point Po* for which the ordered
Qi(i = 1,2,. . ,) lie alternately at the origin and at Po*. Finally, we show
that T* satisfies condition (iii) above ; this completes the proof.

* This appendix is based in large part on work by J. K. KUO [IS] and by

C. H.PAK.
234 R. M. ROSENBERG

We use the following notation;

q = (ql,. . . ,qn) is an n-dimensional physical displacement vector;

q = (g,, . . . ,Qn) is an dimensional physical velocity vector;
0
K = 1/2CmiQia is the kinetic energy of the physical system;
i=l
-0= - #(q) is the potential energy stored in the springs;
h =K 0 is the total energy of the physical system;
VU = grad U is an n-dimensional vector.
We assume that 0 satisfies the following conditions, called “admissibility
conditions” :
(i) U(q) = 0(- q) is of class P ;
(ii) - 0 is positive definite;
(iii) PO = 0 only at q = 0.

(iv) Trajectories are the vectors q(t) which satisfy the equations of
motion,
a
miqi = - U(q).
%i
(i = 1,. .. ,n)
where the mi are real, positive constants. The equations of motion transform
under
-
(A4 x; = vmiqi
into
.. = -
X;
a U ( X )= U,i
ax,
or, in vector notation,
(A.3) P = VU.
With Kuo [18], we transform (A.3) into the system of 2n first-order equations

(A.9 ji = fib)
where
Yi = xi, yi+u = 2i I
fi=Yn+i, f,+n= UIi.
 NONLINEAR VIBRATIONS 236

Since the system is admissible, the functions f j are of class C1, or the Jacobian
J is continuous, where

..

In consequence, (A.4) has a unique solution under specified initial conditions :

y = Y when t = t; that solution is
(A4 y =y(t,t,Y), - =< t< + =.
Conversely, if the initial condition Y = y at t = t is given, one has uniquely
(A.7) Y = Y(t,t,y), - oo<z< + m.
Finally, since J is of class C1. the functions y and Y in (A.6) and (A.7) are
once differentiable with respect to all arguments. If one regards (A.6) as
a mapping which, for fixed t , maps the initial point Y onto y , and (A.7)
as the inverse of this mapping, one notes that this mapping and its inverse
are continuous, or (A.4) is a topological mapping in the phase space. Clearly,
the energy integral of (A.4) exists and is
(A4 - U + 4 IlkllS = h = const
where / [ 11 denotes the Euclidean distance function, h 2 0 is the total
energy of the system, and it is bounded. The configuration of maximum
potential energy is
(A.9) U+h=O.
This is the maximum equipotential surface and it is the bounding surface
for all possible trajectories of admissible systems (Theorem V-11).
All trajectories satisfy
Lemma A-I: No integral curve of (A.4) termirsates at a point.
Lemma A-11: If an integral curve of (A.4) passes through ( X , 8 ) at t = 0
a d it reaches the bounding surface at t = t o , then it passes through ( X , - 2)
at t = 2t0 .
The proofs of these lemmas are simple and will only be indicated. To
prove the first, one supposes to the contrary that a trajectory does terminate
.
at a point x when t = 'cO, Then, for all t > t o ,x = E = . . = 0. But,
when x = 0, the configuration is that of maximum potential energy, and
when j ; = 0, x is a t the origin. Since the origin does not lie on the bounding
surface under our admissibility condition, Lemma A-I is true.
236 R. M. ROSENBERG

For proof of the second lemma one need only observe that trajectories
which attain the bounding surface intercept that surface orthogonally
(Theorem V-V). Hence, the initial conditions (position and slope) are
defined; then, in view of the uniqueness of (A.6), the trajectory must retrace
itself.
I t is now convenient to rewrite the energy integral in terms of the initial
energy when the initial point is not on the bounding surface:
(A.lO) - + tl141s= - U X ) + i l l q *
W)
where x and as functions of the initial values X and d are given in'(A.6)
because y and Y correspond, respectively, to ( x , i ) and (X,ft).
Differentia-
ting (A.lO) with respect to X and ft, we have:

-ax,
__ ... ax, _an,_...- an,,'
ax, ax, ax, ax,
ax,
- ax,
(A.ll)
ax,, axu
ax, ax, aa, an"
_-
ad, ax, a f t , aft,

We shall denote by A(t) the 2n x 2n matrix on the left-hand side of (A.lI), or

and we shall demonstrate the existence and regularity of A(t), because these
are essential for the proof of the central theorem. For instance, the transpose
A* may be found from

(A.13)

with the initial condition A*(O) = I where I is the identity matrix. Since
J is of class C1, the existence, uniqueness, and continuity of A(t) is assured.
But J is a function of the configuration x ; hence the solution of (A.13)
depends on the initial configuration X, and we write it as
 NONLINEAR VIBRATIONS 237

(A.14) A = A(t,X).

Let us consider A as a mapping in the phase space which, for fixed t and X
maps the point (- VU,$ onto the initial point (- Po U , x ) where voU
denotes grad U at X . Then, A has the important property, formulated in
Lemma A-111: For fixed t and X , the mapping A i s volume-preserving;
i.e., the determinant of A is unity. Conseq.uently, a unique inverse of A exists.
To prove this lemma, we make use of the Jacobi-Liouville formula

(A.16)
I
det IA*(t)I = exp tr J(s)ds
0

where J is defined in (A.6), and tr J denotes the trace of J . Now, the elements
in the main diagonal of J are all zero. Therefore, det A is unity and the
inverse A-l exists. Thus, we may write (A.ll) as

(A.16) [-xvu]= A-W [-8vou1’

and A-1 is a one-to-one mapping in the phase space which maps the point
(- r o U , f t ) , for fixed t and X , onto a point (- V U , i ) .
In the following, we shall ogly consider trajectories originating from an
initial point Po lying on the bounding surface; i.e., at x = X , 5 = 0. These
are called T-curves (see Section VII.2) ; they correspond to motions with
a rest-point. We shall now prove

Theorem A-11: Along every T-curve there is a countable infinity of

isolated points Q1 ,Qz,. . . , at which the configuration is one of stationary
potential energy with respect to points on T neighboring on Q1,Qs,. , . .
The proof of this theorem is lengthy, and we shall only sketch it. At
every Qi ,(i = 1,2,. . .) one has

(A.17)
au - 0
--
at

and the theorem holds if (A.17) has a countable infinity of isolated t i .

.
( j = 1,2,. .).
First, we note that (A.17) may be rewritten in the form

(A.18)
238 R. M. ROSENBERG

which shows that the gradient VU is orthogonal to the velocity vector 2

at all Qjor, at the Qi, the velocity vector lies in a tangent plane (actually
the intersection of the tangent surfaces) to an equipotential surface.
Now, T-curves are either tangent to equipotential surfaces at finite times
ti or, if they are not, they tend to an equipotential surface as t + 00. To prove
Theorem A-11, we need only show that the latter cannot arise. We suppose
that a trajectory (not necessarily a T-curve) lies in an equipotential surface.
Such a trajectory either is a closed orbit, or it tends asymptotically and
monotonically to a closed orbit. This latter is true because the transversals
(see VII-1) are, by definition, orthogonal to the equipotential surfaces;
thus, no trajectory, lying in an equipotential surface, can have curvature
components in a tangent plane (as above) to that surface.
Next, we observe that, if there exists a closed trajectory in an equi-
potential surface, it must lie in a two-dimensional plane which contains
the origin. This is true because such an orbit must be symmetric with
respect to the origin by virtue of Theorem V-111. Thus, if a T-curve tends
to a closed orbit which lies in an equipotential surface it will, after a finite
time, have a position and slope which differs from that of the closed orbit
by as little as we please. Since that T-curve satisfies the system M of (3.12).
the closed orbit must satisfy (3.12)within O(E). But it can be shown that
the only two-dimensional orbit lying in an equipotential surface which
satisfies (3.12)is a circle. That demonstration is based on the observation
that [ (211s= const for any trajectory in an equipotential surface. Finally,
we show that equipotential surfaces of admissible systems do not contain
circles in a plane containing the origin, because such equipotential surfaces
occur only in systems which are not elastically coupled throughout; i.e.,
which satisfy (2.4).Thus, no trajectory satisfying (3.12)can be asymptotic
to an equipotential surface; consequently, we have proved that aUlat
vanishes for bounded 4 . By the same arguments as above, no arc of finite
length of a trajectory can coincide with an equipotential surface because,
if it did, the arc would have to be that of a circle about the origin, and no
equipotential surface contains such arcs. This proves Theorem A-11.
Let us denote by hj the energy levels of the Qi,and let us write
(A.19) ti tj(X)
because it has been shown that the ti depend on the initial point X.
We wish to show that every hi is continuous and differentiable with
respect to the initial point X. The differential of U is
n
(A.20) dU = Utdt + 2 UxadXa
a=l

for variations in transit time through a given energy level, and for variations
in the initial point. The existence of this differential is assured by the prop-
 NONLINEAR VIBRATIONS 239

erties of U and the uniqueness of (A.6). Consider now one of the hj , say h, .
Its differential is the same as (A.20) except that U,= 0 by definition.
Hence,

(A.21)

Thus, the hi are continuous and differentiable with respect to the initial
point Po(X).
To prove the existence of a simple trajectory we extremize h j ( X ) subject
to the equation of constraint
(A.22) U(X) + h = 0.
In other words, we determine the stationary values of the potential energy
for a trajectory and we insure by the constraint equation (A.22) that this
trajectory is a T-curve. Thus, we define a function
(A.23) g ( X ) = hj(X) - ilU(X)
where il is a Lagrangian multiplier. Since we wish to deal here with the
phase space we may write, instead of (A.23)

with = 0. Then, the necessary conditions for the extremum are

(A.26)

and, in view of (A.21), we have at t = ti

(A.26)

Then, we may rewrite (A.26) as

(A.27)

where A ( t , X ) has been defined in (A.12). But, if we write (A.16) at t = 4

we have

(A.28)
240 R. M. ROSENBERG

or, combining the last two,

(A.29) A(t, ,X) ['"-"' 7

R = 0.

But, since det A = 1, the only possible solutions to (A.27)are

The first of these states that U,= 0 at the origh since VU = 0 at the origin
only, and the second, that Ut = 0 on the bounding surface because that
surface is defined by R = 0. Hence, a T-curve T* exists which connects
an initial point on the bounding surface with the origin. To prove Theorem
A-I, it remains to show that the T*-curve is a simple trajectory.
Let Po* be the initial point of T* on the bounding surface. Let its points
of first, second, etc. tangency with equipotential surfaces be Q1,Q2,. .. .
Now, suppose that Q1 is not a t the origin. Then, in view of the above
demonstration, Q1 is on the bounding surface. In that case, T* moves at
first toward configurations of lower potential energy and then toward
configurations of higher potential energy. Thus, there is a point R between
Po* and Q1 at which U,= 0. But this contradicts the fact that Q1 is the
first point along T* where Ut= 0. Hence, Q1 is a t the origin. This com-
pletes the proof.

ACKNOWLEDGMENT

This work was supported, in part, by a grant from the National Science
Foundation.

Retmcnces

1. MINORSKI,N.. "Nonlinear Oscillations," Van Nostrand, N. J.. 1962.

2. SYNGE,On the geometry of dynamics, Phil. Trans. A, 226, 31-106 (1926).
3. RAUSCHER. M., Steady oscillations of systems with nonlinear and unsymmetrical
elasticity, Journ. A@. Mech. 6, A169-Al77 (1938).
4. STOKER, J. J., "Nonlinear Vibrations," Interscience Publ., New York. 1950.
5. LEDOUX, P.,Personal communication, Lihge, 1964.
6. B~AWHIN.J., Oscillations en modes normaux de systhmes dynamiques nonlintbires
B plusieurs degrC de liberth, Bull. SOC. Royule Sciences Lilge, 9-10. 540-557
(1964).
7. ROSENBERG, H. M., On normal mode vibrations, PYOC.Cumb. Phil. Soc.. 80, 595-611
(1964).
 NONLINEAR VIBRATIONS 241

R. M. and Kuo, J .
8. ROSENBERG, K.,Nonsimilar normal mode vibrations of non-
linear systems having two degrees of freedom, Trans. ASME. Journ. Awl.
Mech., 81, E, 2, 283-290 (1964).
9. ROSENBERG, R. M. and Hsu. C. S., On the geometrization of normal vibrations
of nonlinear systems having many degrees of freedom, in Proc. IUTAM Symp.
on Nonl. Vibr.. Ukrainian Academy of Sciences, Kiev, 1. 380-415 (1963).
10. KAUDERER, H., Nichtlineare Mechanik, Springer, Gtjttingen, 1968.
11. ROSENBERG, R. M., The normal modes of nonlinear n-degree-of-freedom systems,
Trans. ASME. Journ. A w l . Mech., 29, E, 1, 7-14 (1962).
12. KINNEY,W. D., On the geometrization of the forced oscillations of nonlinear
systems having many degrees of freedom, Ph. D. Thesis, Univ. of Calif., Berkeley,
1905.
13. KINNEY.W. D. and ROSENBERG,
R. M., On steady-state forced vibrations in
strongly nonlinear systems having two degrees of freedom (Les vibrations
forcbs dans les systbmes nonlin6aires). Proc. CNRS I s t . Coll. 148, 351-373
(1965).
14. DARBOUX, G.. “Thbrie a n b r a l e des Surfaces,” vol. 2, pp. 438-509. Gauthier-Villar,
Paris, 1889.
16. J ., “Oscillatory Motions” (Translation of “Les Mouvements Vibratoires”),
HAAG,
vol. 1, pp. 36-39, Wadsworth Publ. Co., 1962.
16. ROSENBERG, R. M., La g6ometrie de la dynamique et les vibrations des systbmes
non-linbaires. L’Onde Ebctrique 48, 433, 425-434 (1963).
17. COURANT, R. and HILBERT,D., ,,Methoden der mathematischen Physik‘, vol. 1,
pp. 24&245. Springer, Berlin, 1931.
18. Kuo, J . K., On non-similar normal mode vibrations of nonlinear systems having
many degrees of freedom, Ph. D. Thesis, Univ. of Calif., Berkeley, 1964.
19. ROSENBERG, R. M., On the existence of normal mode vibrations of nonlinear systems
with two degrees of freedom, Quart. Appl. Math., 22, 3, 217-234 (1964).
20. ROSENBERG, R. M., The ateb(h)-functions and their properties, Quart. Appl. Math.
el, 6, 37-47 (1963).
21. ROSENBERG. R. M., “Les vibrations non-lineaires de systbmes B plusieurs degrbs
de libert6.” Centre de Rech. Phys., Marseille, 1984.
22. SIDERIAD~S, L. and ARGBMI,J.. Sur la singularit6 multiple des systhmes dyna-
miques plans, Comp. Rend., Acad. Scie. 268, 2037-2039, Gauthier-Villar, Paris
(1961).
23. ARGEMI,J. and SIDERIADkS, L., Sur la singularitd multiple dbfinie par les formes
homogbnes des systkmes dynamiques plans, Comp. Rend. 864, 4135-4136.
Paris (1962).
24. ROSENBERG, R. M., On motions with a rest point and the existence of normal
mode vibrations in nonlinear systems, in “Proc. Eleventh. Int. Congr. Appl.
Mech., Munich” 1964, (to appear).
25. MAHAREM, N., On homogeneous systems and the stability of their normal modes,
Ph. D. Thesis, Univ. of Calif., Berkeley, 1965.
26. CODDINGTON, E. A. and LEVINSON, N., “Theory of Ordinary Differential Equations,”
McGraw-Hill, New York, 1966.
27. HAUGHTON. K. E., Similar motion of multi-degree-of-freedom vibrating systems
with nonsymmetric springs, Ph. D. Thesis, Univ. of Calif., Berkeley, 1964.
28. ROSENBERG, R. M., On normal vibrations of a general class of nonlinear dual-mode
systems, Trans. ASME. Joum. Awl. Mech.. S8, E, 2, 275-283 (1961).
30. ROSBNBERG, R. M. and ATKINSON.C. P., On the natural modes and their stability
in nonlinear two-degree-of-freedom systems, Trans. ASME, Journ. Appl. Mech.,
B6, E, 3, 377-386 (1959).
242 R. M. ROSENBERG

E. T. and WATSON,G. N., “Modern Analysis,” Cambridge Univ. Press,

30. WHITTAKER,
London and New York, p. 624, 1927.
31. MIISHKES, A. D., On the exactness of approximate methods of analyzing small
nonlinear vibrations of one degree of freedom, in “Dynamic Problems” (coll.
papers) vol. 1, p. 139, publishers, Riga, 1963 (in Russian).
32. MCDUFF. J. N. and CURRERI,J. R., “Vibration Control,” McGraw-Hill, New York,
p. 72, 1968.
83. LANDAU, L. D. and LIFSHITZ,E. M., “Mechanics,” Pergamon Press and Addison-
Wesley, London, Reading, Mass., p. 21, 1900.
34. PBARSON, K., “Tables of Incomplete Beta Functions,” Cambridge Univ. Press.
London and New York, 1927.
36. ROSBNBERG, R. M.,On a geometrical method in nonlinear vibrations (Les vibrations
forcQs dans les systhmes non-lin6aires), Proc. CNRS In#. ColE.. 148, 69-79
(1906).
86. TIMOSHENKO, $3.. “Vibration Problems in Engineering,” Van Nostrand, Prince-
ton, N. J., 1956.
37. DENHARTOG, J. P., “Mechanical Vibrations,” Third ed., McGraw-Hill, New York,
1947.
38. KLOTTER,K., Steady-state vibrations in systems having arbitrary restoring and
arbitrary damping forces, in “Proc. Symp. Nonl. Circuit Anal. 234-267, Poly.
Inst. Brooklyn” (1963).
39. STRUBLE, R. A., “Nonlinear Differential Equations,” McGraw-Hill, New York.
1962.
40. Hsu. C. S., On a restricted class of coupled Hill’s equations and some applications,
Trans. A S M E , Journ. A+@. Mcch., 28, E. 4, 651-666 (1961).
41. LOUDE,W., Personal communication (1961).
42. MACLACHLAN. N. W., “Ordinary Non-Linear Differential Equations in Engineering
and Physical Sciences,” Clarendon Press, Oxford, 1960.
43. Hsu. C. S., On the parametric excitation of a dynamic system having multiple
degrees of freedom, Trans. A S M E , Joum. Appl. Mech., 80, E , 3, 307-372, (1963).
44. Hsu, C. S.. On the stability of periodic solutions of nonlinear dynamical systems
under forcing, Proc. CNRS Int. CoR., (Les vibrations forc6es dans les systhmes
non-lindaires). 148, 125-141 (1966).
 Fundamental Problems in Viscoplasticity
BY PIOTR PERZYNA
Institute of Basic Technical Research. Polish Academy of Sciences
Warsaw. Poland
Page
I. Physical Foundations . ........................ 244
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2 . Experimental Results for Metals . . . . . . . . . . . . . . . . . 245
3. Experimental Results for Soils . . . . . . . . . . . . . . . . . . 253
4 . Some Theoretical Ideas for the One-Dimensional Case . . . . . . . 254
5 . Assumptions and Definitions . . . . . . . . . . . . . . . . . . . 259
.
I1 Constitutive Equations . . . . . . . . . . . . . . . . . . . . . . . 260
1. The Behaviour of a Material in the Viscoelastic Region . . . . . . . 260
2 . The Yield Criterion for a Viscoelastic Material . . . . . . . . . . . 262
3 . A Definition of Stable Inelastic Material . . . . . . . . . . . . . 264
4 . Convexity of the Flow Surface . . . . . . . . . . . . . . . . . . 266
5. The Direction of the Vector of Plastic Strain Rate . . . . . . . . . 269
8. Constitutive Equations for an Elastic-Viscoplastic Material . . . . . 270
7. Constitutive Equations for Rate-Sensitive Plastic Materials ..... 272
8. Special Cases of Constitutive Equations . . . . . . . . . . . . . . 275
.
9 The Dynamical Condition for a Stable Elastic/Viscoplastic Material . 282
10. Comparison with Experimental Dafa . . . . . . . . . . . . . . . . 283
11 . Temperature-Dependent and Strain-Rate Sensitive Plastic Materials . . 290
12. Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . 297
13. Relaxation Processes . . . . . . . . . . . . . . . . . . . . . . . 298
14. Fracture of Time-Dependent and Temperature-Dependent Materials . . 300
.
I11 Stress Wave Propagation in an Elastic/Viscoplastic Medium . . . . . . 302
1. Mathematical Preliminaries . . . . . . . . . . . . . . . . . . . . 302
2 . Formulation of the Problem . . . . . . . . . . . . . . . . . . . 316
3. Application of the Method of Finite Differences . . . . . . . . . . 316
4. Application of the Method of Successive Approximations ...... 330
5. Solution in the Elastic Region . . . . . . . . . . . . . . . . . . 336
6. Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . 340
.
IV Quasi-Static Solutions . . . . . . . . . . . . . . . . . . . . . . . . 350
1. Spherical Problem . . . . . . . . . . . . . . . . . . . . . . . . 350
2. Viscoplastic Flow of a Circular Plate . . . . . . . . . . . . . . . 352
V . Other Dynamical Problems . . . . . . . . . . . . . . . . . . . . . 361
1 . Wave Problems for Rods and Beams . . . . . . . . . . . . . . . 361
2 . Impulsive Loading of a Spherical Container . . . . . . . . . . . . 362
. . . . . . . . . . . .
3. Strain-Rate Sensitive Beams under Impact 365
.............
4. Longitudinal Impact on Viscoplastic Rods 367
References ............................. 368

243
244 PIOTR PERZYNA

1. PHYSICAL
FOUNDATIONS

The fundamental assumption of all theories of plasticity-that of time

independence of the equations of state-makes simultaneous description
of the plastic and rheologic properties of a material impossible.
I t is well known that in many practical problems the actual behaviour
of a material is governed by plastic as well as by rheologic effects. It can
even be said that for many important structural materials rheologic effects
are more pronounced after the plastic state has been reached.
Every material shows more or less pronounced viscous properties. Today
this is a common place. In some problems the influence of viscous properties
of the material may be negligible while in other problems it may be essen-
tial.
Both sciences- plasticity and rheology-are concerned with the descrip
tion of very important mechanical properties of structural materials. Each
of them has created its own methods of investigation and has developed
within the framework of certain assumptions which, unfortunately, cannot
always be satisfied in reality. The results of rheology are confined to cases
where plastic strain is of no decisive importance. On the other hand, the
results of the theory of plasticity permit correct description of only such
problems where the influence of rheologic effects may be considered unessen-
tial. In other words, if methods of rheology are used we should confine
our considerations to the study of those states of stress that do not produce
plastic flow of the material. If methods of plasticity are used we must
limit ourselves to quasi-static processes the duration of which is sufficiently
short, so that creep or relaxation effects do not occur. However, recent
research concerning the description of dynamic properties of materials has
shown that the application of the theory of plasticity, in which rheologic
effects are disregarded, leads to too large discrepancies between the theoret-
ical and experimental results.
Thus there is no need to point out the advantages that can be gained
by simultaneous description of rheologic and plastic effects, and the general
problem is that of viscoplasticity. Before we proceed to a more detailed
description let us emphasize the fact that the methods of viscoplasticity
belong neither t o rheology nor to plasticity. It will be seen that the specific
character of the problem and the complicated way in which rheologic and
plastic effects are interconnected require special methods, based on careful
analysis of the physics of solids and special mathematical methods.
However, the difficulties of combined treatment of rheologic and plastic
phenomena are enormous. The viscous properties of the material intro-
duce a time dependence of the states of stress and strain. The plastic prop-
erties, on the other hand, make these states depend on the loading path.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 245

Thus, as a result of simultaneous introduction of viscous and plastic proper-

ties, we obtain a dependence on the load history a d on the time. A descrip-
tion of strain in viscoplasticity will therefore involve the history of the
specimen, expressed in the type of the loading process, and the time. Differ-
ent results will be obtained for different loading paths and different duration
of the process.
The aim of the present survey is to discuss the new problems, in which
combined treatment of rheologic and plastic phenomena is essential, and to
describe the principal lines along which the methods of viscoplasticity
develop. Although viscoplasticity cannot yet be regarded as a theory in
its definitive shape, certain research trends can be discerned. The funda-
mental problem of viscoplasticity is the determination of an adequate yield
criterion for a viscoelastic material. Another important problem is that of
establishing suitable constitutive equations.

2. Experimental Results for Metals

It is not the aim of the present section to give a survey of experimental
methods. Rather, a critical review of the results obtained by various research
workers using different apparatus and different methods will be presented.
However, the validity of the results obtained cannot be appraised without
taking into consideration the accuracy of the method applied ; therefore,
in the course of our discussion, we shall try to give a brief characteristic of
the method and an analysis of the validity of the interpretation.
Often the experimental results are obtained correctly by an accurate
method, but their interpretation gives rise to some doubts. Indeed, the
dynamic phenomena occurring in a test piece during the impact process
is influenced by so many factors that the selection of the dominating one
is very difficult. Often a phenomenon is determined simultaneously by
several factors of equal importance. Thus, for instance, for a dynamic
tension or compression test the propagation of stress waves is essential
which, in turn, depends to a large extent on the length and the form of
the test piece, the stress concentration, and the distribution of the strain
rate. The strain rate during an impact test is not constant but varies along
the bar within broad limits and is also variable in time. As a result, the actual
process is extremely complicated. In addition, the entire phenomenon
lasts often only a few tens or hundreds of a microsecond.
An ideal solution would be to eliminate completely the stress-wave
effect, thus allowing the separation of the influence of the strain rate on
the mechanical properties of a metal. This has been pointed out by E. H. Lee
and H. Wolf [104], D. S . Clark [40], and J. D. Campbell and J. Duby [27].
Correct interpretation of results can also be obtained by taking into account
the wave phenomenon in the test piece. This method would, of course,
be more difficult.
246 PIOTR PERZYNA

Very great progress in the experimental investigation of dynamic proper-

ties of materials was made during the last few years; research methods
and measuring apparatus were considerably improved. Thus, errors of
measurement have been reduced, and a variety of causes influencing the
behaviour of a metal during dynamic and static loading has been recognized.
The general interest in the dynamic properties of metals is stimulated not
only by the complicated character of the dynamic problems, but the differ-
ences between the static and dynamic behaviour of a metal are of such
practical importance that the results of static tests can no longer be used
for an appraisal of a dynamic phenomenon.
Depending on the main object of research, experimental investigations
may be classed in various groups. Let the first group be tests which aim
at a detailed analysis of the influence of the strain rate on the mechanical
properties of a metal. Another group are tests concerned with basic research
with the aim to determine the dynamic stress-strain curves of materials.
The third group comprises studies of the distribution of permanent strain
along a test piece. Next, let us mention investigations of the role of the
transversal motion during the propagation of longitudinal stress waves.
The influence of the transversal motion is the increase of the dispersion
effect. It has also been observed that stresses higher than the upper yield
point can be applied for a very short time without producing plastic flow.
This phenomenon may be called the delay-time phenomenon. There are
many investigations in which the delay-time effect is analysed quantita-
tively and qualitatively and which try to explain its causes. An interesting
group of papers are those, in which changes in the static properties of metals
due to previozls dynamic loads are studied. The last two groups of experi-
mental works show how the dynamic behaviour of metals is influenced by
temperature and irradiation.
The results of the tests of G. I. Taylor and A. C. Whiffin [172, 1891,
P. E. Duwez and D. S. Clark [67], J. E. Johnson, D. S. Wood, and D. S.
Clark [SS], and J. D. Campbell [24, 251, have shown that higher stresses
are needed for metals to reach plastic state for a sudden than for a slowly
acting load.
Many theoretical and experimental investigations have pointed out
that metals having a distinct yield limit are particularly sensitive to the
strain rate (cf. for instance M. P. White [lW], J. Miklowitz [116], and
H. G. Hopkins [83]). Very good examples of metals behaving in a different
way during static and dynamic loading and showing considerable rate
sensitivity are mild steel and pure iron.
The influence of the strain-rate on the yield limit for mild steel has
been examined in detail by G. I. Taylor [171], M. J. Manjoine [112],
D. S. Clark and P. E. Duwez [39], F. E. Hauser, J. A. Simmons, and J , E.
Dorn [77], and K . J. Marsh and J. D. Campbell [113].
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 247

G. I. Taylor [171], was able to perform tensile tests over a wide range
of average strain rates by firing bullets against an anvil bar held as a simple
beam between two tensile specimens. The other end of each specimen was
fastened to a ballistic pendulum. Since the velocity of the anvil after im-
pact decreased from the impact velocity to zero, he assumed a mean rate
of strain equal to 1/2(velocity of anvil)/(length of specimen). Applying
the law of conservation of energy and assuming that the kinetic energy
equals the plastic strain energy he determined for steel the variability of
the ratio of the dynamic to the static yield limit as a function of the mean
strain-rate, Fig. 1.

4:
9

*.
9

4 I I I 1 I I I I ,

FIG. 1. Effect of strain rate on dynamic yield in steel (G. I. Taylor [171]).

The results of D. S. Clark and P. E. Duwez, [39],for mild steel (0.22%

carbon, 39,000 lb/ins upper static yield limit) have shown (cf. Fig. 2) that
the proportionality limit, which can in this case be identified with the upper
yield limit increases with increasing strain-rate until it coincides with the
ultimate strength of the metal. The latter quantity increases until the
strain rate reaches the value of about 200 sec-l. In these tests the specimen
was loaded in circumferential tension due to the action of compressed mer-
cury in the bore of the tube.
The properties of low-carbon structural steel of 28,000 lb/ins lower yield
point were studied by M. J. Manjoine [112]. His results are somewhat
different from those obtained by D. S. Clark and P. E. Duwez.
The paper of F. E. Hauser, J. A. Simmons, and J. E. Dorn [77], presents
a method by which the plastic properties of materials can be investigated
at strain rates up to 1.5 x lo4 sec-' through impulsive-loading techniques.
Some of these data, for high purity aluminium at 295' K, are plotted in
Fig. 3 where the stress is represented as a function of the strain for a given
strain rate.
K. J. Marsh and J. D. Campbell [113] using a rapid-loading hydraulic
test machine obtained interesting results of constant-stress tests on mild-
steel specimens of different mean ferrite-grain sizes. These results suggested
248 PIOTR PERZYNA

that the behaviour of the material may be represented by a functional

relationship between stress, mean strain and mean strain-rate, but this

FIG.2, Effect of strain rate on the proportional limit and ultimate strength of a 0.22%
carbon steel (D. S . Clark and P. E. Duwez [39]).

FIG.3. Effect of stress on strain at constant strain rate of high purity aluminium (F. E.
Hauser, J. A. Simmons and J. E. Dorn [77]).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 249

cannot be considered a fundamental one when the strain distribution is

non-uniform.
A direct measurement of the effect of strain rate on the character of the
u, &-curvewas carried out by T. E. Tietz and J. E. Dorn [177].
Experimental investigations of J. D. Campbell [26, 281, J. D. Campbell
and J. Duby [27], J. Harding, E. 0. Wood, and J. D. Campbell [76],
K. J. Marsh and J. D. Campbell [113], have shown that the upper yield
limit for mild steel during a dynamic loading-process may reach a value
2.5-3 times as high as during the static test. These tests show clearly that
the yield limit increases with increasing strain rate. At the same time,
the very important pb<iomenon of a definite reduction of the strain-hard-
ening effect during the process of dynamic loading as compared to the
strain-hardening effect observed during the static test is pointed out.

FIG.4. Super-purity iron based 0.21 per cent carbon steel curves: (a) strain rate against
strain for dynamic test; (b) stress against strain, A. Static, B. Dynamic; numbers
denote time in microseconds; impact velocity 198 in./sec. (J. Harding, E. 0.Wood
and J. D. Campbell [75]).

Dynamic u,C-CUNIS were obtained by J. E. Johnson, D. S. Wood, and

D. S. Clark [88], ana 1. D. Campbell [24], for aluminium and by J. D.
Campbell [26] and J. U. Campbell and J. Duby [27], for mild steel.
J. Harding, E. 0. Wood, and J. D. Campbell [75], determined the dynamic
characteristics for a few metals, among others for mild steel and pure iron
(cf. Figs. 4 and 6). H. Kolsky and L. S. Douch [96], measured the dynamic
stress-strain curves of annealed specimens of pure copper, pure aluminium,
and an aluminium alloy by experiments in which short bars of these mate-
rials were fired from a compressed-air gun at a steel pressure bar. It was
found that for copper and aluminium the dynamic CUN- lay appreciably
above the static ones whereas for the aluminium alloy there appeared to
be no appreciable strain-rate effect.
280 PIOTR PERZYNA

Detailed analysis of the distribution of permanent strain along a specimen

subjected to a dynamic load was performed by many scientists (cf. for
instance Th. KBrmPn and P. Duwez [92], J. D. Campbell [26], and
H. Kolsky and L. S. Douch [MI).J. D. Campbell [26], showed that if
a specimen (12.7 mm diameter and 266.7 mm length) undergoes an impact
a

Stmin-per mi Strain-percent
FIG.6. Super-purity iron curves: (a) strain rate against strain for dynamic test; (b)
stress against strain, A. Static, B. Dynamic; numbers denote time in microseconds:
impact velocity 198 in./sec. ( J . Harding, E. 0. Wood and J. D. Campbell [76]).

Distance hrn impact end (cm]

FIG. 6. Distribution of permanent strain in specimen ( J , D. Campbell [26]).

with initial velocity of 400 inlsec, yielding occurs only in the neighbourhood
of end of the test piece under impact and is limited to about lOmm (Fig.6).
These tests showed that there is a strong dispersion due to rheologic effects.
The influence of the transversal motion on the phenomenon of propa-
gation of stress waves in a bar is discussed in a paper by H. Kolsky and
L. S. Douch [96].
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 251

The phenomenon of the delay time was studied in detail for mild steel
by J. E. Johnson, D. S. Wood, and D. S. Clark [87]. The delay time which
varies within the limits of 40 p e c to 1.5 msec is a function of the initial
impact stress varying from 80,000lb/ins to 5000 lb/in2, respectively. It was
also shown that for low carbon steel the delay effect in the dynamic com-
pressive test is of the same nature as in the static tensile test (Fig.7; cf.
also [28, 32, 38, 401).

o Cornpmsion impact
Rapid load tension
m-4 lo" 10-9 lo-' I
Delay time,sec
FIG.7. Stress versus log delay time, 73 "F (J. E. Johnson,D. S. Wood and D.S.Clark
~371).

The results of tests of D. B. C. Taylor [173]. showed that the delay

phenomenon can be observed not only for metals in the non-plastic state
but also in a partially plastic state.
In a paper by J. D. Campbell and C. J. Maiden [29], the influence of
previous dynamic loadings on the static properties of the material was
studied. In these experiments annealed medium-carbon steel specimens
were subjected to rapidly applied compressive loads maintained for periods
of the order of 10-4 sec. The applied stresses were between two and three
times the static upper yield stress of the steel and the permanent deforma-
tion varied from 1.2 percent down to very small amounts. Static stress-
strain curves were obtained by reloading the specimens in compression
immediately after impact. It appears that the upper yield stress can be
considerably reduced by the application of an impact stress of magnitude
262 PIOTR PERZYNA

and duration insufficient to cause appreciable permanent deformation

(see Fig. 8).
The influence of temperature on the dynamic behaviour of metals was
the object of tests of, among others, D. D. Sokolov [169, 1601, M. Manjoine
[112], J. F. Alder and V. A. Philips [l], J. M. Krafft, A. M. Sullivan, and

FIG. 8. Static stress-strain curves obtained after dynamic loading. Curve A : annealed
specimen (not pulsed). Curves B, C, D: specimens pulsed at 277, 293 and 308 in./sec.,
respectively. Curves E, F, G, H : specimens impacted at 326, 340, 362 and 366 in./sec.,
respectively ( J . D. Campbell and C. J . Maiden [29]).

C. F. Tipper [96], R. J. Mac Donald, R. L. Carlson, and W. T. Lankford

[60], C. J. Maiden and J. D. Campbell [log], and J. L. Chiddister and
L. E. Malvern [34]. It was observed that if the temperature is decreased
the rate-sensitivity of the material increases. This is illustrated by some
results taken from the work of C. J. Maiden and J. D. Campbell [log],
and collected in Table 1.
The influence of neutron irradiation on the dynamic properties of metals
has not yet been investigated in a satisfactory manner. Some results are
contained in the paper by J. D. Campbell and J. Harding. [31]. The influ-
ence of neutron irradiation on the dynamic properties seems to be of a more
complex nature, and more detailed conclusions are so far not possible.
The paper by J. D. Campbell and J. Harding has shown, however, that
these changes are very pronounced, and decisive for the behaviour of the
materials.
 FUNDAMENTAL PROBLEMS IN VI SCOPLASTICITY 263

TABLE1. THEP R O P E R T I E S OF A CARBON STEEL A T LOW T E M P E R A T U R E

(C. J. MAIDEN AND J. D. CAMPBELL[log])
~

Tem- Ratio of Per-

Dynamic Dynamic
pera- Time dynamic Maximum Average ma-
ture Impact to upper lower tostatic strain strain nent
of
velocity yield yield yield defor-
upper rate rate
test inlsec stress stress yield sec-1 sec-1
ma-
"C psec 10slb/ina 1O8lb/in~ stress tion
%

603 25 113 94 2.57 1220 670 5.2

+15 470 28 . 108 93 2.45 970 570 4.6
430 30 105 92 2.38 800 440 3.3

503 30 125 107 2.49 960 470 3.7

-41 470 35 119 105 2.42 700 380 2.9
430 45 116 104 2.36 660 280 1.8

-84 503 40 130 120 2.32 650 350 2.6

470 47 127 120 2.26 540 260 1.6

-121 503 54 141 135 2.13 470 200 0.9

3. Experimental Results for Soils

Only a limited number of measurements of plastic constants of soil

have been made under dynamic conditions, but recent experimental data
show clearly that soils exhibit rheological effects and are sensitive to the
change of the strain rate. Thus the characteristic feature of the behaviour
of soils, especially in their dynamic response, is the time-dependence of
the deformation process.
Of particular value are data discussed by P. Chadwick, A. D. Cox and
H. G . Hopkins [33] (see also A. W. Skempton and A. W. Bishop [167]).
These data are presented in Fig. 9 and show *at, as the rate of strain in-
creases from zero to about 1.6 sec-I, tbsfrengths of clays and sands increase
by factors of about 2 and 1.2, respectively. The authors of paper [33]
note that strain rate effects in soils are not well understood, but in clays
the resistance to the flow of pore water is probably one reason for their
existence.
The influence of the strain rate effects on the behaviour of soils have
also been investigated by N. A. Alexiev, Ch. A. Rakhmatulin and A. Y.
Sagamonian [2].
264 PIOTR PERZYNA

It is clear that in real soil the inelastic strain-rate tensor depends on the
time-history as well as on the path-history of the stress.

2-
/
/

a/’
0
co
.
€ ? - 0
’

if.-----
-

I
01 I I I I I I I I t
40-4 10-2 1 102 104
hte of shin (pemntqeper minute)
FIG.9. Relation between strength and rate of strain for (a) clay and (b) sand (dataof
D. W. Taylor, A. Casagrande and W. L. Shannon, from A. W. Skempton and A. W.
Bishop [167]).

4. Some Theoretical Ideas for the One-Dimensional Case

The difference in behaviour of a material during static and dynamic

loading is also of interest to theoreticians. They have tried, above all, to
describe the influence of the strain-rate on the yield limit of the material.
The general physical relation for the one-dimensional problem may be
represented in the form (cf. the papers of H. G. Hopkins [83-851)
(1.1) u = +(ef’, #),

where u is the nominal tensile stress and ef’ and if’are the nominal plastic
strain and strain rate, respectively. P. Ludwik [107], and H. Deutler
[68], on the basis of experimental investigations, and L. Prandtl [147],
on the basis of some theoretical considerations, proposed a logarithmic
+
expression for the function (see also A. L. Nadai [llS]).
In the analysis of the problem of propagation of plastic stress waves
in bars, the effect of strain rate was first taken into consideration by V. V.
Sokolovsky [161, 1621, and L. E. Malvern [110, 1111. The physical law
of L. E. Malvern can be expressed thus

(1.2) 6= w + (@[a- f(e)I),

where e denotes total nominal strain, E denotes Young’s modulus, and
u = /(a) is the static curve for simple tension or compression.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 266

The symbol ( 0 )is defined thus

L. E. Malvern discussed in detail two cases of the function @: the expo-

nential function

(1.4) G = a{exp b[o - f(43 - l}

and the linear function

(1.6) @ = c[o - f(e)l,

where a, b and c are constants depending on the material. Equation (1.6) had
been introduced earlier by V. V. Sokolovsky [161, 1621.
For the solution of the problem of propagation of plastic-stress waves
in a bar a more general physical relation was used by L. E. Malvern:

(1-6) d = ci/E + (g(0,e)).

An analysis of the physical relation (1.2) with (1.6) reveals easily that
an increase of the plastic strain rate proportional to the difference between

t in psec
- 8

--
I I
I"

x (inches)
FIG.10. Strain distribution in a long bar (S. Rajnak, F. Hauser and J. E. Dorn [ISO]).

the actual stress and the stress computed from the static curve is the assump-
tion of L. E. Malvern [110]. This difference produces the strain rate in
agreement with the viscosity law. The elastic strain component is considered
to be independent of the strain rate. In agreement with these assumptions
266 PIOTR PERZYNA

and in contrast with the results of the theory assuming the static charac-
teristic as a basis for dynamic considerations, the front of a plastic wave
propagates with the same velocity as the front of an elastic wave.
The theoretical results obtained by L. E. Malvern show that in this
strain-rate dependent theory there is no region of constant strain in the
neighbourhood of the end subjected to the impact load. I t is worthwhile
to observe that in many experimental investigations indeed no such region
I 1 I I I
1 I I I

x (inches)
FIG.11. Strain rate distribution in a long bar (S.Rajnak. F. Hauserand J.E.Dorn [lSO)).

was found (see for instance the tests by J. D. Campbell [26], Fig. 6). A
detailed discussion of this problem can be found in the paper by H. G.
Hopkins [83]. S. Rajnak, F. E. Hauser. and J. E. Dorn [150], examined
theoretically the strain distribution in a long bar subjected to dynamic
load. As a basis for their study they took the physical relation of L. E.
Malvern (1.6), the function g(a, 8 ) being determined from experimental
investigations with aluminium. Their results confirm the validity of the
assumptions of the theory that accounts for the influence of the strain rate.
Fig. 10 represents the strain distribution along the bar, computed for various
times. Similarly, Fig. 11 represents the distribution of the strain rate.
Fig. 12 represents a comparison of theoretical and experimental results.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 257

The experimental results obtained by H. Kolsky and L. S. Douch [95],

show, in comparison with the results of the theory disregarding the strain rate,
that permanent deformation at the end of the bar is in agreement with the
theoretically expected results. The theoretical predictions of strain distri-
bution were less satisfactory, and there was some indication in the results
for copper at low impact velocities that the simple theory needed modi-
fication in the manner suggested by L. E. Malvern.

x (inches)
FIG. 12. Final strain distribution in a long bar (S. Rajnak, F. Hauser and J . E. Dorn
[ 1501).

B. E. K. Alter and C. W.Curtis [3], carried out a number of tests with

the aim to determine how impulses producing plastic strain propagate
along cylindrical lead bars. These tests have shown considerable influence
of the strain rate on the propagation of plastic strain. They are also in
agreement with the assumptions of L. E. Malvern’s nonlinear theory.
E. J. Sternglas and D. A. Stuart [166], who tested copper specimens,
showed that stress waves produced by an increase of dynamic load super-
posed on a static load at the yield limit propagate with the elastic velocity.
A similar phenomenon was observed by J. F. Bell [lo] with steel specimens.
268 PIOTR PERZYNA

These observations are contradicted, however, by recent tests carried

out by H. Kolsky and L. S. Douch [W],who point out the fact (cf. also
H. G. Hopkins [83-86]) that the load conditions in [lo] and [166] were
not purely dynamic but partially static.
Most experimental results show clearly that the adoption of the static
tension curve as a basis for the description of dynamic phenomena leads,
in general, to erroneous results. This does not mean that the influence of
the strain rate is equally important for every material. There are many
materials for which this phenomenon has not been observed to be essential,
and the assumptions of Khrmhn, Taylor, and Rakhmatulin are thus justified.
Many special tests concerned with this problem could be quoted. In a series
of tests for heat-treated aluminium it has been shown by J. F. Bell [ll-171,
who used an original test method based on diffraction, that the results
obtained are in good agreement with the assumptions of the strain-rate
independent theory. A similar conclusion has been obtained by H. Kolsky
and L. S. Douch [%I, for an aluminium alloy.
The physical relations (1.1)-(1.6) have purely phenomenological char-
acter. I t will be of interest to analyse them in the light of the dislocation
theory of crystalline materials. Much recent theoretical and experimental
work is concerned with the detailed investigation of this problem.
The idea of explaining the phenomenon of plastic flow of metals on the
basis of dislocation theory has been communicated by A. M. Cottrell and
B. A. Bilby [48]. General foundations for the explanation of many phe-
nomena during dynamic loading processes of metals are given in a later
paper by A. M. Cottrell, [49]. Generalizing this idea J. D. Campbell [26,97]
has shown that the phenomenon of delay before yielding can be explained
satisfactorily by dislocation theory.
Although this theory has not yet been developed sufficiently to ensure
an adequate basis for the derivation of physical relations describing the
dynamic behaviour of a material, it has been shown by J. A. Simmons,
F. Hauser, and J. E. Dorn [166], that from the viewpoint of dislocation
theory the physical relation (1.1) may be written as

(1.7) +
dB = f(a,s,l)do g(o,E,B)dE + h(a,s,i)dt,
where the influence of strain acceleration is taken into account in addition
to that of strain rate.
Another interesting question is : what is the interpretation of the existing
physical relations in the light of the dislocation theory. This is the subject
matter of a paper by J. A. Simmons, F. Hauser, and J. E. Dorn [166].
A study by means of the dislocation scheme of the classical theory of K k r m h ,
Taylor, and Rakhmatulin and the viscoplastic theory in which the influence
of the strain rate is taken into consideration has shown the advantages of
the viscoplastic theory. It is hoped that some of the experimental tests
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 259

now being carried out and a tentative interpretation of their results by

means of the dislocation theory will lay the foundations for the derivation
of general physical relations which describe the dynamic behaviour of
plastic materials.
The behaviour of Frank-Read dislocation sources under the action of
impact stress was studied by J. D.Campbell, J. A. Simmons, and J. E. Dorn
[30]. This analysis has shown that if the initial accelerating part of the
dynamic process of dislocation can be disregarded, the nonlinear theory of
L. E. Malvern can be adopted, at least formally, as a basis for the description
of the behaviour of crystalline materials.
A particular form of the relations (1.7) has recently been analysed by
J. Lubliner, [106]. Taking as a basis the equation

he proposed the following modification

(1.9)

It is evident that these modified physical relations become, in partic-

ular cases, those of the Karmin-Taylor-Rakhmatulin or of the Sokolovsky-
Malvern theory.
Physical relations similar to (1.9) were also the object of the investi-
gations by N. Cristescu [66, 671 who used them in problems of propagation
of stress waves in thin strings.

6. Assumptions and Definitions

We shall make a distinction between an elastic-viscoplastic material

and an elastic/viscoplastic material. Elastic-viscoplastic we shall call a
material showing viscous properties in both the elastic and plastic regions.
The term eZustic/viscoplustic will be reserved for materials showing viscous
properties in the plastic region only.
A similar distinction was introduced by P. M. Naghdi and S. A. Murch
in [llS]. The notion of an elastic/viscoplastic material is evidently an
idealization that simplifies the argument considerably. To see this consider
the problem of choice of an adequate yield criterion. The determination of
the yield condition for an elastic-viscoplastic material is very difficult and
has not yet been done. The assumption of the elastic/viscoplastic scheme
solves this problem at once, because the initial yield condition can remain
260 PIOTR PERZYNA

the same as in flow theory. Difficulties are encountered only if the varia-
bility of flow surfaces in the course of a deformation process of an elastic/
viscoplastic body is considered.
The distinction just introduced requires that the considerations con-
cerning the constitutive equations should be subdivided into two parts.
Thus, methods of description of elastic-viscoplastic materials will be dis-
cussed first in Secs. 2 to 6. Sections7 to 13 will be concerned with elastic/visco-
plastic materials and with the application of that scheme to the description
of the dynamic properties of plastic materials.
It will be assumed throughout the entire work that the displacement
gradients in the straining processes under consideration are infinitely small
(cf. the definition in the monograph by C. Truesdell and R. Toupin [183]).

11. CONSTITUTIVEEQUATIONS

1. The Behaviow of a Material i n the Viscoelastic Regiolz

Let us denote by cii the strain tensor and by uij the stress tensor in
Cartesian coordinates. For an elastic-viscoplastic body it will be assumed
that the strain tensor can be represented in the form of the sum

where E$, e,; 4j denote the elastic, viscous, and plastic strain components,
respectively.
Let us emphasize the fact that the equation (2.1) is not satisfied in
general. Its introduction will be treated as a fundamental simplifying
assumption.
The stress and strain deviator tensors used in what follows are defined
thus :

where dij is Kronecker's delta.

The components of the strain-rate and stress-rate tensor will be denoted
by dii and aij, respectively.
Problems of linear theory of viscoelasticity have been extensively treated
in numerous monographs and,,survey papers. Let us quote several recent
fundamental investigations giving a systematic analysis of physical relations
between the stress tensor and the strain tensor for viscoelastic bodies, within
the frame of the linear theory. These are the papers by D. R. Bland [18],
B. D. Coleman and W. No11 [42], M. E. Gurtin and E. Sternberg [73],
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 261

A. C. Eringen [68], and W.Noll and C. Truesdell [122]. They also contain
extensive lists of literature.
With the notations of [73] the physical relations between the tensors of
strain and stress in the viscoelastic region read

where f l and j zdenote the distortion and the bulk expansioncreep-functions,

respectively. The notations sij(x,t) and soi i ( %) stand for sii(xk,t), sii(xh,O +),
respectively.
The physical relations (2.4) and (2.6) express the dependence of the state
of stress at the time t on the history of that state in the entire time interval
in which the phenomenon takes place, that is from 0 to t .
The authors of [73] have shown that the initial elastic state of a visco-
elastic body obeying (2.4) and (2.6)is described by the generalized Hooke’s
law
e l 1
e.. - -s*, t? = -s,
” - 2p 3K

where p and K are the elastic constants of the material (shear and bulk
modulus).
Making use of this observation, the viscous strain components can be
separated off in a simple manner. From (2.4). (2.6) we find (cf. P. M.Naghdi
and S. A. Murch [llS])

t
262 PIOTR PERZYNA

2 . The Yield Criteriolz for a Viscoelastic Material

In order to explain the differences between the establishment of the

plastic state in an elastic body and in a viscoelastic body, let us consider the
load path in the nine-dimensional space of stresses (Fig. 13).

A2

FIG.13. Load path and yield surface for viscoelastic material.

In the elastic case the plastic state produced will be represented by the
same point independently of the time in which the state represented by A is
reached, provided that the load path is the same. If the material is visco-
elastic the plastic state may be reached at different points A, or 4, say,
depending on the time in which the load path is through. This difference
is caused by the viscosity of the material and the dependence of the load
history on time. It is obvious that by passing through the path OA in the
same overall time but with different strain-rates at the same points of the path,
different yield limits will be obtained.
In order to describe the complicated problem of a viscoelastic material
becoming plastic the notion of flow wrface has been introduced by P. M.
Naghdi and S.A. Murch [118]:

(2.9) f = f (q,&GP) *

The elastic-viscoplastic state is determined by the condition f = 0, while

the viscoelastic states correspond to the condition f < 0. The function f
depends on the state of stress q ,the state of plastic strain &, the parameter
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 263

= / I ( E ~ ; ~which
) represents viscous effects and the strain-hardening para-
meter K. The latter depends, in turn, on the plastic strain and is defined in
the same manner as in the theory of plastic flow describing isotropic strain-
hardening of the material (cf. the definition (2.44)).
Let us now consider the time-Variability of a flow surface. The time-
derivative of the function f is

(2.10)

If the state under consideration is elastic-viscoplastic and undergoes a

change such that f < 0, this change leads to a viscoelastic state because
f+ idt gives a new value of f which lies below zero. Such a change of the
state of stress will be called an unloading process. During such a process
there is no increase of plastic strain, therefore = 0 which involves k = 0
because no change of the strain-hardening parameter can take place if
l$= 0.
Since 1can be expressed by means of the physical relations (2.7) and
(2.8) as a function of the stress rate dij, we can introduce

(2.11)

The mathematical condition for unloading can now be written as

(2.12) f = 0, S(6ij) < 0.
A change of state of stress from one elastic-viscoplastic state to another
elastic-viscoplastic state accompanied by no increase of plastic strain will
be called a neutral process. A neutral state is characterized by
(2.13) f = 0, U(&j)= 0.
An active loading process, which is accompanied by an increase of plastic
strain, takes place if
(2.14) f = 0, S(6ij) > 0.
By considering the flow surface in the stress space we observe easily
that a neutral process does not correspond to the direction of the stress
increment 6&t which is tangential to the flow surface at the point con-
sidered. This is different from flow theory.
A neutral state will now be realized if the vector of stress increase ciijdt
deviates from the direction normal to the flow surface by the angle 6, where

(2.16)
264 PIOTR PERZYNA

The expression (2.16) shows why the load criteria for the viscoplastic
case are changed, if compared to those known from the classical theory of
plasticity; the influence of the viscous properties of the material on the
yield condition is here responsible. There is also an influence of the strain-
rate on the yield condition, of which a detailed discussion will be given
later.
M. Reiner [152] discussed the problem of attainment of the plastic state
with a constant load, that is, the possibility of a material becoming plastic
in a creep process.
An original analysis of yield criteria or, more generally, criteria of ob-
taining a critical state was proposed by W. Olszak and 2. Bychawski [124,
1261. This idea is based on the observation that the fracture of some visco-
elastic bodies depends not only on the elastic energy but also on the rate a t
which the dissipation energy varies. Examples of bodies were analysed
where destruction occurs if the rate of variability of the dissipation energy
reaches a certain value. For such bodies the strain rate is decisive. For an
analysis of a sufficiently large class of real bodies the authors of [126] propose
the following condition for reaching the critical state* :

(2.17) O E W = WE, Wuw = WD,

where W Eis the reduced elastic distortion energy and W , the rate of change
of the dissipation energy expressed in the dimension of energy. The quan-
tities w ~ , w E and k are certain constants characterizing the material.
In [126] several models are considered, the form of the criterion (2.16)
being analysed in detail.

3. A Definition of a Stable Inelastic Material

It appears that a general approach to describing the properties of an
elastic-viscoplastic material is given by the ideas of D. C. Drucker set forth
in [62, 641.
D. C. Drucker uses certain observations in the domain of the theory of
plasticity and points out: If we want to give a uniform and unambiguous
description of the viscous and plastic properties of a material, we must
introduce certain additional postulates which actually are limitations intro-
duced in a judicious manner.

* It should be observed, however, that the attainment of the critical state is under-
stood in [125] in a more general manner than the attainment of the plastic state. The
plastic state may constitute a particular case of the critical state.
 FUNDAMENTAL PROBLEMS IN VISCOPLA STICITY 266

By introducing the postulate of stability of the material we can obtain

fundamental conditions the satisfaction of which permits a consistent
derivation of the physical relations. These conditions restrict the consider-
ations to a certain class of materials but, at the same time, allow a consistent
mathematical description of that class.
Consider a body of volume V bounded by a regular surface S subject to
surface tractions Ti and mass forces Piwhich are functions of time. The
states of displacement ui, strain E , ~ , and stress aijproduced by these boundary
conditions are also functions of time. Assume now that the boundary con-
ditions undergo certain variations, determined by the surface tractions
+ +
Ti ATi and the mass forces Pi AP, to which correspond: the state
+ +
of displacement u, Au,, the state of strain .sii A.sij, and the state of
+
stress aij Aaij. The definition of a stable inelastic material (elastic-visco-
plastic) is furnished by the following postulate of Drucker:
The work performed by the increments of the external forces on the
corresponding increments of the components of the displacement vector
must be non-negative.
This may be expressed as follows:

where t = 0 is the instant at which the external load increments are applied.
It is often more convenient to introduce two loading paths T i l ) , P i l )
and Ti2),Pie), which become different after t = 0 (Fig. 14). Then (2.18)
can be written in the form
(2.19)

If in the expressions (2.18), (2.19) the choice of t k is limited by no addi-

tional condition and can be assumed to be sufficiently large, we are concerned
with “stability in large.” If t k must be close to t = 0 we have the mathe-
matical expression of “stability in small.’’
Using the principle of virtual work, the surface and mass forces and the
velocity can be replaced by the stresses and the strain rates. The principle
of virtual work asserts that for any continuous velocity ti+ we have

(2.20)
S V V
266 PIOTR PERZYNA

where Ti,Pi, aij represent a set of mechanical quantities in equilibrium

and tii, d, constitute a consistent set of kinematic quantities. It should be
stressed that the two systems of mechanical and kinematic quantities are
not interdependent.

f’ fk

FIG.14. Two paths of loading. Arrows join points of the same time.

Assuming that only homogeneous states of stress and strain are con-
sidered, we obtain from (2.19), on the basis of (2.20), the condition

(2.21)
i‘
0
[aij
(2) - a;{(1)3 [its’- 49at 2 0.

4. Convexity of the Flow Surface

Convexity of the flow surface should be assumed in the theory of elastic-

viscoplastic bodies, since it is required by condition (2.21) which follows
from the postulate of Drucker.
Assume that the state a$’ is a steady state a; at the instant t = 0 and
that the state a!;) is time-variable; denote it by aii. Now consider the fol-
lowing closed load cycle. At t = 0 the state of load aij coincides with the
steady state a;. Next, aii varies along the path M&, (Fig. 15) reaching
at t = t, the point M, which represents a plastic state. Increments of plastic
strain occur on the path M,M,. The state M, is reached at t = ta. Starting
from t, unloading takes place along the path M a M o . At t = th the state
again coincides with the initial state, thus aij = a;.
The condition (2.21) for the closed cycle MoM1M&f0 in the time interval

i *
0 t 4 tk yields

(2.22) (aij - aij)(di, *

- i i j ) dt 2 0,
0

and 6: replace .@ and

where iij .@, respectively.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 267

Bearing in mind that 12: = 0 and 2.. *I = i$’”’ +

i$, the expression
(2.22) can be written in the form (cf. P. M. Naghdi and S. A. Murch [llS]):

(2.23)

(2.24)

Expanding the first term of (2.23) in a Taylor series a t the point t = t, we

obtain in O(dt)

If we assume At = tz - t, sufficiently small (i.e. restricting ourselves to

“stability in the small”) the inequality (2.26) implies the inequality (2.23).

FIG. 15. Closed loading cycle in stress space.

In order to discuss the problem of flow-surface convexity completely we

shall introduce, in agreement with P. M. Naghdi and S. A. Murch [118],
the notion of rapid load path and instantaneous loading surface. Rapid
load path in the nine-dimensional stress space is a finite process of stress
change realized in an infinitely short time, with the parameter P(&il)remaining
unchanged, i.e.

(2.28) lim dP(&k)= 0.

At -0

For a rapid load path the behaviour of the flow surface is exactly the
same as in the inviscid theory of flow for an ordinary path.
268 PIOTR PERZYNA

Instantaneous loading surface will be defined now. Let us assume that

at a given time t, 2 0 we have a state characterized by the variables a$),
E $ ( ' ) , K ( ~ ) , ~ (such
~) that the body is viscoelastic, i.e.

(2.27) f(a~',E~I"',P"',K"') < 0.

The equation

(2.28) fa = f(aii,&~I"),P('),K("))= 0

is the instantaneous loading surface corresponding to the state which in

the stress space is represented by point A (Fig. 16). I t should be observed

c----

-0

FIG. 18. Direction of the plastic strain-rate vector.

that the state A not only depends on the components of the stress tensor
but also on the quantities E $ @ ) , , ~ ( ~ ) , K (which
~), describe the history of the
loading process. The latter is characterized by the load path and is followed
by the varying state of the body at the time at which the point A is reached.
Let us now assume that a loading process takes place a t t > t, accompanied
for t , > t, by an increase of plastic strain ,4$ dt; unloading then starts so
that the stress state returns to the initial state a t t = tk. The application
of the inequality (2.26) to the load cycle at the instant t, t. <
t - t,, gives

For any rapid load cycle the flow surface is on the basis of (2.26) identical
with the instantaneous flow surface
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 269

(2.30) lim /(aii,ei?), !(a), ,(a)) =fa.

('I - 'J -+Ll

If it is assumed that. the quantity {Y}; is a continuous function a t the

point t = ta, then, with the notation
'k

(2.31)

- -
'a

it can easily be shown that in the limit t k -+ ta, At 0, dp 0

(2.32)

Thus the inequality (2.29) for a rapid load path can be assumed in the same
form as in the inviscid theory of plastic flow. In particular, if ta is assumed
to be equal to zero and a$) is treated as an arbitrary point in the visco-
elastic region, identical with a,; we have

(2.33) (Uii - .;)it 2 0.

For rapid load paths the direction of 6; is fixed and independent of the
time. By considering every possible rapid load path from an arbitrary point
0; in the viscoelastic region to the point ai at the instantaneous flow surface
f a = 0 we can prove, on the basis of the inequalities (2.33), the convexity
of the instantaneous flow surface.
The direction of 4. is, for every rapid load path, orthogonal to the in-
stantaneous flow surface.
The above proof of convexity of the instantaneous flow surface is valid
for every point and every path in the stress space, hence the flow surface
f = 0 must also be convex. This leads to the conclusion that the convexity
of the flow surface depends on neither the load path nor the time. But
there is no such conclusion for the orthogonality of the vector which is
orthogonal to the flow surface for rapid load paths only. For a real loading
process the direction of the vector i$ remains an open question.

6. The Direction of the Vector of Plastic Strain Rate

Let us again consider the load path in the stress space and assume that
the instantaneous stress point moves along the same path with various
velocities. The duration of each loading process will be different, and each
will involve different rheologic effects. The plastic state will therefore be
270 PIOTR PERZYNA

realized at various points of the path and determined by different flow

surfaces. As a result of the various rheologic effects for the same load path
we obtain a family of flow surfaces depending on the parameter p.
In order to simplify the considerations assume that all flow surfaces of
the family under consideration pass through the same point, for instance A
(Fig. 16). Then it is easily observed that the actual direction of the vector
of plastic strain rate may deviate from the direction normal to the instan-
taneous flow surface and is contained within a cone whose apex angle depends
on {YC.
Making use of (2.26) in the case of Fig. 16 and assuming that we have
uii = 0; for t = 0, we infer that the maximum apex angle of the cone is
a function of {Y>”,”.
The influence of rheologic effects on the variability of the flow surface
leads to a certain indeterminacy, the position of the instantaneous flow
surface in the stress space being unknown as well as the point, at which the
plastic state is attained. The direction of the hyperplane tangential to the
instantaneous flow surface a t the point considered is not uniquely deter-
mined. These problems are fully discussed in a paper by P. N. Naghdi and
S. A. Murch [118].
Certain conclusions have also been drawn by these authors in the case
of inviscid plasticity which admits, however, the existence of irregular flow
surfaces. In such a theory the vector of the plastic strain rate 4.$ lies in the
fan formed by the normals to the smooth surfaces which constitute the
”plastic corner.” If an experimenter observes that the vector of plastic
strain rate has a direction different from the direction normal to the expected
flow surface or that this direction varies within certain limits, he will con-
clude, on the basis of inviscid plasticity, that a “plastic corner” exists since
by viscoplasticity the flow surface should be regular.

6. Constitutive Equations for an Elastic-Viscoplastic Material

Assuming that the vector of plastic strain rate .4$ is directed along the
outer normal t o the instantaneous flow surface f = 0, P. M. Naghdi and
S. A. Murch [118] proposed the relation

(2.34)

Determining A from the condition = 0, we obtain

 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 271

where

(2.36)

On introducing the notation

(2.37)

and bearing in mind the load criteria we obtain

(2.38) i$ = l o a/
if f < 0,

Ih(B(akl)) aai, if f = 0,

where ( x ) is defined thus:

0 if x 5 0,
(2.39)
x if x >O.

In the equations (2.38) just derived the difficulty mentioned earlier,

which results from the indeterminacy of direction of 4.
in relation to the
actual flow surface is not accounted for. To correct this we introduce the
following artifice. The loading surface f is first replaced by a characteristic
yield condition fo , for which hardening is assumed to take place isotropically.
A second loading function g is then introduced, which differs from fo in
such a way that bounds on the direction of 8,are implied by the constitutive
equations. T h e constitutive equation (2.38) is then replaced by

(2.40)

where L and M are scalars dependent on {Y>b”.

We shall generalize the constitutive equations (2.38) in a different way
by introducing two loading functions h and f and the following relations:

(2.41)

The function h plays now the role of the viscoplastic potential, and the
function f represents the yield criterion for the elastic-viscoplastic material.
In order to determine the coefficient N and the connection between functions
h and f certain conditions must be imposed on the constitutive equations
(2.41).
A possibility seems to exist to develop the theory of viscoplasticity in
a broader sense than that for a stable inelastic material. To do that, an
272 PIOTR PERZYNA

idea given first by B. D. Coleman and W. No11 [41]may be used. I t leads

to constitutive inequalities or thermostatic and thermodynamic inequalities
which ensure a physically natural response. This idea was extended by
C. Truesdell [184],B. D. Coleman [a],R. A. Toupin and B. Bernstein [182],
C. Truesdell and R. A. Toupin [la], L. ‘E. Bragg and B. D. Coleman [21]
and B. D.Coleman and W. No11 [44](see also W.No11 and C. Truesdell [122]).
Methods describing the behaviour of elastic-viscoplastic bodies are not
a means by which we could solve practically important problems. This is
a very real general shortcoming and its correction will require considerable
work. Many questions remain at present unanswered; some of them have
only been touched. The merit of the general studies lies in their considerable
cognitive value. Essential difficulties are indicated and a procedure that is
most likely generally correct and consistent is developed. These studies
unquestionably succeeded in the establishment of general foundations for
the solution of problems of viscoplasticity in a broad sense.
Among the basic problems to be solved we mention, above all, detailed
discussion and study of the yield criterion for viscoelastic bodies, the estab-
lishment of the function representing the yield condition, and the deter-
mination of how it depends on rheologic effects. The latter problem is
connected with the variability of the flow surface, depending on the duration
of the loading process. Its solution will remove the indeterminacy mentioned
above. Other problems are: the full explanation and discussion of the ortho-
gonality of the vector of plastic strain rate, and the study of the character
of the constitutive equations in order to approximate better real materials.
One of the fundamental problems is of course whether the fundamental
assumption expressed by (1.1) is at all valid.
Let us now mention the problems of generalization of considerations and
results already obtained.
Generalization to-or a new study assuming-finite strain and also
extension t o the case of non-isothermal processes is desirable. No full fun-
damental study of the thermodynamics of elastic-viscoplastic deformations
has as yet been done. This would enable us to base all phenomenological
considerations on more accurate physical foundations. The question of unique-
ness of the solution of the fundamental boundary-value problem of viscopla-
sticity remains also open. No variational theorems that would permit
systematic application of approximate methods have as yet been demonstrated.

7. Cmstitlttive Equations for Rate-Sensitive Plastic Materids

Studies in this domain were initiated earlier than those described in the
foregoing section. The general foundations for the study of viscoplastic
problems were laid by K. Hohenemser and W. Prager [80] already in 1932
(see also W.Prager [142,1481).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 273

This work was not fully appreciated for a long time. It was only in the
years 1948-1960 that it attracted attention subsequent to the observations
of V. V. Sokolovsky [162], and later of L. E. Malvern [110], who showed
that the assumption of K. Hohenemser and W. Prager may be used as a
basis for description of certain dynamic properties of rate-sensitive materials.
This concerned, however, one-dimensional problems only (see Section I).
Further development of the idea of K. Hohenemser and W. Prager is
contained in the references [132-1341.
Owing to the assumption that the viscous properties of the materials
become manifest only after the passage to the plastic state and that these
properties are not essential in elastic regions, the basic concepts of the
description of viscoplastic properties differ from the methods explained
before. It will be assumed that the strain rate can be resolved into an elastic
and inelastic part

(2.42) j 6; + 6;.
ii=
The inelastic part of the strain rate, which is denoted by ig., represents
combined viscous and plastic effects.
Since the material has no viscous properties in the elastic region, the
choice of an adequate yield criterion will be much simpler than in the case
of an elastic-viscoplastic material. The initial yield condition, which will
be called the static yield criterion, will not differ from the known condition
of the inviscid theory of plasticity.
In order to keep our considerations sufficiently general we introduce a
static yield function in the form

(2.43)

where the function f(ui,&) depends on the state of stress uij and the state
of plastic strain4. The parameter K is defined by the expression

This quantity is called the strain hardening parameter (detailed discussion

of the two fundamental definitions of that parameter may be found in
R. Hill's book [79], or in a paper by P. M. Naghdi [117]).
The flow surface F = 0, which is again in the nine-dimensional stress
space, is assumed regular and convex, and we propose the constitutive
equations for work-hardening and rate-sensitive plastic materials in the
following form [134] :
214 PIOTR PERZYNA

i.'j - -1 &j 1 --
+- 2Y s + yO(@(F))-
&j
aF I

2P E a0,
where the symbol (@(F))is defined as follows:

The function @(F) may be chosen to represent the results of tests on the
behavio~rof metals under dynamic loading.
It will be much more convenient to write the constitutive equations
(2.46) in the slightly different form

(2.47)

where y = ~ O / Kdenotes a viscosity constant of the material.

The relations (2.47) involve the assumption that the rate of increase of
the inelastic components of the strain tensor is a function of the excess
stresses above the static yield criterion.
This function of stresses above the static yield criterion generates the
inelastic strain rate according to a viscosity law of the Maxwell type. The
elastic components of the strain tensor are considered to be independent of
the strain rate.
The constitutive equations (2.47) describe the work-hardening effect of
the material. Owing to the introduction of the function F,it is possible to
study the anisotropic and the isotropic work-hardening for a broad class of
compressible materials.
To discuss the constitutive equations more accurately let us consider the
inelastic part of the relations (2.47)

(2.48)

Squaring both sides of ( 2 4 , and denoting by I 2 = (1/2)$$ the invariant

of the inelastic strain-rate tensor, we obtain

(2.49)

According to (2.49) we have

 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 276

This expression implicitly represents the dynamical yield condition for

elastic/viscoplastic, work-hardening materials, and describes the dependence
of the yield criterion on the strain rate.
It can be seen from (2.48) and (2.60) that the inelastic strain-rate tensor
considered as a vector in the nine-dimensional stress space is always directed
along the normal to the subsequent dynamic loading surface (Fig. 17).

[Subsequent dynamic !w

F-O(Jnitia1 yield surF-1

FIG. 17. Dynamic loading surface and strain-rate vector.

The change of the yield surface during the deformation process is caused
by isotropic and anisotropic work-hardening effects, and by the influence of
the strain-rate effect.

8. Special Cases of Constitzctive Equations

We shall first study the elastic/viscoplastic material with isotropic

work-hardening. Let us aSsume the function F in the form:

(2.61)

where the function f(uH)depends now on the state of stress only. Thus

(2.62) f h j ) = fWiJwJa),

where J< denotes the first invariant of the stress tensor udf;Ja and Js are
the second and third invariants of the stress deviator sii. respectively. We can
write
276 PIOTR PERZYNA

where
(2.64) tij= sighj - #Ja &j.

In this case the constitutive equations (2.48) take form

(2.66) &j = A(J:t J g , J S , K ) &j + B(J;Jp,Js,~)sij + C(J;IJa,JalK)tij,

where the following notations have been introduced

(2.66)

A . Application to Metds. Assuming plastic incompressibility, i.e.,

A(J:,Js,Js,~)= 0 or af/aJ1' = 0, the general constitutive equations for
strain-rate sensitive metals may be written as follows:

(2.67) ifj = B*(Ja,Ja,K)Sij + C*(Ja,Ja,K)tij.

If we then assume the Huber-Mises yield condition, i.e., that the function
f(c7q) = (Ja)"', we obtain by (2.47)*:

According to (2.50). the dynamical yield criterion has the form:

(2.69)

For one-dimensional states the relations (2.68) furnish the strain-rate

law

A similar case of constitutive equations has been recently studied by S. Kaliski

[as]. The difference between Eqs. (2.68) and those of S. Kaliski lies in the definition
of the work-hardening parameter K.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 277

where

(2.61) Y * = (2/1/3)Yt w? VWP).

=
Relation (2.60) was first introduced by Malvern [110].
The expression (2.69) now gives

(2.62) u = #?) [+
1 m-1(;)] ,

where u = +(@) represents the static stress-strain relation for simple tension.

FIG. 18. Dynamic stress-strain cur-

ves for work-hardening and strain-
rate sensitive material, i = const.

FIG.19. Dynamic stress-strain curve

for work-hardening and strain-rate
sensitive material, 3 = &(e).

Equation (2.62) describes the dynamic stressstrain relation in the plastic

range for simple tension. The result of this relation is plotted for B = const
in Fig. 18, and for the more realistic case B = 2 ( ~in
) Fig. 19.
As a second case, we shall examine the elastic/visco-(perfectly plastic)
material. We obtain this case by introducing the assumption that the
function I; does not depend on the strain, i.e.

(2.63) F=--W J ~ )
1, c = const.
C
278 PIOTR PERZYNA

The constitutive equations (2.47) can now be written as:

where y = yO/c, as first introduced in [132].

According to (2.60), we can write:

By assuming

(2.66)

FIG.20. The dependence of E o n m.

where k is the yield stress in simple shear, we have the constitutive equations

(41.
and the dynamical yield criterion

(2.68) = k [1+ 0-1 VIgP '

The dependence of
Qi in Fig, 20.
n9 l/v
on as given by (2.68) is plotted for some function

It can be seen from (2.68) that the constitutive equations (2.67) lead
to a similar result as the inviscid theory of plasticity for isotropic work-
hardening material. But in the inviscid theory of plasticity for isotropic
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 279

work-hardening, the radius R of the cylindrical yield-locus in stress space

depends on the plastic strain, whereas in the case considered here R depends
on the inelastic strain rate, according to

(2.69) = R.[ 1 +@-I m

(T)] 8

where R, = Kv? denotes the radius of the static yield cylinder (Fig. 21).
In the inviscid theory of plasticity for isotropic work-hardening material
we have three possibilities according to whether Jg > O (loading), J2 = 0
(neutral loading), or j g < 0 (unloading).

FIG. 21. The cylindrical yield locus for elasticlvisco-(perfectly plastic) material.

Since in viscoplasticity J 8 is a function of the strain rate, the plastic

flow (relaxation effect) occurs if J 2 >ha, independent of j , $0. Owing to
the relations (2.67)the material will be elastic on the path OP,,but on the
path P,$'iP,'Pg it will be elastic/viscoplastic (Fig. 21).
For one-dimensional states, (2.67)and (2.68)lead to the relations
(2.70) d =d/E + Y*(@(u/u~
- I)),
(2.71)

respectively. Law (2.70)describes the dynamic stress-strain relation in simple

tension for an elastic/visco-(perfectly plastic) material. The result of (2.71)
is plotted for S = const in Fig. 22 and for S = C(E) in Fig. 23.
280 PIOTR PERZYNA

When the elastic deformations are negligible in comparison with the

plastic deformations, then for a linear function @(F) (2.67) furnishes the
constitutive equations first given by Hohenemser and Prager [SO].

“t
i
I f 1
c-const

FIG. 22. Dynamic stress-strain curves for elastic/(visco-perfectlyplastic) material,

i = const.

Prager [la61 showed that the constitutive equations of the viscoplastic

and perfectly plastic materials have the same relations to each other as the
constitutive equations of the viscous and perfect fluids.

c
&

FIG. 23. Dynamic stress-strain curve for elastic/(visco-perfectly plastic) material.

i = i(e).

It may be interesting to see that the constitutive equations of the flow

theory result as a limiting case from the general constitutive equations of
the elastic/viscoplasticmaterial. With this object let us assume that y + 00.
According to (2.50) this means that

(2.72) /(a&) =K or F = 0.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 281

From (2.72) and by definition (2.46) of the function O(F),O(F)+ O in

the limiting case y --* 00; thus y@(F) = A becomes indeterminate. In this
case, we have

(2.73)

It is possible to determine the parameter A by satisfying the static

yield criterion
(2.74) F(aij,&) = 0.
Since the subsequent static loading surface (2.74) passes through the loading
point, and since during loading

(2.76)

we have (see, eg. [117]) by (2.73)

(2.76)

I ae,acr,,
In the special case of elastic/visco-(perfectly plastic) material we obtain
by putting the result (2.68) into the relations (2.67)

(2.77)

By setting y = 00 in (2.77), we obtain the constitutive equations

(2.78)
of the incompressible perfectly plastic material [146]. These constitutive
equations hold only when the rate of deformation does not vanish.
B . A$plicatiort to Soils. As a particular case of the constitutive equations
(2.66), we shall study the elastic/visco-(perfectly plastic) soil for the following
static yield function*:

(2.79) F=
aJi + J;'' - 1,
k
where a is a constant describing the dilatation rate of the soil.

* For the discussion of the other cases see papers [127, 1281.
282 PIOTR PERZYNA

Here the constitutive equations (2.47) give

(2.80)

The rate of cubical dilatation takes the following form

(2.81)

Equation (2.81) shows that the inelastic deformation is accompanied by an

increase of volume, if a # 0. This property is known as dilatancy.
The dynamic yield condition following from (2.80) has the form

In the limiting case, y + oo, we obtain from (2.80) the known consti-
tutive equations for an elastic-perfectly plastic soil (see D. C. Drucker and
W. Prager [201]):

(2.83)

where

The plastic rate of cubical dilatation is then expressed by the relation

(2JW & = 3d.

9. The Dynamical Conditiolt /or a Stable ElasticlVisco$lastic Material

A definition of stable inelastic materials with inclusion of dynamic terms

was introduced by Drucker, [62] ; it leads to the following condition :
(2.86)
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 283

where the stress tensors u$) and ui:) and the corresponding strain-rate
tensors ij;) and St,)refer to two distinct mechanical paths. The first integral
is extended over a finite interval of time 0 - t - I,tk, the second term is the
value of the volume integral at tk minus the value at the instant of the
divergence of paths, t = 0. Necessarily, in all cases the velocity &@) through-
out the volume is the same as @) at t = 0,hence the second term of (2.88)
is positive or zero at all times tk. Therefore, the inequality

is much more significant (compare with (2.19)). It is obvious that the in-
clusion of dynamic terms does not affect the conclusions concerning the
stress-strain relations to be drawn as a result of neglecting these terms.
From the inequality (2.87),we reobtain the condition of orthogonality
of the inelastic strain-rate vector to the yield surface and that of convexity
of the subsequent dynamic loading surface. Both were assumed as basis
for introducing the general constitutive equations (2.47).

10. Comparison with Experimerttal Data

Let us introduce the following special types of function O(F) (see [132,
1331) :
@(F)= Fd, @(F)= F, @(F)= e x p F - 1,
(2.88) N N
O(F)= 2 Aa[expFa-
a=l
11, @(F) = 2 BaFa.
a-1

When the elastic deformations are negligible in comparison with the

plastic deformations, then we obtain by (2.70)and (2.88,)for a one-dimen-
sional state of stress the equation

(2.89)

The relation (2.89)is equivalent to the Cowper-Symonds-Bodner strain-rate

law [19, 661. In the case (2.88,) the constitutive equations (2.67)are equiv-
alent to Freudenthal’s relations [69].
For the one-dimensional case we have by (2.70)and (2.88,)

S=d/E+y*
[exp:(-- l)-lID
which, in slightly different form, was introduced by Malvern [llO].
204 PIOTR PERZYNA

We shall examine the types of function @(F) (2.88) in the light of exper-
imental data. At first we shall try to determine the constants y*, d, A,*
and B,* (a= 1,2,. ..,N) according to the results of D. S. Clark and P. E.
Duwez [39], so as to describe best the dependence of yield stress upon
strain rate, but we must remember that the strain rate in impact problems
accompanied by the wave propagation effect may change from zero to about
lo00 sec-l.

__-----

FIG.24. Comparison of the experimental data with the prediction of the power strain-
rate law.

The experimental investigations of D. S. Clark and P. E. Duwez [39],

gave the curve cr versus B in the range of strain rate from zero to 200 sec-1
and it was pointed out that no further increase of yield stress was observed
when the strain rate exceeded 200 sec-l (see Fig. 2).
According to the first three types of function @(F) (2.8%)-(2.8%)and
by Eq. (2.71)we can write
(2.91)

u = a, 1[+ (+)l’d], 0 = uo(l + ),; = 0, [I+ log( 1 +]); .

In Fig. 24 the experimental data of Clark and Duwez for mild steel are
compared with the predictions of the power law (2.91,) for d = 6, y* = 40.4
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 286

sec-1 (see S. R. Bodner and P. S. Symonds assumption [19, 20]), 6 = 3,

y* = 180 sec-1; 6 = 3, y+ = 240 sec-1.

0’ ‘ 4D’ ’ ’ I I I I I I 1 1 1 -
80I ’
a0 I
.(80 pw MI PBI) am
stmin mte (sBc.-’)
FIG.26. Comparison of the experimental data with the prediction of the linear strain-
rate law.

FIG.26. Comparison of the experimental data with the prediction of the exponential
strain-rate law.
286 PIOTR PERZYNA

The analogous comparisons with the linear law (2.91,) and the exponential
law (2.91,) are made in Fig. 25 for y* = 180-300 sec-l and in Fig. 26 for
y* = 80-240 sec-l.
To determine theconstants A,* and B,* (a = 1,2,. . ., N)wemay take the
data of D. S. Clark and P. E. Duwez [39], given L. as a function of a/ao- 1.
Hence we can write two systems of linear equations for A,* and B,*

..,N,
N
2 A,*[eXp(Xi)a - 11 = B ( X J .
a-1
i = 1,2,,
N

where Xi = ai/u0- 1 for some points of Q = o,,u,,. .., QN, and d(X,)
.
(i = 1,2,. .,N) are known from experimental data.
I

0 " " "

4
'
80 a #
sn
' I
w
I '

min mte(kc:')
m
I I
m s
I #
a '
m
I -
FIG.27. Approximation of the experimental curve by the functions (2.88,) and (2.88,).

The approximations of the experimental curve Q versus B by the functions

(2.88,) and (2.88d for N = 6 are plotted in Fig. 27. The values of constants
.
A,* and B,* (a = 1,2,. .,N) are given in Table 2.
We assume here that the influence of the elastic part of the strain rate is
very small, and that the curve versus vx vp
is the same as thecurvea
versus 6. In the light of the recent experimental results the first assumption
seems acceptable.
Since all experimental investigations concerning the dynamic behaviour
of the materials have been performed under one-dimensional conditions,
we cannot discuss the validity of the second assumption and must consider
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 287

it as a hypothesis. This hypothesis will permit the use of the same constants
in the general constitutive equations (e.g. in (2.68) or in (2.67)) as those
determined for one-dimensional states of stress.

TABLE2

A,*(sec-l) 217.66 -664.11 874.62 -484.16 93.66

B,* (sec-I) 337.63 - 1470.66 3271.71 -3339.98 1280.06

From experimental results of J. D. Campbell and J. Duby [27], in which

a mild-steel specimen was subjected to compressive impact load, we have

7lme (psec.)
FI 28. Comparison of the experimental data for strain rate with the prediction of
the power law.

the average stress-time curve and the average (strain-rate)-time curve.

Taking the stress-time curve, we propose to compute the strain rate as a
function of time according to the constitutive equations already introduced:
288 PIOTR PERZYNA

6=6/E+y* exp - -
o: ( J J
(2.93) N r ,

Time (psec.)
FIG.29. Comparison of the experimental data for strain rate with the predictions of
the exponential law.

The best agreement with experimental data can be observed for the power
function (P(F), (2.93,) and for the exponential function @(F), (2.93,). The
results for these functions are plotted in Figs. 28 and 29, respectively
(cf. [133]).
All theoretical curves show that in comparison with experimental data
the maximum value for the strain rate is shifted in time by about 15 psec,
which may be due to the delay time. Indeed these shifts have the order of
magnitude of the delay times observed in mild steel under stress of up to
100,OOO Ib/ins (see J. E. Johnson, D. S. Wood,and D. S. Clark [87]).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 289

In Fig. 30 the experimental data of J. Harding, E. 0. Wood, and J. D.

Campbell [76], for mild steel are compared with the theoretical predictions

8
*
P
h,

9 6
9

4 4

0 P 4 6 8
Rillin(%)
FIG. 30. Comparison between the experimental and theoretical stress-strain curves
for mild steel.

0 4 6 8
Strain (“/o)
.FIG.31. Comparison between the experimental and theoretical stress-strain curves
for pure iron.

concerning elastic/visco-perfectly plastic material (curve A , @(F)= exp F

- 1, y = 240 sec-l) and concerning elastic/viscoplastic, work-hardening
material (curve B, @(F)= exp F - 1, y = 360 sec-l). The analogous
comparisons for pure iron are given in Fig. 31 (curve A , O(F) = exp F - 1,
y = 300 sec-1; curve B, @(F)= expF - 1, y = 500 sec-1).
290 PIOTR PERZYNA

In Fg.32 the experimental data of F. E. Hauser, J. A. Simmons, and

J. E. Dorn [77], for pure aluminium are compared with the predictions of
elastic/viscoplasticwork-hardening theory.

FIG.33. Comparison between the experimental and theoretical atreaa-strain curves at

i = const for pure aluminium.

11. Ternfievatwe-Defielzdentand Strain-Rate Smsdive Plastic Materids

The influence of the strain rate on the mechanical properties of metals
depends markedly on the value of absolute temperature. It has been dem-
onstrated experimentally [149], that at -18OOC the low yield point of
pure iron is equal for all strain rates. A similar result was obtained by
J. M. Krafft, A. M. Sullivan, and C. F. Tipper, [W].On the other hand,
even a slight change of the strain rate at high temperatures causes a con-
siderable rise or drop in the yield point.
Recent years have brought a number of improvements in measuring
the dynamic properties of metals at various temperatures. Papers [34,
109, 1491 may be quoted here. Unfortunately, these investigations do not
cover the entire range of variation of strain rate, strain and temperature.
Therefore the effect of strain-hardening of materials cannot be studied and
all results derived are only certain in the range of strain rate B and tem-
perature 8 already examined.
The constitutive equations describing the temperature-dependent and
strain-rate sensitive plastic materials have been discussed in paper [126].
A detailed analysis of some particular cases of the constitutive equations
and a comparison of theoretical with experimental results can be found in
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 291

[138]together with a complete discussion of the problem of an appropriate

selection of the temperature dependent coefficients. But it should be noted
that the considerations given in [138]are purely phenomenological, without
regard to thermodynamics of irreversible processes. Irreversible thermo-
dynamics for viscoplastic materials has been presented by H. Ziegler [197]
and by C. Wehrli and H. Ziegler [188]. A review of thermodynamic con-
siderations in continuum mechanics has been given by H. Ziegler [198,1991,
in which also information concerning the original papers on thermodynamics
of irreversible processes can be found.
I t appears that the simultaneous influence of strain rate and temperature
can be described by (2.47)if only the quantities appearing there are taken
in their dependence on temperature. In general, temperature-dependent
quantities can be y , F, as well as the function 0 itself. With Z denoting
the coefficient of thermal expansion, the constitutive equations for tempera-
ture-dependent and strain-rate sensitive materials take the form

Consider now the inelastic part of the strain-rate tensor. For perfectly
plastic materials with Huber-Mises yield condition (2.94)gives

or equivalently

For a one-dimensional state of stress (2.96)and (2.96)transform to (compare

with (2.70)-(2.71))

(2.97)

(2.98)

Since all experiments of interest were carried out under tension or compression,
(2.97)and (2.98)will be useful in order to give experimental evidence of
the present approach.
292 PIOTR PERZYNA

The most general case in which the quantities y and a, and the function
@ are temperature-variable has not been studied in the literature. The main
reason for this is that the next two cases considered, where only two quan-
tities vary with temperature, were found to give satisfactory correspondence
with experimental data over the entire range of temperature. Let us con-
sider in detail the following three particular cases:
A. a, = a,(@ and y = ~ ( 6and ) @ is independent of temperature. It was
shown that the viscosity of the plastic material proposed by Hohenemser
and Prager can be used to describe the strain-rate sensitivity of metals.
Since the viscosity constant y , similarly as the yield point a,, depends on

FIG.33. Compariaon of the theoretical predictions with the experimental data for
mild steel.

temperature, the case A is a further generalization of the above-mentioned

phenomenological approach. This assumption allows, as will be indicated
later, for the description of the dynamic properties of a number of metals
over the whole investigated range of B and 8 (Figs. 33-37). A particular
case of this assumption is the hypothesis [187] that

(2.99) a(w= + a,(e),

which can be derived from (2.98) if we assume that the product
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 293
Mild steel

(d)- 73wNI- "40(8-125)'I3

m-

m-

80-

go-

40-

Pure imn
1oQ-
Q (~~=4saoo-~ao(e-m)1~z
4?l-
8aT

80- "'~i77/45?1-Sl(s-l.R)
'\ n-10

FIG.36. Comparisonof the theoretical predictions with the experimental data for pureiron.
294 PIOTR PERZYNA

t .WmJ

Pure imn

o pVgh,Chung 4 Hopkins [Wg]

0 1 I I I I I
-2m -#o 0 m sp]’
FIG.36. Comparisonof the theoreticalpredictions with the experimentaldata for pureiron.

arm-%

*s -
25-

Id-

29 -
22 -

w-
40 -

0.9-

08-

FIG. 37. Comparison of the theoretical predictions with the experimental data for
aluminium.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 296

is independent of temperature throughout the whole range of C. This

assumption can never be true for real materials, and the results of the tests
quoted here do not conform with the foregoing hypothesis.

B. uo = a,(@, @ depends on temperature, and y is independent of tem-

perature. This is the conception of L. D. Sokolov [169]. J. F. Alder and
V. A. Phillips [l] and many others. These authors studied the power func-
tion
(2.100) o(i,e)= uo(e)( i + y e ) ,
where the exponent n is also a function of temperature. Such a form of a
state equation has a certain thermodynamical justification and therefore
attracted the attention of a number of investigators, but no consistent
results have as yet been obtained. L. D. Sokolov [169]investigated four
kinds of steel and found the exponent n as a linear function of the homologous
temperature with the extrapolated line coming through the origin. Exper-
imental data [l] on aluminium, copper and steel show that n(0) can be
approximated by two straight lines. Recent experiments [34]indicated the
non-linear relation between 1z and 8 for aluminium in the temperature
range 0 6 6 0 ° C . No algebraic representation of the function n(8) has been
proposed.
C . uo = ao(8),y and @ are independent of temperature. This is a gener-
alization of W. Prager's [la]idea for materials exhibiting strain-rate
sensitivity. Eq. (2.98) yields

(2.101)

This hypothesis suggested in [186] has been criticized in [187]. The

analysis of experimental data presented later shows that neither (2.99) nor
(2.101)holds over the whole range of B and 8. However, for narrow ranges
of temperature (2.101)can give a fairly adequate description. From the
point of view of practical applications (2.101)leads to the simplest results
but, as already mentioned, its range of validity is very limited. Although
cases A and B both give a good description of the mechanical properties of
metals, it is more convenient to use relations where not the function but
only the coefficientsof the material depend on temperature. Thus, the case A
seems here more appropriate.
In the following the available experimental data for aluminium, pure
iron and mild steel are discussed in order to show the validity of A.
296 PIOTR PERZYNA

The results of tests on aluminium, iron, and steel taken from [l, 34,
96, 109, 1121 and [149], have been elaborated for our purpose. Only the
range of temperature is considered, in which the material is metallurgically
“stable.” In order to use the results of tests the form of the function Q, has
to be assumed. The power function has been found to be preferable (see
(2.88,)), and the introduction of more involved function is unnecessary,
since the power function @ gives a very good agreement with all tests con-
sidered.
Circles and triangles in Figs. 33-37 are experimental points both from
static and dynamic tests. All static tests of u,, were approximated by either
power or exponential functions and the lower solid lines are plots of u,(0).
The upper solid lines represent u(i,0) for one or several values of strain
rate. It is seen, that by a suitable choice of y ( 0 ) very good agreement with
experimental data can be achieved. This is evidence of the correctness of
the formula (2.96),case A. In the same figures the dash and dash-dot lines
correspond to the hypotheses of Vishman, Zlatin, and Joffe [186] and Volo-
shenko [187], respectively. In both cases a definite deviation from the
experimental points is evident.
It should be pointed out here that the comparison with experiments has
been carried out under the assumption that the material is perfectly plastic,
hence the strain-hardening effect is disregarded according to (2.96). If we
include strain-hardening, (2.94) yields

(2.102)

or

(2.103)

for a one-dimensional state of stress. Eqs. (2.102) and (2.103) could not be
compared with experimental data because of lack of suitable tests. On the
other hand, an excellent agreement of (2.96) with all tests here presented by
no means denies the existence of strain-hardening properties when the rate
of strain is high. This is because of the arbitrariness in the choice of the
function y ( 0 ) which was taken to fit the experimental data. If we had in-
cluded the strain-hardening in the analysis, we would also have changed
the function y ( 0 ) . I t seems most probable that (2.102) and (2.103) might
give a better description of the dynamic response of certain metals in larger
ranges of variables.
Recent investigations indicate the strong interaction between yield
stress, strain rate and temperature in dynamic tests of various metals.
The present approach shows how a simple temperature modification of the
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 297

constitutive equation for strain-rate sensitive materials developed in

[132-1341, can describe the effect of simultaneous change in 8
' and 8 .
Once more, our conclusions refer only to ranges of 4. and 8 already
examined; particular attention should be paid to extending these conclu-
sions to other ranges.

12. Linearization

In the theory of perfectly plastic'solids, the quadratic yield condition of

Huber-Mises is often approximated by a piecewise linear yield condition
to simplify the analysis of boundary value problems. W. Prager in [146]
discussed an analogous linearization in the theory of viscoplastic solids.
Assuming @(F)= F and neglecting the elastic part of the strain rate
in (2.45) we obtain

(2.104)

Piecewise linear approximations to the constitutive equation (2.104)

can be obtained as follows. According to (2.104), viscoplastic flow can only
occur if
(2.105) F > 0.
This flow condition, which has the form of a single nonlinear inequality,
is now approximated by the set of m linear inequalities

(2.106) L(v)= ap4spq

(v) - p ( v ) > 0, (Y = 1,2,. . .,m),
in which the symmetric deviators 4)
and the scalars ptv) are independent of
the state of stress. Through a suitable choice of these deviators and scalars
and of the number m,the flow condition (2.106) can be made to approximate
(2.105) with any desired degree of accuracy. Finally, the constitutive equa-
tions (2.104) are replaced by

(2.107)

Since 4)is a deviator, 6 = 0, and it follows from (2.106) that

(2.108)
298 PIOTR PERZYNA

The constitutive equation (2.107) can therefore be written in the form

m

(2.109)

13. Relaxation Processes

To discuss a relaxation process for general states of stress, let us consider

an elastic/Viscoplastic body, occupying the three-dimensional region V with
the regular surface S, and investigate the following boundary value problem.
Consider first the loading process in which the surface tractions Ti are
prescribed on the part S, of S and vanish on the remainder S., This loading
is to be followed by a relaxation process in which the surface velocities vi
vanish on S, while the surface tractions Ti continue to vanish on S., In a
relaxation test that is to furnish useful information about the constitutive
equations, the states of stress and strain must be homogeneous. A relaxation
process defined in this manner will be called an A-process (see [132]).
First consider the tensor rijdefined by

(2.110)
s
During an A-process the tensor rijvanishes. Thus,

(TiVj + TjVi)dS = (UiknkVj + 0jknkVi)dS

S S

(2.111) +
ak(0ikvf 0jhvi)dV = (uikakvj + ujk&vi)dV

'5
V

=
2
(0&j + 0jkkki)dV = 0.
V

The states of stress and strain are homogeneous during this type of relaxation
test ; we therefore have
(2.112) i(u&' + a&i) = 0.

These conditions enable us to determine the state of stress as a function of

time during the relaxation process.
Assume now that a certain state of stress characterized by u$) (or s$)
and uiy) and JJo) = s$%i?)/2> K g has been reached a t the time t = 0 and
 FUNDAMENTAL PROBLEMS IN VI SCOPLASTICITY 299

the relaxation process follows. Then from (2.67) and (2.112) we obtain a
system of six differential equations with respect to sij and akk

By letting i = j in (2.112) we have the condition

Multiplying (2.112) by aij we obtain

(2.115) $7 i k E'kr - 0.
u-a.
Consider now the simpler boundary value problem in which the loading
process is characterized by the surface tractions T i being prescribed on the
entire surface S, while during a relaxation process the surface velocities v j
vanish on the entire surface S. This kind of relaxation process we shall call
a B-process.
Because the states of stress and strain are homogeneous during a relaxa-
tion test, we now have conditions

(2.116) d..$7 -
- 0.
From (2.67) and (2.116) we obtain a system of five differential equations
for sii in the form

(2.117)

The conditions (2.116) lead to two useful scalar conditions

(2.118)

and it is worth noting that these conditions are valid for A-processes in
incompressible materials because then

(2.119) in = 0.

The validity of (2.118) for an A-process under condition (2.119) follows

readily from (2.114) and (2.116).
300 PIOTR PERZYNA

According to (2.67) and (2.11S1) we have during A-processes in incom-

pressible materials and during B-processes in any materials a relaxation
equation for J 2 in the form

(2.120)

This can be written as a nonlinear Volterra integral equation of the second

kind
t

Under the assumption that the integrand J:/W(Ji/2/k- 1) satisfies a Lip-

schitz condition, that is that

where No is a positive constant, the solution of (2.121) can be obtained by

iteration based on the recurrence formula

It is easily verified that the series

Jd0)+ nC
W

(2.124) L l ~ n + l ) ( t )- J ~ ( 0 l
=O

is absolutely and uniformly convergent, and its sum

(2.126)

is the solution of the integral equation (2.121), and hence of the differential
equation (2.120). Of course, the solution (2.126) is valid only in the nonelastic
region J s >ka.
Full discussions of the solutions of the relaxation equations have been
presented in the papers [132, 128, 891.

The detailed discussion of the continuum approach to fracture of metals

has been presented by D. C. Drucker in papers [66, 661. Fracture is a very
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 301

complex phenomenon which is preceded by some viscous flow or plastic

deformation or both in almost all brittle as well as ductile materials. Both
flow and corresponding fracture values depend upon the temperature, the
strain rate and, in fact, on the entire time and temperature history of stress
and strain. It has been'observed that an increase in the strain rate raises
the flow curve and promotes fracture, Fig. 38. A decrease of temperature
has an effect analogous to an increase in strain rate, Fig. 38.

t a
t b

Strain m@ d
c L
Temperature 6

t C

L -
strain
FIG. 98. Increasing strain rate i,or decreasing temperature 0 promotes fracture
(D. C. Drucker [SS]).

It seems, however, that the phenomenological prediction of fracture

must be based on microscopic and atomic results. More elaborate models
are desirable, in particular, models which contain the essential features of
crystal dislocation (cf. [83, 66, 661).
302 PIOTR PERZYNA

111. STRESS WAVE PROPAGATION IN AN MEDIUM

ELASTIC/VISCOPLASTIC

1. M a t h m t i c d Preliminaries

A. General Considerations. We shall consider problems with initial and

boundary conditions for a quasi-linear system of partial differential equations
of the form

(34 U,+AU,+ B=0,

where U is a column vector with the n components U,,U,, . . . ,Us,A is an

n x n matrix and B is an f i element column vector; A and B depend on the
spatial coordinate z, on the time t and on the components of the vector U
(in the case of semi-linear system the matrix A is independent of the com-
ponents of vector U).
The system (3.1)is assumed hyperbolic, that is, the matrix A has n real
eigenvalues (i = 1,2,. . .,n), and possesses a full set of linearly independ-
ent eigenvectors (see R. Courant [a] and A. Jeffrey [SS]). The left eigen-
vectors of A, l(i,$)with k = 1,2.. ..,s corresponding to the eigenvalue 1,)
with multiplicity s satisfy the equations

They may be used to display the equations (3.1) in characteristic form

and to introduce the n characteristic curves C,) as follows. Pre-multiply
equations (3.1)by lP)and assume for the moment that the n eigenvalues of
A are distinct; the n equations are now in the characteristic form

+
The operator a/at 1,) (a/az) in the ith equation represents differentiation
along the ith characteristic curve determined by

(3.4)

If = const the conditions (3.3) along the Characteristic lines (3.4)

may be wntten as follows:

(3.5)

where qi), b,) and c& depend on t , a, and u$.

 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 303

The solution of the initial value problern for the system (3.1) has been
studied by numerous authors. First existence and uniqueness were discussed
in full (see K. 0. Friedrichs [70], R. Courant and P. Lax [511, P. Hartman
and A. Wintner [76], P. Lax [103], A. Douglis [61] and R. Courant [54]).
Some theorems regarding existence and uniqueness may be obtained as
special cases of the same theorems proved for the system of first order partial
differential hyperbolic equations involving n independent variables (see
J. Schauder [154, 1551, M. Cinquini-Cibrario [36, 371 and R. Courant [54]).
In practice, to obtain the solution of the initial and boundary-value
problems for the system (3.1) the method of finite differences is used (cf.
R. Courant and K. 0. Friedrichs [50]). For a study of convergence of this
method and the conditions of its application see R. Courant, W. Isaacson and
M. Rees [52] and G. Prouse [l48].
Some modifications of the finite difference method, and its application
to the solution of the initial and boundary-value problems for quasi-linear
and semi-linear systems (3.1)were proposed by H. B. Keller and V. ThomCe
[93, 94, 174-1761, R. Courant [53, 541 investigated the possibility of ap-
plication of successive approximations to the solution of initial and boundary-
value problems for the system (3.1).

B. A non-linear boundary-value problem for a semi-linear hyperbolic

partial differential equation. Let Q be the closed rectangle (0 x 2 xo, <
O < y < yo} [opposite vertices at (0,O)and (xo,yo)],and consider-the
normalform of the semi-linear partial differential equation of hyperbolic
type in two independent variables

(3.6) u,y = f(x,y,u,ux,uy),

where u is the unknown function and f(x,y,u,uX,uy) is a given continuous

function in Q for arbitrary u,u, and uy. A solution of (3.6)is a function u(x,y),
which is continuous in Q together with its partial derivatives ux,uy and uxy
and satisfies (3.6).
We require that u(x,y)satisfy the boundary conditions

.xb?J(x)) =g ( x ~ . u ( x ~ q , ( x ) ) ~ ~ Y ( x , q ( ~ ) ) ) ,

(3.7) uY($(Y)JY)= k(y*zl($(Y)sy)s

.X($(r)JY))J
u(x*,y*) = u*,

where (x*,y*) is a point in Q, u* is a given constant, q ( x ) and $(y) are ar-

bitrary curves within Q, and g and h are given continuous functions in Q for
arbitrary u,u, and u,,.
The boundary-value problem (3.0)-(3.7) was first stated by 2. Szmydt
[167], who proved it as follows:
304 PIOTR PERZYNA

Theorem 1 . If

1. the functions y ( x ) and $ ( y ) are continuous in Q, and O<- y ( x ) s y o ,

0 5 #(r)5 xo;
2. /, g and h are continuous real-valued functions of x , y , u, u,, u,, in
Q, and satisfy the following Lipschitz conditions:

1
I / ( ~ , Y # W X , % )- f ( X , y , ~ , ~ z , U 5 - UI
y ) L( (4 + 14, - u,[ + lay- zcyl).
(3.8) Ig(X,Wy) - g(x,u,u,)l 5 LI4 - + M ( 4 , - uyl,
(h(y,n,cx)- h(y,u,ux)I 5 L ( 4 - 211 + N(4, - 21x1 ;

3. the constants L, M ,N satisfy the inequalities

where X, = max(xo,yo);

there exists a unique solution u(x,y) E CQ1 to the boundary value problem
(3.6), (3.7).

Proof. We shall discuss here only the basic elements of the proof,* which
we shall use to construct the solution of the nonlinear problem (3.6)-(3.7)
by means of the iteration method.
Consider the linear space CQ' of the functions u ( x , y ) . The norm in this
space we shall define in the following form:

(3.10) l SUPINXJJl
l l ~ ( X J J ) l=
Q
+ SUPl%(XJJ)
Q
I+ IUY(X>Y)I.

Convergence in the space CQ1stands here for uniform convergence; this

should hold for sequences of functions u(x,y), together with their first
derivatives, in Q. Due to the fact that for uniform convergence the Cauchy
criterion is fulfilled, the space C& is a complete metric space.
Let us now introduce in the space Col the integral transformation 9,
mapping any function u(x,y) E C& into the function

(3.11) fi(X,Y) =9 [ U ( X , Y ) I ,

* For a complete proof see 2 . Szmydt [167].

 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 306

where

W[.(x,y)l = *. + i
X*
g(s,.(s,~(s)),.y(s,~(S)))ds

(3.12) +1
Y

Y.
~(Zl.(~(Z),Z),.X(~(Z),Z))dZ + 5I
Y X'

Y* P(.,
~(s,z,.(s,z~,~,(s,z),.y(s,z))dsd%

+ j j~~s,z,.~s,z~,.,~s,z~,.~~s,
X* CPW

The functions 6 ( x , y ) determined by (3.11), (3.12) and their first derivatives

are continuous in Q. Thus, the transformation W maps the space C,l into
itself.
This shows that every function which is a fixed point of the mapping W ,
i.e.,
(3.13) W , y )= W Y )
satisfies (3.6) and the boundary conditions (3.7). To complete the proof,
it is sufficient to show that there exists only one fixed point of the mapping W .
By Banach's theorem,* every contraction mapping defined in a complete
metric space has in this space exactly one fixed point. Since the metric
space Csl is complete, we have to prove that W is a contraction mapping.
Using the definition of the norm (3.10),we may easily obtain the following
result :
(3.14) 11% - 6211 = kI.l1 - %ll*
where
k = max[2L(XO2+ 2X0 + l),2L(Xo2+ X,) + N(X,+ 1),2L(Xo2+ X,)
(3.15) + M(X, + 1)1,
and from assumptions (3.9) we have kc 1 ; hence the mapping W is con-
tractive.
Taking advantage of the mapping W , we may apply the method of
successive approximations to obtain the solution of the non-linear problem
(3.6)-(3.7). The iterative scheme is as follows:

* For a discussion of the Banach theorem see Ref. [91].

306 PIOTR PERZYNA

(3.16)

In view of our assumptions, the sequence { U [ ~ ) ( X , ~given

)} by (3.16) is a
Cauchy sequence. Hence the iterations (3.16) converge uniformly to the
solution

In practice we have the following iterative scheme :

Note also that by assuming the Lipschitz condition (3.8) in the weaker form

5 -4
I f ( x , Y , u * M y )- f ( x , Y , ~ , w y ) lL((4, + (21, - uyl),
3.18) 1 5 Mp, - uyl,
Ig(x,%cY) - g ( x , w y )
/h(Y,%%)- h ( Y , W %I )5 Np, - %I
and using the Schauder theorem we may prove the existence of the solution
for our boundary-value problem.*
Consider now the special case of the boundary-value problem (3.7),
where we assume

(3.19) h ( y w 4 = b ( y , w 4 - ~,$~(y)on x = y!(y).

Then the boundary conditions have the form

* For details see 2. Szmydt [lee, 1701

 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 307

(3.21)

This special case of the non-linear boundary-value problem has great

practical significance for the solution of the stress-wave propagation problem
in an inelastic medium.
Another special case of the boundary-value problem (3.7) is the linear
relation
ux(-%Pl(x))= U l ( 4 + + Pl(x).(x>(?J(x))>
%(X)~y(X,(?J(X))

(3.22) uY(#(Y)JY) = a2(r) + %b)'x(#(y)3y) + PZ(r)'(#(y)Py)l

u(x*,y*) = u*,

where a,, q, Pi E CQO for i = 1,2. The same problem as (3.6), (3.22) was
posed by J. Conlan [46] who solved it by a polygon method.*
The mapping for the problem (3.6), (3.22) takes the form

(3.23)

* The Euler-Cauchy polygon method for proving the existence of a solution for

was extended by J. B. Diaz [69] t o the characteristic boundary-value problem for a

hyperbolic partial differential equation [3.6] and further extended by J. Conlan [as]
t o the Cauchy problem, and to the mixed boundary-value problem (Picard's problem).
308 PIOTR PERZYNA

It has been shown (see [166-168, 991) that all classical problems (i.e. the
Cauchy, Darboux, Picard and Goursat problems) can be analogically treated
as special cases of the nonlinear problem (3.6)-(3.7). In such a procedure,
particular note must be taken of the restrictions on the functions which
describe the boundary conditions.
Note that all the results discussed here are valid for a system of second-
order semi-linear hyperbolic partial differential equations.*
C, A non-linear boundary-value problem for a linear hyperbolic equation.
Consider the normal form of the second order linear partial differential
equation of hyperbolic type in two independent variables :
(3.24) s(u) uxy + a ( x , ~ ) u+x +
b ( x ~ ~ ) U yC ( x , Y ) U = d(x,y).

The coefficients a, b, c , d are given as real-valued functions of x and y defined

in Q.
The purpose of the present section is to construct a solution u(x,y) of
Eq. (3.24), which is continuous in Q together with its partial derivatives
u,, uy and u,,,, and satisfies the boundary-value conditions (3.7).
The problem (3.24), (3.7) was discussed in [137]. It has been shown
that the solution of this problem may be obtained by means of the procedure
proposed by S. C. Chu and J. B. Diaz [35] for solving the linear problem.
Let V(E,q;x,y) be the Riemann function' for the homogeneous equation
U ( u ) = 0. Any solution of Eq. (3.24) in Q may be given in the form:
Y

+ + lWq)Vx*,q;x,y)dq,
(3.26)

where
4x,y) =S(x,y)
X*j Q(E)W,y*;x,y)@
Y*

(3.26) * ( x , y ) = u*V(x*,y*;x,y) + X'ij y*

d(E,q)W , q ;x,y)dqdE.

The functions Q ( x ) and IT@) should be determined so that the solution

(3.26) satisfies the boundary conditions (3.7). Then the boundary-value
problem (3.24),(3.7),reduces to the equivalent system of non-linear functional-
integral equations :

For the discussion of these general cases see Z. Szmydt [167-1701.

For a review of methods for obtaining the Riemann function see E. T. Copson [47].
He has also made a survey of equations for which the Riemann function is known in
closed form. See also R. Courant [64] and A. Wintner [196].
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 309

where
310 PIOTR PERZYNA

In what follows we shall use the notations:

KO = [IK~(~,x)I.IK~(~,~)I,IK~(€,~)~.~K~(~~Y)II.
8
(3.29) i = 1,3,6; i = 2,4,6; K = 7,9,11; 1 = 8,10,12;

K* = max [ I V ( ~ , ~ ' ; X , F ( ~ ) ) I , I;%,~(x)J,(V(x*,y;~Cy),y)IJ

V(~*,~(~)
0
lV(w,Y*;KY)9Y)ll.
The precise formulation of the problem to be solved is given in the
following
Theorem 2. If
1. the real-valued functions a(x,y), b(x,y), c(x,y) and d ( x , y ) are
continuous in Q ;
2. the function ~ ( x is) continuous on 0 2 x < x, and the function
< <
+(y) is continuous on 0 - y - yo, Gith-0 - q ( x ) - yo; 0 6 <
W )I*o;
3. the functions g and h are continuous in Q and satisfy the Lipschitz
conditions
- g(xSM,My)I 5 M*lI" - + N*Ia.V- u y l .
lg(xJa14Y)

\h(y,a,ax)- h(y,M,Mz)I 5 M*lii - MI + N*I% - M x l ;

4. the constants KO, K*,X,,M* and N* satisfy the inequality

(
2K0X0 1 + -
M*&")- + N* < 1;
there exists, throughout Q, a unique solution w(x,y) E CQ1to the boundary-
value problem (3.24),(3.7).
Proof. We shall show by successive approximations that the system
(3.27)possesses a unique continuous solution set a(%)
and n(y).
We define
the sequences {Q(,,](x)} and {II(,,)(x)} as follows

5 o(n
X rp(Z)

~(rn,(x)= ~ ( x ) + -~ ) ( E ) K ~ ( E , w + -*)(q)Ka(q~dq
X* Y*

1
x

+ [v(%,Y*; x , q ( % ) ) l -(~~ ~ ~ ( % . +~ ~ ( x ) ) otn- l ) ( l j K S ( t . % ) d ~

I*
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 311

for n = 1,2,3,.. ., and for arbitrary continuous functions Q,(x) and I l , ( y ) .

.
Equations (3.30) for n = 1,2,3,. ., yield:

If& +I)(%) - Q(&) I

312 PIOTR PERZYNA

With the notation

(3.32)

we obtain by adding the inequalities (3.31)

The inequality

(3.34) 2KOX0(1 + M * & N * ) + N* < 1

implies that {$&)(x)} and {n(,)(y)} converge uniformly to the continuous
limit functions Q ( x ) and 17(y), which constitute a solution to the system
(3.27).
The above procedure shows simultaneously that this system has a unique
solution*.
We now turn to the following particular case of the boundary-value
problem (3.7):

(3.36)

In this case, using the same procedure as above, we can prove

Theorem 2a. If
1. and 2. of Theorem 2 hold;

3. the function g and h are continuous in Q and satisfy the Lipschitz

conditions

* That the problem (3.24), (3.7) has a unique solution can be shown easily.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 313

Ig(W-4 - g(x,w)l 2 Mop - UJ,

(3.36)
Jk(y,4)- h(y,u)1 5 M014 - U J;
4. the constants KO. K*. X, and M o satisfy the inequality

(3.37)

there exists, throughout Q, a unique solution to the boundary-value problem

(3.24),(3.35).
The non-linear generalized Picard problem stated first by G. Majcher [IOS]
may be treated in a similar way, as a special case of the problem (3.24).
(3.7). In that case the boundary-value conditions have the form:
4 X N =g ( 4 ,
(3.38)
UY($(Y)lY) = hCyP($(Y)IY)PX($(Y)#YH.
We shall now discuss in detail a linear mixed boundary-value problem
for a linear hyperbolic partial differential equation. This problem can be
stated in the following form*:

q@)
= d(x,y),

(3.39)
aO(Wx,v(x)) + .l(x)%(x,V(x)) + a,(x)Uy(X,v(XN = %(X)J

P ~ ( Y ) N $ ( Y ) A+ P i ( ~ ) u x ( $ @ )+
A P a @ ) % y ( # ( y ) , ~=
) us@),
zl(x*,y*) = u*.
The solution of the problem (3.39)has been given by C. Chu and J. B.
Diaz [36]. The case (x*,y*) = (0,O). was treated in the paper [S]by A. K.
Aziz and J. B. Diaz.
Using the solution (3.2)-(3.3).we can reduce the boundary-value problem
(3.16)to an equivalent system of linear functional-integral equations :
X 9(*)

J
X*

(3.40)

It is obvious that the problem (3.39) can be treated as a linearized form of the
problem (3.241, (3.7).
314 PIOTR PERZYNA

where

Theorem 2b. If

1. and 2. of Theorem 2 hold:

3. a,, 9, Pi are continuous in Q for i = 1.2 and j = 0,1.2;

4. al(x) # 0 for all x in O <- x x,,, P2(r) # 0 for all y in 0s

- y <- y o ;
6. the foIIowing inequalities hold:

1~2(X*)/.l(x*)l -= 1, JBlb*)IP&.Y*)l<
1;
6. Y ( x * ) = y*, $@*) = x * ;

7. xo, yo are positive constants such that A, +

2NX, < 1, where A,
is the maximum of both /A(%)]and 1BCy)l over all x in 0-
s x<- xo
5
and y in 0 5 y yo, and

N= "0" [ l ~ l ( e ~ ~ ; ~ ) ~ , ~ ~ * ( e l ~ ; y ) ~ l ~
there exists, throughout Q,a unique solution rs(x,y) E CQ1to the boundary-
value problem (3.39).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 316

Proof. By the definitions of A ( x ) and B(y), the conditions 5. and 6.

imply that \A(x*)I < 1 and \B(y*)I < 1. Since A ( x ) and B ( y )are continuous
this means that, if X, is sufficiently small, the inequality A , +
2mX0< 1
of the condition 7. holds.
Define the sequences {Q(&)} and {17(fil(r)} as follows

Q(s)(x) =G*(x)

P(4
i
+ A ( ~ ) n ( f i - ~ ) ( p (+x ) )klO,r*;x)n(s-l)(5')dE
I.

J
Y*
(3.42)

for n = 1.2,3,. . ., and for arbitrary continuous functions Q(,(x) and n(&y).
Proceeding as in the proof of Theorem 2, we easily obtain from (3.42):

(3.43) Z"+l + - % + I s (A,+2fiX,)(Z"+&),

for 1z = 1,2,3,. . ., from which the conclusion of the theorem follows im-
mediately.
For the case

(3.44)

Theorem 2b is valid (see C , S. Chu and J. B. Diaz [36]). In that case, the
constants N and X, should satisfy the following inequality: SI?X,< 1.
G. Majcher [lo81 has considered the boundary-value problem consisting
of the partial differential equation (3.24) subject to the boundary conditions
(linear generalized F'icard problem)
,

u(x,O) = a&),
(3.45)
BoCr)MlbCr)A + A(r)4lb(r)d + A ( r ) N v ( l b b ) A = OaCy).
A. K.Aziz and J. B. Diaz [6] have shown that the problem (3.46)stated
by G. Majcher [lo81 may be treated as a particular case of the linear problem
(3.39).
316 PIOTR PERZY NA

Full discussion of the other particular cases of the linear problem (3.39)
has been presented in paper [a], where the history of the linear problem
and an extensive bibliography is also given.
All theorems discussed in sections B and C are valid “in the small.”
The inequalities (3.9), (3.34), (3.37), and (3.43) required to prove the existence
and uniqueness of the solutions are very strong and impose important rest-
rictions on the size of the rectangle Q and on the boundary conditions.

2. Formulation 01 the Problem

The object of this chapter is to discuss solutions of certain boundary-
value problems for a work-hardening and rate-sensitive plastic material.
Four types of waves will be considered ; spherical waves, cylindrical radial
waves, cylindrical shear waves, and plane waves.
We assume that the rate-sensitive plastic material can be treated as
elastic/viscoplastic according to the constitutive equations (2.58).
In papers [130, 131, 136, 1381 it has been shown that the solution of the
propagation problem of stress waves of all types in an infinite elastic/visco-
plastic medium may be reduced to one mathematical problem. By changing
the coefficients of the differential equations and the boundary conditions,
we may obtain the mathematical description of the four wave types. This
treatment has been applied in both the elastic and inelastic regions. In each
case the problem in the range of elastic/viscoplastic deformations reduces to
the solution of the corresponding boundary-value problem for the quasi-
linear or semi-linear first-order hyperbolic system (3.1).
Similarly, all problems of one-directional propagation of a stress wave
in a work-hardening elastic-plastic medium may be described by the system
(3.1) (see for instance G. Hopkins [SZ]).
Full particulars concerning the stress-wave propagation in elastic/visco-
plastic soil have been discussed in [127].
Refs. [130, 1311 are concerned with the analysis of wave propagation
in a non-homogeneous elastic/viscoplastic body. The solution of the problem
of propagation of stress waves in a work-hardening and rate-sensitive plastic
material has been given in [go, 1381. Paper [go] also discusses the reflection
of an elastic/viscoplastic wave from a plane.
In Section 3 we shall follow the solutions obtained in Refs. [136, 1361.

3. Application of the Method of Finite Differences

A. Spherical Waves. Consider an infinite, elastic/viscoplastic medium

with a spherical cavity of radius r,,. To the spherical surface of the cavity
the radial uniform, time-dependent pressure p(t) is applied. With reference
to the spherical coordinates r , p, 0 we thus have
 FUNDAMENTAL PROBLEMS IN VISCOPLASTLCITY 317

(3.46) = %(I$), U, = Ng = 0,

where u,,, uq, Ue are the spherical components of the displacement vector.
Assumption (3.46)(spherical symmetry) yields the following components
of the strain and stress tensors:

au
- a4
(3.47) &" =
C ' Epnp=Eee=-' I

(3.48) aw(r,t), apnp(r4 = aee(r4

Denote by v = aulat the radial velocity and by p the density of the

material. In that case the system of differential equations take the form
(3.1),where

0 --
l o o 0
P
-(K+Qp) 0 0 0 0
U= , A=
#ru - K 0 0 0 0
-1 0 0 0 0
0 0 0 0 0

(3.49)

B=

--
V
I
J
T h e characteristics (3.4) have the form

3p )
(3.60) I = const, I = I , f2 + const, where A = (
4p + 3 K 'Ie

and I = const is a triple characteristic.

318 PIOTR PERZYNA

According to (3.6), along these lines the following conditions must be

satisfied :

dE,
1
- -au,,
2P
+-1
2P
au,, - (+ + vy.) at = 0,

(3.61)

+
- ( 4 ~ 3K)dv f3Ad~pp= 0,
respectively.
If a pressure exceeding the plasticity limit, p > p , is suddenly applied to
the surface of the cavity, then the characteristic line Y - Y, - & = 0 will
be a strong discontinuity. Along the discontinuity additional conditions
must be satisfied, which will be called the conditions of kinematic and
dynamic continuity. For the spherical discontinuity these conditions have
the form (see for instance G. Hopkins [82]):

(3.62)
(3.63)
respectively.
Since the strain rate 6 , at the discontinuity must be regarded as infinite,
the constitutive equations give the relation (see [llS])

(3.54)

Using the relation which is satisfied along the characteristic Y - Y, - At = 0,

and (3.62) through (3.64), we obtain

da,
--
(3.66)
dr
- - qY,U")*
with the condition
(3.66)
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 319

where
(3.67)

To solve Eq. (3.66) one has to show that on the front of the shock Y = ro + At
the hardening parameter K = K(W,) appearing in (3.57) depends only on the
variables Y and u,,. By definition (2.44) we have for spherical waves

(3.68)
J J
0 0

but the last term vanishes, since

(3.69) [Ew] =
1: 1= o ;
-

here [ ] denotes the jump of the quantity in brackets across the discontinuity.
Separating the component of the strain tensor E,, into the elastic part E:,
and the plastic part f and using (3.52) and (3.53) we can write

(3.W EP, = u,/(Pp) -EL.

By Hooke’s law and (3.54) we obtain

(3.61) 4 = uw/(E.ap)s
and after a straightforward calculation
(3.62) w,= 0.
Since on the front of the wave the plastic work equals zero, the problem
of the work-hardening plastic material is reduced to the perfectly plastic
material, the hardening parameter K being equal to the yield stress k.
In the case of cylindrical waves and plane waves the proof that along
the shock K(WJ = k is fully analogical and will be omitted.
Equation (3.65) with the condition (3.56) leads to the nonlinear Volterra
integral equation of the second kind

(3.63) 0,
1
= Po - ‘Y[E,u,(E)ldS.
VI

where Y(Y,U,,)
has now the form

(3.64)
320 PIOTR PERZYNA

If Y satisfies the Lipschitz condition then the solution of (3.63) may be

obtained by
(3.66) a,, = lim a:)
n+m

with the recurrent relation

FIG.39. Elastic and plastic regions in the t,r-plane for the case of discontinuous front
wave.

and
v(n) = - &)/@A),
(3.67)
= a!?/@P);
&$I Ew = 0.

The solution given by (3.66)-(3.67) is valid for Y < Y* (Fig. 39) where I*
satisfies the condition
(3.68) V
m= k.
The solution along the discontinuity for r >r* has the closed form

1 Y*
a,, = -
R*
-
Y '
v=--- 1
plR*
r*
1 '
(1-- ;;);*:I --
(3.69)
r*
e, = @ASR*)-' ;
, eW = 0,
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 32 1

where

(3.70) R* = 2p@Aak1/37-1.
ConsLucrnow the solution in the elastic/viscoplastic region, ..e., region
P+ in Fig. 39 which is bounded by the discontinuity r - ro - At = 0 and

FIG.40. Elastic and plastic regions in the t,r-plane.

FIG.41. The characteristic net in the t,r-plane.

the straight line I = yo. On the discontinuity the quantities b,, ow, u, E,,
and e, are determined by means of the solution (3.66)and (3.67), while
a,, is known on the line I = r,, from the boundary condition.
Denoting the intersection points of the characteristic net as shown in
Fig. 41 we approximate Eqs. (3.61) by the following difference equations:
322 PIOTR PERZYNA

(3.71)

+ 3K)
-( 4 ~ [v(i,m,N) -~ ( -1 l.m,n +1)I - 31[~r(l,m,n)
- c,p(I - 1,mn,n+1)1= 0.
In every point of the characteristic net we have a system of five algebraic
equations with respect to five unknowns a,,, aw, v , E,, and
The procedure of finite differences can be applied to the elastic/visco-
plastic region in the case shown in Fig. 40. In this case we know the quantities
a,,, aPp,v , err and ew at the point t = t, on the line r = Y, from the elastic
solution in the region E*.

vm
The boundary *'I of the elastic/viscoplastic region (see Figs. 39 and 40)
can be obtained by an approximate method using the condition = K(Y).

B. Cylindrical Radial Waves. Consider an infinite Cylindrical cavity

of radius yo in an infinite elastic/viscoplasticmedium. To the surface of this
cavity let there be applied the radial pressure +(t), variable in time and
independent of and z. In cylindrical coordinates r, cp, z, we have
(3.72) u, = U(Y,t), up= u, = 0,

where u,,up,u, are the cylindrical components of displacement. The compo-

nents of the strain tensor and the components of the stress tensor are
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 323

(3.74) a,, = u,,(~,t). a,, = uw(r,t), a, = a&$).

The system of differentialequations which describes the problem has the
form (3.1) with

0 - AP 0 0 0

0 0 0 0

U= A= 0 0 0 0 ,
0 0 0 0
0 0 0 0

0 0 0 1
(3.76)

B=

- -V
r

where

(3.76) JS = 4 [(& + a& + d )- (udpp+ +G A I .

The characteristic lines (3.4) now have form

(3.77) I = const, r = Y,, fiU + const,

where r = const is a fourfold characteristic. Along these lines the following
conditions hold by (3.6):
324 PIOTR PERZYNA

a&,, = -
V
7
at,
(3.78)
1
derr- -(da,,
3K
+ da,, + da,,) + -dt = 0,
V
7

&-
33, or,- aw
P r 1
dr = 0.

On the cylindrical discontinuity the conditions of kinematic and dynamic

continuity have the same form as in the spherical case (cf. (3.52) and (3.53)).
The property that d,, tends to infinity on the cylindrical shock wave
leads to the relations
aw = a,, - 2 p r r ,
(3.79)
,, - 2 p w .
a,, = a

Using the relation satisfied along the characteristic line Y - r, - & = 0,

and (3.62) and (3.53), and relations (3.79) we can reduce the problem to the
same Volterra integral equation as in spherical case (cf. (3.63)), where

(3.80) YJ(r,fJ,?)
=- 1-
a?,
2 7
+ 31 0 (m-
2wr
1).
and

The solution on the cylindrical discontinuity for r > Y* has the following
closed form:
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 326

where R* is determined b y . (3.70).

The problem in inelastic regions is the same as in the spherical case and
the difference equations are now (see Fig. 41)

z(;[ - y)+ y
q 2y@ - 1) - -
31 a,,
c1
- a,
r 1
(I-l.m.n+l)
Ar

C . Cylindrical S h a r Waves. Assume now that shearing tractions p(t)

are uniformly distributed on the cylindrical surface with radius r, of the
326 PIOTR PERZYNA

cavity in an infinite elastic/viscoplastic medium. In cylindrical coordinates

r,gl,z we have

(3.84) u, = @ ( I $ ) , u, = u, = 0.

The component of the shear stress is

(3.86) rrp, = rr&$),

whereas all remaining cylindrical components of stress vanish identically.

For the system (3.1) we now have

(3.86)

(fr.
The characteristic lines (3.4) are

(3.87) I = const, I = yo f At + const, where 1 =

along them, by (3.6), we have the relations

Using the conditions of kinematic and dynamic continuity

(3.89) v + A8w= 0,
(3.90) Apv + zr, = 0,
and the first relation (3.88), we obtain on the shear discontinuity the equation

(3.91)
 FUNDAMENTAL PROBLEMS IN VISCOPLASTLCITY 327

The iteration of the order n is

(3.03)

and

(3.94) v(nJ I- -r‘”’ 1

&(*I - -*(n)
p l w’
The condition for r* i s
(3.95) t&*) = k.

The solution on the shear discontinuity for Y >Y* is

In the inelastic regions we can again apply the method of finite

differences along the characteristic lines (see Fig. 41) by using
328 PIOTR PERZYNA

D. Plane Waves in a Half Space. Let 3, 7, 2 be Cartesian coordinates and

consider an elastic/viscoplastic medium occupying the half space f 2 0.
Suppose that the plane f = 0 is exposed to the uniform pressure T(t).
The displacement field is characterized by
(3.98) 243 = @(a$), t+= 248 = 0.

The strain tensor has only one component that is not identically zero, namely
(3.99) Eli = &aa(f,t),

and the normal stresses are independent of 7 and 2: ui3(f,t),agp(f,t)

= upm(f,t),
while the shearing stresses vanish identically.
The system (3.1) describes the plane problem in the inelastic regions,
where
. -
0

U , A=

. I
-1 0 0

r
(3.100)
0
1
B = p 5 y @ [ 3 ( u 3 3 2- VK5 eK i 3 ) - l

The characteristics (3.4) now take the form

(3.101) 5 = const, f = f& + const;
along them the following conditions must be satisfied by (3.6):

Using the condition of kinematic continuity

(3.103) err + -A1v = 0,

 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 329

the condition of dynamic continuity

(3.104) p b + Of3 = 0,
and the relation which is satisfied along the characteristic line 2 = 2,we
obtain the following equation on the plane discontinuity:

(3.106) (ILL = Po
j
- v[S,aLa(S)l&
0

where

(3.108) Y(Z,a,,) = A7 + y@ [q 1 - -+)ULL - I] *

and

(3.107)

Using iterations we may write the solution of Eq. (3.105) in the form

(3.108) uLL= lim a g ,

n d m

where

v(n)
1 (*)
= -- (4- 1 (n)
(3.110) a**I ELL- __ U L L .
PAa
The condition for f * (see Fig. 39) is

(3.111) VJa(x*) = k-
The solution on the plane discontinuous wave for f > 2* is
330 PIOTR PERZYNA

Denoting the intersection points of the characteristic net as shown in

Fig. 41, we can write the difference equations

4. Applicatiolt of the Method of Successive Approximations

Some problems of one-directional propagation of stress waves in an inelastic

medium described by the system (3.1) may be reduced to second order
partial differential equations. In these cases, the application of the method
of successive approximations is easier, and solutions of the very general
initial-boundary-value problems are available (cf, Section 1A and 1B).
It will be proved that some problems previously discussed may be reduced
to the solution of one of two general non-linear problems (see Section 1,
Theorems 1 and 2).
The method of successive approximations permits a full discussion and
examination of the convergence of the solution obtained.
As an example of application of the successive approximation method,
we shall study the problem of propagation of shear waves in an infinite
elastic/viscoplastic medium (see [139]).
The problem of propagation of shear waves in an infinite elastic/visco-
plastic medium was first treated by V. V. Sokolovsky [163]. Certain gener-
alizations were discussed in [130, 131, 1361. In all previous papers, the
solution was obtained by means of finite differences taken on the char-
acteristic net (compare with Section 3).
Assume now that shearing tractions p ( t ) or shear strain rates w(t) are
uniformly distributed on the cylindrical surface with radius ro of the cavity
in an infinite elastic/viscoplastic medium. In the cylindrical coordinates
r , cp, z, we have
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 331

(3.114) l((p = u(r,t), 24, = u, = 0.

The component of the shear stress is

(3.116) t = .r&.t).

whereas all the remaining cylindrical components of stress vanish identically.

- 13.
CascA.*Non-homogeneousmatmial; dinearficnctionQ)=@[(r/k(r))
It will be assumed that all functions describing the mechanical properties
of the material vary with the radius Y only. We then obtain the semi-linear
system of differential equations

(3.116) Uc + AU, + B = 0,
where

(3.117)

and either one of the following two boundary conditions

(3.118)

The system (3.116) is hyperbolic, hence the eigenvalues of the matrix A

(3.119) h ( d = f [ru(*)/p(~)l”e
are real. If shearing tractions p(t) or shear strain rate w(t) exceeding the
plasticity limit, p , >$, or a,,>a,,,are suddenly applied to the surface

The problems which will be studied in the Case A are much more general than those
discussed in Refs. [ISO, 131, lS5].
332 PIOTR PERZYNA

of the cavity, then the characteristic line r - r, - 4 7 ) t = 0 will be a strong

discontinuity (see Fig. 42).
By introducing new coordinates

I
r

(3.120) x =t + j A-'(l)dt, y = t - A-l(&)d€,

re

the equations of motion (3.116) valid in the elastic/viscoplastic region 9*

(see Fig. 42) may be reduced to the second order partial differential hyper-
bolic equation :

FIG. 42. The plastic region 9*in the l,r-plane.

where

(3.122)

and

(3.123)
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 333

The elastic/viscoplastic region 9*at Fig. 42 now takes the form of the
region 9 at Fig. 43. The region 9 is bounded by the characteristic y = 0
and the line x = y (not a characteristic).
The boundary conditions (3.118) now become

FIG.43. The plastic region 9 in the r,y-plane.

where

Thus, in either case the problem is reduced to the solution of a non-

linear boundary-value problem for a semi-linear hyperbolic equation (cf.
(3.20)).
Let us study in detail the second case of (3.124). Consider the linear
space Cgl of the functions z(x,y). The norm in this space has the form

(3.126) Il*(%Y)ll = SUP I*(%Y)l+ SUP I*x(xtr)l+ SUP I*Y(XIY)l.

9 9 9
Assuming that the function f given by (3.122), the integrand in (3.124,).
and the function h given by (3.125), satisfy the Lipschitz inequalities with
respect to z, t , and t,,,and the constants involved in these inequalities
satisfy the restrictions (3.9), we can write the mapping W for that problem
in the form
334 PIOTR PERZYNA

In practice, we shall have following iterative scheme :

t(r+&,Y) = @[w(xJ~I,

(3.128) ~ * ( , , , ) ( ~ S Y=) @z[t,,(x.Y)I,

rY(n+l)(X,Y) = WY [.ccn,(XJ41*

The above restrictions imply that (q,,)(x,y)}, {tx(,,,(x,y)}and {tY(,,) (x,~)}

converge uniformly to the continuous limit functions t ( x , y ) . t,(x,y) and
r,,(x,y), which constitute a solution of our problem.
To give an interpretation of these restrictions, and the limitations which
they impose on the class of the materials, let us first study the Lipschitz
conditions.
The functions f, h, and the integrand in (3.1%) satisfy the Lipschitz
inequalities if they have at every point of the region D bounded partial
derivatives with respect to 7 , t,, and t,,.Thus we have to assume that the
function @ [ ( t / k ( r ) )- 11 belongs to the class CBwith respect to [ ( t / k ( r ) -
) 11,
and that the functions describing the physical properties of the material
satisfy: [&), p ( r ) ] E C1, [ Y ( Y ) , ~ ( Y ) ] E Co, with respect to Y .

Case B. Non-homogeneous materid; linear fwwtiorr @ = (t/k(r))- 1.

In this case, the equation of motion takes the linear form
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 336

(3.130)
We seek the solution of (3.129), which satisfies the following boundary
conditions

(3.131)

Thus the problem in the region 9 (see Fig. 43) for the linear function @
is reduced to the solution of the special case of the linear problem (3.39).

Case C. Homogeneous material; linear function @ = t/ko- 1. This

case was earlier discussed by V. V. Sokolovsky [163], who solved it by the
finite difference method along characteristic lines.
For that case, the coefficients of (3.129) have the form

(3.133) b ( w )=-
3.0
2 { P o + &(x - r)l-l+ &Yofo/Q.
C ( X , y ) = 4&'[2y0 -k &(X - y)]-'.
The boundary conditions have a form similar to that of the previous case.
338 PIOTR PERZYNA

6. Solution i n the Elastic Region

Solutions in the elastic region by successive approximations have been

discussed in [130,131, 135, 1361. In this section we shall follow the results
presented in [135, 1361.
By introducing new coordinates

for spherical waves, cylindrical radial waves, and cylindrical shear waves,
and new coordinates

(3.136)

FIG.44. The elastic region E , in the x,y-plane.

(where x,, = const) for plane waves, the equation of motion in the elastic
region may be written in the form
A
(3.138) Wxy = ~

x- Y
(ux - M y ) + ( x -6y)2 @*

The considered elastic regions in the x,y-plane are shown in Fig. 44.
Equation (3.136) is valid in the regions El and E,.
Consider first the problem in the region El. This region is bounded by the
characteristic y = 0 and the curve x = ~ ( y ) .On y = 0, we have u = 0 ;
on x = q(y) in the cases of spherical waves, shear waves, and plane waves
the linear condition
(3.137) ci(%x -~ y + caw =
) %s

holds, and for cylindrical radial waves the nonlinear condition

(3.138) uy = X ( y ’ , W . h
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 337

where

must be satisfied.
Hence for all four types of waves the problem in region Elreduces to the
generalized Picard problem (see G. Majcher, [lOS]).
The solution of this generalized Picard problem may be expressed as
follows

(3.139)
i
U(%Y) = W , Y ;xo,q)Q(r])dr],
0

where V ( x , y ; l , q )denotes the Riemann function for (3.139) and xo = const.

It is obvious that the solution (3.139) satisfies the condition u = 0 on
y = 0. The function Q ( y ) should be determined so that the solution satis-
fies the conditions (3.137) and (3.138) respectively. We obtain a Volterra
equation of the second kind

(3.140)
0

where for spherical, cylindrical shear and plane waves

and for cylindrical radial waves

The solution of the integral equation (3.140)may be written in the form

(3.143)
338 PIOTR PERZYNA

where R(y,q) is the resolving kernel of Eq. (3.140). that is,

+c
oc
(3.144) RCy4) = J C r ( Y B l 7 ) n i l~ ( & J d D

with
Y

(3.146) J ' - ( n ) ( ~ ~=qI) ~ Y , O J ' - ( l)(E,r])dE,

~- "(0) = 4.
11

Thus in the case of the nonlinear condition (3.138). Eq. (3.140) is equivalent
to the equation

Equation (3.146) is a nonlinear integral equation with the unknown function

Q ( y ) . The method of successive approximations will be used to define the
functions
(3.147) Q ( o ) ( Y ) J w Y *) *3. ,Q(")(Y),'*
with the following recurrence formula for Q("+&J) :
FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 339
lh
+
eu CI
- 0
I I
El #-I
m o
I I
340 PIOTR PERZYNA

The Riemann function for Eq. (3.136) has the following closed form (see
R. Courant [64])
(3.149) Y(x,y;t,q)= ( x - Y)”(t - r)%- q)JF(- A- B;l,t),
where F(- 8,- B;l,() is the hypergeometric function and is determined
by the relation

(3.160)

and
(3.161)

The values of the constants A , 6, LO and p and the coefficients c,, cg and cg
are presented in Table 3. The case of a plane wave can be treated separately.
In this case the Riemann function V(x,y;{,q)= 1 and the solution is trivial.
In the region E, (Fig. 44) the problem for all four types of waves
may be reduced to a similar generalized Picard problem. But in this region
it is useful to apply the well-known Fourier transform method (see A. Kromm
[98] and G. Hopkins [82]).

6. Numerical Examples

We shall discuss here the numerical examples presented in the paper [136].
As a first example, the propagation of plane waves in the half space is
considered. In this case the geometric dispersion does not have such a great
influence on the solution as in the other cases. In the spherical problem the
convergence of the method of successive approximations, due to the geo-
metric dispersion, is so quick that the difference between the first and the
second iteration is practically insignificant, whereas in the case of plane
waves the first and the second iteration and for certain small regions of 2
even the next iterations can lead to different results (see curves 1-6 of
Fig. 46). Thus i t seems that in the case of plane waves the influence of the
work-hardening of the material may be studied more effectively.
For simplicity of the equations in practical applications, linear and
power-law forms of the function Q, and linear work-hardening of the material
have been assumed. Thus Q, has the form (cf. (2.88,))

(3.162)
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 341

the coefficient of the work-hardening K ( W+)for linear work-hardening

material may be determined by the relation (see [79], and [ 8 2 ] )
(3.163) .W,) = (mw, + l)k,
where m is a constant of the material depending on the tangent modulus
and the yield stress in simple shear k. The tangent modulus and the coefficient
of viscosity y have been determined from the experimental data of J, Harding,

0 4 6 8

FIG. 45. The nth iterations of vm versus curve.

E. 0. Wood, and J. D. Campbell [ 7 6 ] , while S = 1 and S = 3 have been

assumed. On the basis of the consideration of Section 3D,the solution
(3.108) for perfectly plastic and for work-hardening material has the same
closed form
(a) for linear @:

and
(b) for a power-law form of qZ

(3.166)
342 PIOTR PERZYNA

The data assumed for mild steel are collected in Table 4.

TABLE4

Upper yield Shear yield

2040 k G cm-' 1 180 k G cm-*
stress a, stress k

Young's Tangent
2.1 * l@k G cm-a 0.014062* loe k G cm-s
modulus E modulus

Density p 7.8. l e e k G seca ~ m - ~

Poisson's
0.29 y1 = 45Osec-'
ratio u
Viscosity yp = 680 sec-'
Bulk constants y3 = 736sec-'
1.663- 106 kG cm-* y, = 7OOsec-I
modulus K

Shear
0.820- l@kG cm-*
modulus p

-4

----- exact limr solution

e m ~ paner
t mlution

I I I I

Vm
'OO 2 4 6 8 1 0 2 -
, Curves
h ~46. versus I . Comparison of the exact linear solution with the exact
power solution.

In Fig. 46 the curves V x / k versus L at the discontinuity are given. Full lines
represent the results of the linear solution, broken lines correspond to the
power-law solution. Material constants y1 and ys have been assumed here
the same as in the case of a perfectly plastic material.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 343

The object of the second example is to find the curve (see Fig. 39) dividing
the plastic and elastic region. The influence of such parameters as duration
time of loading, the shape and sign of the loading curve, constant of material
y and work-hardening parameter K on the character of the curve P* is also
discussed. The comparison of the results for work-hardening and perfectly
plastic material has been presented.
Computing the third iteration by (3.65) and (3.66),we obtain the fol-
lowing result a t the discontinuity:

(9) 17
o,, = nya(R*$o)

- i@y2+ $,(l - D )

{ + 150 + 18)
- A?y,R*Po [(R*p,)9(D3+ t3Dz

- R*Po(3De+ 120 + 16)+ 3(D + 2 ) ] ~

- D [(R*$o)z(D' + 3 0 + - R*p0(2D + 3) + 1 rz
1
+ R*p,,Dz [R*po (D +):
I
- I] r3 -
1
(R*@,)'D44}

f 3flY$*$o{ [(R*$O)O(D' + 4 0 + 5, - 2R*$o(D + 2) + 11l o g y

- R*#o [R*$o(0 + 2) - l](logr)*+ 31 (R*fio)s(logrp
- 3nys(R*$o)'g {[ (. ):
R*$o f - I ] log 1

--
2
1
R*$,(log .Iz} I* + L ' @ ~ , ( R * $ ~ ) ~logD % ~
1,

(3.168)
where
(3.167) D=- *"
A
(R*$, - 1)8.
344 PIOTR PERZYNA

Eqs. (3.67) are employed for the determination of the quantities a*. v ,
E,,and eCC at the discontinuity.
Condition (3.67), according to (3.152) and (3.153) together with (3.166),
may now be written as follows:

(3.168) R*&)(r*) = 1.

L
tm m
I
.uo
I I
*w
I I I I L-

.1m PaJ rh
FIG. 47. Curves a,, versus v / r o on the discontinuous wave.

FIG. 48. Curve r*/v,, versus VJl(r,)/k for spherical wave.

In Fig. 47 we have two curves a,, versus r at the discontinuity for applied initial
pressures p , of 5800 Kg/cma and 3740 Kg/cma, all material data being assumed
as for the plane wave. The character of the curve r*/ro versus V x ( y 0 ) / k
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 345

is presented in Fig. 48. In the elastic/visc plastic region, i.e. region P* in

Fig. 39, the quantities arr,oPV,v , E,, and E~ are determined from (3.71).
The condition defining the curve r* in the plane is

and

A more convenient formula for calculating the parameter K may be used,

if we takz into account that

(3.162)
0

Using the constitutive equation valid in the plastic region

(3.163)

and the condition of plastic incompressibility

(3.164) 6; +2 4 = 0,
from (3.162), we obtain
1

a m - a,,
(3.165)
0

With the notation P(l,m,n) for the intersection point of the characteristic
net as in Fig. 41, the following recurrence formula for determination of the
parameter K for any fixed no has been employed:
346 PIOTR PERZYNA

49. Influence of the time duration on the character of boundary r*for linear form
FIG.
of preeeure.

Fx0.60. Influence of the time duration on the chsracter of boundary P for power form
of pressure.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 347

This way of obtaining the sth iteration has been chosen in the numerical
computations in order to satisfy the condition I K ( ~ + ' ) - &)I < 1. The
absolute value of an intensity a,, - oqqhas been taken because W , from the
definition of the plastic work, should be positive. The following pressure
functions have been assumed

where t = 2t/t, is half of the loading period.

FIG.51. Influence of the viscosity constant y on the shape of the curve r*.

In Figs. 49-51 and in Table 1 the results are presented. In Figs. 49 and 50
the curves r* for linear and quadratic form of P(r) and for various loading
periods have been plotted. I t is seen that if the period increases, the character
ofr* changes. The influence of the material constant y on the shape of the
r*
curve may be observed in Fig. 51. The difference is seen only at the
shock and in points lying close to it. From the results collected in Table 6
the small influence of the work-hardening effect is seen. However, it should
be taken into consideration that the hardening parameter K depends first
of all on the duration of the process of plastic deformation and on the amount
 TABLE5 w
%
AV = 0.04 AY = 0.01

- Po. perfectly - Po, work-hardening - Po, perfectly - Po, work-hardening

plastic material material plastic material material

t = 1.34576- 10-7
-4833.33 - 1406.65 - 1407.28 - 1436.07
1.00
1.04 - 4635.99 - 1382.42
-4833.33
-4635.80 - 1382.66
-4833.33
-4642.19
- 1435.70
- 1409.03
-4833.33
-4642.05 - 1409.18 z
1.08 -4451.33 - 1354.87 -4451.05 - 1354.92 -4463.17 - 1379.60 -4462.98 -11379.63 8
1.12 -4278.79 - 1325.46 -4278.49 - 1325.41 -4295.68 - 1348.64 - 4295.47 - 1348.61
1.10 -4117.69 - 1295.19 -4117.38 - 1295.10 -4139.03 -1317.03 -4138.80 -11316.97 8
1.20 -3967.28 - 1264.78 -3966.96 - 1264.67 -3992.47 - 1285.38 -3992.24 - 1285.31
1.24 -3826.77 - 1234.71 -3826.46 - 1234.59 -3855.27 - 1254.12 - 3865.05 - 1254.04
1.28 -3695.41 - 1205.31 -3695.11 - 1205.19 -3726.73 - 1223.54 -3726.51 -1223.46
1.32 - 3572.49 - 1176.82 -3572.20 - 1176.70 -3606.17 - 1193.83 -3605.96 - 1193.76
1.36 - 3457.33 - 1149.37 - 3457.05 - 1149.25 -3492.97 -1165.12 - 3492.76 - 1165.04
1.40 -3349.31 - 1123.06 -3349.03 -1122.95 -3386.53 - 1137.47 -3386.34 - 1137.40
1.44 -3247.84 - 1097.95 -3247.57 - 1097.85 - 3286.33 - 11 10.92 - 3286.14 -1110.85
1.48 -3152.40 - 1074.07 -3152.14 - 1073.96 -3191.79 - 1085.45 -3191.60 - 1085.38
1.52 -3061.72 - 1051.10 -3061.41 -1051.00 -3102.50 - 1060.27 -3102.32 - 1060.20
t = 2.69152- 1 0 - 7
1.00 -3866.67 -541.45 -3866.67 -543.27 -3866.67 -604.02 -3866.67 -605.10
1.04 - 3691.25 - 575.89 -3690.98 - 576.67 -3698.83 - 633.44 - 3698.66 -633.90
1.08 -3527.76 -598.05 - 3527.39 - 598.34 - 3542.21 -651.28 - 3541.97 -651.41
 TABLE5 (continued)

Ar = 0.04 AY = 0.01
-
- Po, perfectly - p,, work-hardening - Po. perfectly - pc work-hardening
plastic material material plastic material material

Qw Qw UY, aw QW =w Qw

1.12 -3375.94 -611.45 - 3375.55 -611.51 -3396.45 - 660.93 - 3396.20 -660.97

1.16 -3235.20 -618.59 -3234.81 -618.55 -3260.97 -664.76 -3260.72 -664.74
1.20 -3104.79 -621.27 -3104.41 -621.18 -3155.04 -664.47 -3134.79 -664.42
1.24 -2983.89 - 620.79 -2983.53 - 620.68 -3017.91 -661.27 -3017.67 -661.21
1.28 -2811.70 -618.10 -2871.34 -617.98 -2908.85 - 656.05 - 2908.62 - 655.98
1.32 -22761.42 -613.88 - 2767.08 -613.76 - 2807.16 -649.45 -2806.94 - 649.38
1.36 - 2670.36 - 608.67 - 2670.02 - 608.55 -27 12.19 -641.93 -2711.97 -641.87
1 = 4.03728. 10-7
1.00 -2900.00 239.89 - 2900.00 236.83 - 2900.00 140.21 -2900.00 138.56
1.04 -2748.47 149.55 -2748.28 148.24 - 2757.62 59.69 -2757.51 56.99
1.08 - 2608.56 81.41 -2608.33 80.86 -2625.84 -0.29 -2625.70 -0.58
1.12 -2480.06 29.16 - 2479.83 28.95 - 2504.35 -45.72 -2504.21 -45.84
1.16 -2362.27 -11.54 - 2362.04 -11.60 -2392.47 -80.66 -2392.33 -80.71
1.20 -2254.28 -43.73 -2254.06 -43.74 -2289.39 - 107.92 -2289.26 - 107.93
1.24 -2155.15 - 69.59 -2154.94 -69.57 - 2 194.30 -129.47 -2194.17 - 129.46
t = -
5.38304 lo-’
1.00 - 1933.33 932.58 - 1933.33 928.71 - 1933.33 795.32 - 1933.33 793.43
1.04 -1808.36 792.31 - 1808.32 790.71 - 1819.27 671.37 - 1819.23 670.57
1.08 - 1695.10 683.97 - 1695.06 683.27 - 1715.33 575.71 -1715.28 575.36
1.12 - 1592.92 598.01 - 1592.90 697.68 - 1620.88 499.77 - 1620.83 499.60 w
P
(D
360 PIOTR PERZYNA

of pressure applied. In our example, the duration of pressure application

is very short, and the process of plastic deformation is not yet fully developed.
The comparison of the results of the work-hardening and perfectly-
plastic theories shows that in practical applications the influence of the
work-hardening may be neglected at least for certain regions of strain rate
and for initial plastic deformation.

IV. QUASI-STATIC
SOLUTIONS

1. Spherical Problem

The quasi-static problem of a thick-walled spherical container with an

elastic/viscoplastic material has been studied in Refs. [191, 1941. Two
cases of boundary conditions have been treated. In the first a constant
pressure p is assumed and in the second a constant displacement uo, both
prescribed on the interior surface of the sphere.
We will consider in detail the first case. The full system of differential
equations involving equation of equilibrium and constitutive equations,
according to (3.1) and (3.49). has the form

a ~ ,
--+2
-I oprp=
a 8

where the static yield function F is defined by F = V x / k - 1.

There exists a closed-form solution of the system (4.1), if we assume
the linear function @(F)= F and a rectangular pressure pulse

(4.2) t )pH(&
~ , ~ ( a ,= u,,(b,t) = 0,
where a and b denote the inner and outer radii respectively and H ( t ) is the
Heaviside function.

[(:r-
The solution of the system (4.1) by means of Laplace transform leads
to the following formulae

y a k=
F - +(iy 11 -'[- p + 2k(l - e ~ p ( -m*t)) log-a
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 361

where

(4.6)

Quasi-static

01
FIG. 62. Curves vJu/k versus v/a for different time parameters (Ref. [194]).

The formulae for the components of the stress tensor a,,, uw and uee are
given in [191], where the influence of viscosity and time on the stress distri-
bution are also discussed. The interaction between elastic and inelastic
362 PIOTR PERZYNA

components of the stress tensor causes a certain levelling off in the distribution
of J 2 along the radius r , Fig. 52.

2. Viscoplastic Flow of a Circdar Plate

E. J. Appleby and W. Prager [4] have treated the viscoplastic flow of

a circular plate that is simply supported along its edge and subjected to a
uniformly distributed transverse load. The material of the plate is supposed
to be incompressible and rigid/viscoplastic according to the linearized theory
proposed by W. Prager [146] (see Chapter 11, Section 12). Thus the
solution has been based upon the Tresca yield condition. A similar problem
based upon the Huber-Mises yield condition has been treated in Ref. [193].
We follow here the presentation given there.
The plate remains undisturbed if the applied pressure p(t) does not reach
the value of load carrying capacity p = 6.51 M,/R2, (see for instance [81]),
where R is the radius of the plate and M , = a,h2 the fully-plastic yield
moment. The thickness of the plate is denoted by 2h and the yield stress
in uniaxial tension by a,. The consequence of the assumed model of the
material is that the load intensity 9' = p(R2/M,) can exceed the value
$' = 6.51 and the displacement rate and displacement fields can be
uniquely determined.
In the cylindrical coordinate system r , q, z , ( z vertically downward)
the only nonvanishing components of the stress tensor are the radiaI and
circumferential stresses a,, and app. The shearing stresses ,z, and t,, vanish
in view of rotational symmetry. Since the thickness of the plate is supposed
to be small as compared with the radius R , u, and z, will be small compared
with a,, and a,,. We assume that the shearing stress t, does not enter into
the constitutive equations.
The generalized stressesare the radial and circumferential bendingmoments
M, and M , and the corresponding generalized strain rates are the radial and
circumferential rate of curvature i , and ip.
The whole analysis is carried out within the theory of thin plates. Ac-
cording to the Love-Kirchhoff hypotheses the rate of strain is related to the
rate of curvature of the middle surface by

Since the deflections of the plate are supposed to be small a linear relation
between rate of curvature and rate of deflection is assumed:

&=--
a% 1 aw
(4.7) up=---
r8r
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 363

If the inertia effects are neglected the equation of equilibrium takes the
form

The constitutive equations for incompressible strain-rate sensitive rigid-

plastic material are used [cf. (241:

(4.9) for F>O. dji = 0,

where F = V&/k - 1.
The principal idea of Eq. (4.9) is that the strain rate should be in general
a non-linear but uni-valued function of the excessive stresses above the
yield surface. Two particular types of the function @(F)will be considered,
namely, the linear function @(F)= F and the power function @(F)= Fd.
In the case of the power function the constitutive equations (4.9) give two
independent equations

(4.10)

where J a = 4, - a,,uw + 4, and y = y0/2k. All components of the stress

and strain-rate tensors are functions of two coordinates r and z. It is therefore
desirable to transform the constitutive equations (4.10) to the generalized
stress and strain-rate space where all quantities are functions of a single
space coordinate Y . The bending moments M , and M, will be expressed as

(4.11)

The stresses a,, and aw can be evaluated from (4.10) and substituted into
(4.11). Taking account of the relations (4.6) the integration can be performed
for any integer exponent 6. This leads to the following formulae relating
rate of curvature k, and kp with the radial and circumferential bending
moments :
364 PIOTR PERZYNA

ri, = B{ [1 - Mo(M,' - M,M, + M,')- "7 [(M,'- M,Mp + M,')l/e/Mo]d -

' (2MT - M,)/MO}#
rip= B{ [l - Mo(M,' [M,' - M&,
- M,Mp -I- M,')-'~2]d + M,')'/'/M '-'
01

' (2M, - M,)/MO},

(4.12)
where the constant B is
(4.13) B = y/1/5h[(26 + 1)/26]d.
Note thatmo formal analom. between (4.10) and (4.12) exists. Equations
(4.12) involve a new term which vanishes only for the linear function, ie.,
for 6 = 1. This example shows that particular attention should be paid
in constructing a stress strain-rate relation appropriate for the generalized
quantities.
It is convenient to introduce dimensionless variables
M, M w
(4.14) m,=-M o '
Mo' P=B* v=- BR'
We have now five equations with five unknown functions m,, mvJv , k, and kv
However only three of them are of interest to us. These functions are m,,
m, I and v. After eliminationof k, and kpthe followingsystem of three ordinary
quasi-linear differential equations is obtained

(4.16)
 FUNDAMENTAL PROBLEMS IN VIXOPLASTICITY 366

At the centre of the plate 7 = 0, m, = mI by rotational symmetry. At the

simply supported edge Y = R the radial bending moment and rate of deflec-
tion vanish. Thus, the boundary condition can be written in the form

FIG.63. Curves m, and %versus i; for b = 1and several values of the loading paramebr
p’ (Ref. [193]).
358 PIOTR PERZYNA

(4.17) m,(O) = m,(O), m,(l) = 0, v(1) = 0.

The computations were carried out for two values of the exponent b and for
several values of the load intensity 9'.

24 t mrtmcp

6-3

FIG.
84. Curves m, and n"p, versus for d = 3 and several values of the loading parameter
p' (Ref. [l03]).

Figures 6366 present the moment distribution and the corresponding

velocity fields for all values of p' and b listed in Table 6. It was thought
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 357

desirable to compare the moment and velocity distribution for the case of
non-linear and linear function @(F). This has been done in Figs. 67
and 68, where the solid line refers to the case 6 = 3 and the broken line
corresponds to the case. 6 = 1. Note that the moment distributions do not

c
P

FIG.56. The velocity fields for 6 = 1 and different values of the loading parameter p‘
(Ref. [193]).

FIG.68. The velocity fields for B =3 and different values of the loading parameter p’
(Ref. [193]).

differ noticeably, but the difference in the rate of deflection is appreciable.

This confirms the supposition that the rate of deflection is far more sensitive
to the change in the function @(F) than the moment distribution.
Most numerical results in [193] were obtained for the linear function
@(F) = F . This has been done to get a detailed comparison of the solution
based upon the Huber-Mises yield condition with the solution of a similar
368 PIOTR PERZYNA

TABLE6

P'

22-

10 -

ae-

06-

04-

02-

0
,

a2
, ,

0.4
I I

06
, I

ne
l l

1.0 F
-
FIG.57. Comparison between the distributions of the moments m, and % for non-linear
and linear function @(F) (Ref. 11981).

problem based upon the Tresca yield condition, given by E. J. Appleby

and W. Prager [4]. The linearization of the yield locus introduces much
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 369

simplification to the governing equations, and the corresponding solution is in

closed form. In this case the plate is divided into three regions that are in
different visco-plastic regimes, the radii p = a, and p = a, being the boun-
daries. A solid line in Figs. 69 and 60 represents the solution obtained in
paper [193] for a chosen value of 9' = 10, while the broken line plots the
corresponding solution of E. J. Appleby and W. Prager. There is an excellent
coincidence in both moment and rate of deflection distribution. The broken
line in Fig. 60 does not exhibit a noticeable deviation from the solid line either
in shape or even in the absolute value. This result is somewhat unexpected.
It is known from the theory of perfectly plastic solids that the linearization
of the yield criterion leads to a fairly good estimate of the stress fields,
while the corresponding displacement fields are usually unrealistic. This is due
to the piecewise linear yield condition since one component of the strain-rate

FIG.68. Comparison between the velocity distributions for non-linear and linear function
@(F) (Ref. [l93]).

tensor has a constant direction for each region while the others vanish. On
the other hand in the majority of cases for visco-plastic material there are
two or more non-vanishing components of the strain-rate tensor and the
direction of the strain-rate tensor is no longer constant. I t can be concluded
that, at least for simply supported circular plates, the linearized theory of
viscoplasticity due to W. Prager [146] may give a close qualitative and
quantitative assessment of the deformation under a given condition of
loading. Of course the above statement relates only to the linear function
@(F) for which the comparison was carried out.
The above procedure involves the assumption that the deflection of the
plate is small. It is straightforward to establish upper bounds on the values
of pressure and duration of impulse so as to remain within the limits of the
theory.
360 PIOTR PERZYNA

A certain peculiarity of the governing equations is the fact that the

solution in dimensionless quantities depends neither on the dimension of
the plate nor on the constants of the material a,, and y . However, the in-
fluence of the type of function @(F) is fundamental.

08 -

06

a4

a2 -
-

-
-Huber -Mises
---- Tmca
1 I
I
I

FIG.69. Comparison between the distributions of the moments m, and m, for the Huber-
Mises and Tresca yield functions (Ref. [193]).

Since the applied pressure p(t) was assumed to be constant the time
is formally eliminated from the solution. Although there are no restrictions
to extending of the solution over the range of variable pressure, a sufficiently
rapid time variation of the load intensities would require consideration of
inertia effects. The dynamics of a visco-plastic circular plate has been con-
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 361

sidered in Ref. [196]. The treatment makes the same assumptions as in the
quasi-static problem, but the transverse inertia motion is taken into ac-
count. It is shown that the dynamic flow of a viscoplastic circular plate
is described by an initial- and boundary-value problem for a quasi-linear
parabolic system.

W/BRp
FIG.60. Comparison between the velocity distributions for the Huber-Mises and Tresca
yield functions (Ref. [193]).

Other solutions for the quasi-static flow of a visco-plastic circular plate

may be found in Refs. [22, 741.

V. OTHERDYNAMICAL
PROBLEMS

1. Wave Problems for Rods and Beams

The problem of the propagation of the longitudinal plastic waves in

rods of strain-rate dependent material, including effects of lateral inertia
and shear, has been considered by H. J. Plass [141]. A similar problem for
elasticlviscoplastic beams, i.e., the problem of propagation of bending and
transverse waves has been treated by L. V. Nikitin [120].
The solution of the wave problem of infinite and semi-infinite elastic/
viscoplastic beams by the finite-difference method has been discussed in
Ref. [8]. In the equations of motion of a beam the effects of shear and
rotatory inertia are included. The material of a beam is described by the
constitutive equations (2.67).
362 PIOTR PERZYNA

Under the assumption that the propagation velocities of the moment

and shear waves are the same, the wave problem for the elastic/viscoplastic
beam may be solved by successive approximation (cf. Ref. [S]).

2. Impulsive Loading of a S9herical Container

Consider a sphere of rigid/viscoplastic material and denote inner and

outer radii by a and b, respectively. Let a time-variable pressure p(t) be
applied on the surface Y = a. Assuming in (3.1) and (3.49) p + 00 and
K .+ m, we obtain the system of differential equations

a v u
r+2r=0.

We shall assume the power law @(F)= Fd [cf. (2.88)] and arbitrary
pressure +(t). The boundary and initial conditions have-now the fo&
(6.2) u,,(a,t) = - p(t). u,(b,t) = 0, u(r,O)= 0.
In this case the general solution of (6.1) given in Ref. [192] has the
following form

8
(6.4) v = (m3)7a(;) At),
where y ( t ) is determined by the differential equation

dY + Ay'ld
- = P(L),
(6.6)
dt
with initial condition y(0)= 0 and
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 303

Using successive approximations, the solution of (6.6) may be given in

the form

(5-7) Y V ) = lim Y(")(t),

U-PW

where y,)(t) is determined by a recurrence formula

Dynamic

t"
-\
'L
0 t

t-a I
I
I
I
t
I
I
I
I
1
I I I 5
9.I .
II 23 q4 25 r/a
FIG. 61. Dynamic curves VJTversus r/a for different time parameters (Ref. [lQ4]).
If the function @(F) is linear, i.e., 8 = 1, (6.6) has a closed-form solution
364 PIOTR PERZYNA

The time dependence of J&) for rectangular shape of the loading curve
and the linear function @(F) are plotted in Fig. 61,
In Ref. [192] the formulae for the components of stress a,,, a=, am,
strain E”,, eW, E80 and strain-rate tensors are given and the solutions for other
types of the function @(F) are discussed.
In order to compare the dynamic and the quasi-static solutions (cf.
Chapter IV, Section l), the same material and identical boundary conditions
in both cases should be assumed (see Ref. [194]). Consider the rigid/visco-
plastic material, the rectangular impulse and the linear function @(F) = F.
The corresponding solution of the quasi-static problem can be obtained
by putting into formulae (4.3)-(4.6) ,u -,00 and K --* 00 with the result

(6.10) v x / k=1 +

The particular case of the dynamic solution given by (6.3). (5.4),(6.6),

r] (g
and (6.9) for 8 = 1 and p(t) = const, yields

(6.12) V&.k = 1 + (- [ (-
1- - + 3 log 11 - exp (- ~ t ) ] ,

(6.13) zt(r,t)
1[7:(= -(); ]3 -1
yi(f+2vrlogz) b

In both cases the pressure p has to exceed the minimum value

b
(6.14) #min=2v3-~10g-.
a

From the comparison it is evident that the quasi-static solution is an

asymptotic solution for the dynamic problem. The final results of these
solutions differ from each other only in terms involving a function of time.
Thus the distribution of stresses and displacements along the radius of
sphere is the same for the dynamic and quasi-static problems.
The above conclusions are illustrated in Figs. 62433 for a mild-steel
spherical container and a/b = 1.6. In these figures the dashed lines are
plotted according to (4.3)and (4.4).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 365

FIG. 62. Comparison of the quasi-static with the dynamic solution (Ref. [1@4]).

FIG.63. Comparison of the quasi-static with the dynamic solution (Ref. [ 1941).

3 . Strain-Rate Sensitive Beams wader Im+act

The elementary dynamic rigid-plastic theory is characterized by the

neglect of elastic deformations, strain hardening, strain-rate sensitivity, shear
deformation, and geometry changes associated with large deformations.
Rapid loading tests on mild-steel beams have been performed by E. W.
Parkes [129] and by T. J. Mentel [114]. Their experimental deformation
366 PIOTR PERZYNA

values were consistently below theoretical predictions by factors ranging

from about 0.3 to 0.8. They showed that approximate corrections for strain-
hardening and dependence of the yield stress on strain rate could be made
so as to bring the calculated deformations close to their experimental results.
G. R. Cowper and P. S. Symonds [65] have treated this problem more
completely with strain hardening and strain-rate sensitivity included sep-
arately. In this treatment a strain-rate law (2.89) has been used, where
the constants have been chosen to agree as closely as possible with the
experimental results.

Experimental Summa
Cum C-SsOO(~ -1)7ax-'
C U M Of FhrkeS -
---
I I 1
c
9-8 a-g W" 1 8
strain mte ~,sec.?
FIG.64. Effect of rate of straining on yield stress ($5. R. Bodner and P. S. Symonds [20]).

A more general discussion of the strain-rate effect on impulsive loading

has been given by S. R. Bodner and P. S. Symonds [19, 201 and by T. C.
T. Ting and P. S. Symonds [178].
The paper [20] presents the results of a test program on the dynamic
loading of cantilever-beam specimens. These tests have been designed to
evaluate the relative importance of the various factors that are neglected
in the rigid-perfectly plastic theory of beams. The most important result
of this work is the conclusion that the rigid-perfectly plastic theory can serve
as a reasonable first-order theory as long as the energy ratio 9 (of the kinetic
energy imput to the maximum possible elastic energy) is not too small
(at least greater than 3). Elastic vibrations do not have much effect on the
results when the energy ratio 9 is greater than about 10. For sufficiently
large S, the trend of the test results indicates that the influence of strain
 IWNDAMENTAL PROBLEMS IN VISCOPLASTICITY 367

rate on the yield stress was primarily responsible for the deviations between
theory and experiment. In theoretical treatment S. R. Rodner and P. S.
Symonds [20] have assumed that the influence of plastic strain rate E upon
yield stress 0 obeys the relation (2.89). The numerical values for y and 6
have been deduced from the experimental data of M. J. Manjoine [112] for
mild-steel specimens, and from experimental values collected by E. W. Parkes
[129] for various aluminium alloys (see Fig. 64). The predictions of the
rate-dependent, rigid-plastic theory are generally in satisfactory agreement
with test results for the final deformation (shape and magnitude), deformation
time, and strain time-history. On the other hand the application of an
overall strain-rate correction factor on the final deformation (cf. T. J. Mentel
[lla]) cannot be generally recommended since it may lead to serious errors.
The analysis in Ref. [20] depends on several special assumptions. In
particular, it has been assumed that the plastic region is finite but small
in length compared to the beam length and the deformations small enough
so that geometry changes could be ignored. Comparisons both within the
theory and with test results have shown that these assumptions are of
doubtful validity in the range of interest. They are not made in the analysis
given by T. C. T. Ting and P. S. Symonds [178] (see also T. C. T. Ting [189,
1901). This study shows that final plastic deformations are in good agreement
with those measured in tests performed by S. R. Bodner and P. S. Symonds
[201.

4. Longitudinal Impact on Viscoplastic Rods

The prablem of plastic deformations in a cylinder striking a rigid target

has been undertaken by several authors (see for instance G. I. Taylor [172],
E. H. Lee and S. J. Tupper [202]). An exact analysis of this problem based
on the theory of elastic and plastic wave propagation has been given by
E. H. Lee and S. J. Tupper [202].
In several papers t!ie attention has been confined to cases in which
plastic strains are much larger than elastic strains, and the validity of rigid-
plastic theory has been assumed. Such treatment is motivated primarily
by the need to understand the essential features of the fields of stress, strain,
and strain rate in a specimen undergoing plastic deformation as a result of
high-speed impact. When a test of this type is performed in order to gain
knowledge of material behavior at high strain rates, the results will be
meaningful only if these essential features and the stress and strain history
of typical elements of the specimen are reasonably well known (cf. A. C.
Whiffin [189] and T. C. T. Ting and P. S. Symonds [ I N ] ) .
An approximate analysis of a cylinder under impact load, disregarding
the elastic strains, has been presented by G. I. Taylor [172] and by E. H. Lee
368 PIOTR PERZYNA

and H. Wolf [104]. These considerations have shown the very important
influence of the strain-rate effect on the deformation.
A rigid-(linearly viscoplastic) formulation of the theory for longitudinal
impact on a bar has been applied by V. V. Sokolovsky [lea] to several
problems of plane shear waves in a semi-infinite medium. Known solutions
of the heat equations were used.
T. C , T. Ting and P. S. Symonds [181] in the analysis of the longitudinal
impact on rigid/viscoplastic rods have used the linear stress-(strain rate)
law (cf. (2.93,))
(5.15) c = y*(a - ao).
Four cases have been solved : constant-velocity impact on a semi-infinite
bar, constant-velocity impact on a bar of finite length, impact of a finite
mass on a semi-infinite rod and impact of a finite mass on finite rod. In
every case the problem has been reduced to the solution of the heat equation.
G . I. Barenblatt and A. Y. Ishlinsky [6] have extended Taylor’s problem
to the viscoplastic material.
In Ref. [7] successive approximations have been suggested as a method
of analysis for the deformation of a strain-rate sensitive plastic cylinder
and the theoretical results have been compared with Whiffin’s experimental
data. Discussions of similar problems may be found in Refs, [loo, 101.
106, 150, 153, 190, 2001.

References
(Titles of Russian publications are translated).

1. ALDER, J. F., and PHILLIPS,V. A., The effect of strain rate and temperature on
the resistance of aluminium, copper and steel to compression, J . Inst. Metals 88.
80-86 (1954-55).
2. ALEXEIEV, N. A,, RAKHMATULIN, KH.A., and SAGAMONIAN, A. Y.,On the fundamental
equations for dynamics of soil, Zhuvlzal Prik. Mekh. Tekh. 8, 147 (1963).
3. ALTER,B. E. K., and CURTIS,C. W., Effects of strain rate on the propagation of
a plastic strain pulse along a lead bar, J . A p p l . Phys. 27, 1079-1085 (1956).
4. APPLEBY, E. J., and PRAGER, W., A problem in visco-plasticity, J . A p p l . Mech. 29,
381 - 384 (1962).
5. A m , A. K., and DIAZ,J. B., On a mixed boundary-value problem for linear hyper-
bolic partial differential equations in two independent variables, Avch. Rational
Mech. Anal. 10, 1-28 (1962).
6. BARENBLATT, G. I., and ISHLINSKY, A. Y., On the impact of a visco-plastic bar
on a rigid target (in Russian), Pvik. Mat. Mekh. 26, 497-602 (1962).
7. BEJDA,J., Analysis of deformation in a short visco-plastic cylinder striking a rigid
target, Avch. Mech. Stos. 16, 879-889 (1963).
8. BEJDA, J.. T h e wave problem of elastic/viscoplastic beams, J. MCan. (1966)
(forthcoming).
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 369

9. BEJDA,J., The method of successive approximation applied to the wave problem

of elasticlviscoplastic beams, Arch. Mech. Stos. 17, 711-726 (1965).
10. BELL, J . F., Propagation of plastic waves in pre-stressed bars, The Johns Hopkins
University, Baltimore, Tech. Rep. no. 5 , 19.51.
11. BELL,J. F., Diffraction grating strain gauge, Proc. Soc. Experimental Stress Analysis
17, 51-64 (1960).
12. BELL,J. F., Propagation of large amplitude waves in annealed aluminium, J . APpZ.
Phys. 81, 277-282 (1960).
13. BELL, J. F.. Study of initial conditions in constant velocity impact, J . Awl. Phys.
81, 2188-2195 (1960).
14. BELL,J. F., Experimental study of the interrelation between the theory of disloca-
tions in polycrystalline media and finite amplitude wave propagation in solids,
J . APPZ. Phys. 89, 1982-1993 (1961).
15. BELL, J. F., An experimental study of the unloading phenomenon in constant
velocity impact, J . Mech. Phys. Solids 9, 1-15 (1961).
16. BELL, J. F., Further experimental study of the unloading phenomenon in constant
velocity impact, J . Mech. Phys. Solids 9, 261-278 (1961).
17. BELL, J. F., Experimental study of dynamic plasticity at elevated temperatures,
Ex$erimeniaZ Mechanics, June 1962, 1- 6.
18. BLAND,D. H.,“The Theory of Linear Viscoelasticity.” Pergamon Press, New York
1960.
19. BODNER,S. R., and SYMONDS, P. S., Plastic deformations in impact and impulsive
loading of beams, i n “Plasticity” (E. H. Lee and P. S.Symonds, eds.), Pergamon
Press, London, 1960, pp. 488-500.
20. BODNER,S. R., and SYMONDS. P. S.. Experimental and theoretical investigation
of the plastic deformation of cantilever beams subjected to impulsive loading,
Technical Report No. 71, Brown University, J u l y 1961; J . A$$Z. Mech. ‘29,
719-728 (1962).
21. BRAGG,L. E., The thermodynamical limitation on compressibility, J . Math. Phys. 4,
1074-1079 (1963).
22. BYKOVCEV, G. M., and SEMYKINA, P. D.. On visco-plastic flow of circular plates
and shells (in Russian), Imesiiya Akademii Nauh SSSR, 4, 68-76 (1964).
23. CAMPBELL, W. R., Determination of dynamic stress-strain curves from strain waves
in long bars, Proc. SOG. Experimental Stress Analysis 10, 113-124 (1952).
24. CAMPBELL, J. D.. An investigation of the plastic behavior of metal rods subjected
t o longitudinal impact, J . Mech. Phys. Solids 1, 113-123 (1953).
25. CAMPBELL, J . D., The dynamic yielding of mild steel, Acta MelaZZurgica 1. 706-711
(1953).
26. CAMPBELL, J . D., The yield of mild steel under impact loading, J . Mech. Pkys.
Solids 8, 54- 62 (19.54).
27. CAMPBELL, J. D., and DUBY, J.. The yield behaviour of mild steel in dynamic
compression, Proc. Royal Soc. 886A. 24-40 (1966).
28. CAMPBELL, J. D., and DUBY,J,, Delayed yield and other dynamic loading phenomena
in a medium-carbon steel, i n “Proc. of the Conference on the properties of
materials at high rates of strain, London 1957,” p. 214, Inst. Mech. Engrs.,
London, 1967.
29. CAMPBELL, J. D., and MAIDEN,C. J., The effect of impact loading on the static
yield strength of a medium-carbon steel, J . Mech. Phys. Solids 6, 53-62 (1957).
30. CAMPBELL, J. D., SIMMONS, J. A., and DORN,J. E., On the dynamic behaviour
of a Frank-Read source, J. A$$l. Mech. 88, 447-452 (1961).
31. CAMPBELL.J. D.. and HARDINC,J.. The effect of grain size, rate of strain, and
neutron irradiation on the tensile strength of a-iron, in “Response of Metals to
High Velocity Deformation.” Interscience,’ New York, 1961, pp. 51 -76.
370 PIOTR PERZYNA

32. CAMPBELL, J. D.,and MARSH,K. J,, The effect of grain size on the delayed yielding
of mild steel, Philosofihical Magazine 7, 933- 952 (1962).
33. CHADWICK. P.,Cox, A. D., and HOPKINS,H. G., Mechanics of deep underground
explosions, Phil. Trans. Roy. SOC.London 2628A. 235-300 (1964).
34. CHIDDISTER, J . L., and MALVERN, L. E., Compression-impact testing of aluminium
at elevated temperature, Ex$. Mech. 8, 81-91 (1963).
35. CHU,S. C.,and DIAZ,J. B., Remarks on a mixed boundary-value problem for linear
hyperbolic partial differential equations in two independent variables, Arch. Ra-
tional Mech. Anal. 16, 187-196 (1964).
36. CINQUINI-CIBRARIO, M., Teoremi di unicitd per sistemi di equazioni a derivate
parziali in pih variabili indipendenti. Ann. di Mat., 48, 103-134 (1959).
37. CINQUINI-CIBRARIO, M.. Sistemi di equazioni a derivate parziali in pi& variabili
indipen’denti, Ann. di Mat. .44, 357-418 (1957).
38. CLARK, D.S.,and WOOD,D. S., The time delay for the initiation of plastic deforma-
tion at rapidly applied constant stress, Proc. A h . Soc. Testing Materials 40.
717-737 (1949).
39. CLARK, D. S.,and DUWEZ,P. E., The influence of strain rate on some tensile prop-
erties of steel, Proc. Amer. SOC.Testing Materials 60, 560-575 (1960).
40. CLARK,D. S.. The behavior of metals under dynamic loading, Trans Amer. SOC.
Metals 46, 34-52 (1954).
41. COLEMAN, B. D.,and NOLL,W., On the thermostatics of continuous media, Arch.
Rational Mech. Anal, 4, 97- 128 (1959).
42. COLEMAN, B. D., and NOLL.W., Foundations of linear viscoelasticity. Reviews oj
Modern Physics 81. 239-249 (1961).
43. COLEMAN, B. D.,Mechanical and thermodynamical admissibility of stress-strain
functions, Arch. Rational Mech. Anal. 9, 172-186 (1902).
44. COLEMAN,B. D.,and NOLL, W.,Material symmetry and thermostatic inequal-
ities in finite elastic deformations, Arch. Rational Mech. Anal. 16, 87-111
(1964).
45. CONLAN, J., The Cauchy problem and the mixed boundary-value problem for a
non-linear hyperbolic partial differential equation in two independent variables,
Arch. Rational Mech. Anal. 8, 355-380 (1959).
46. CONLAN, J., A generalized boundary value problem for uxy = f(r,y , u, uz, u,,),
Arch. Rational Mech. Anal. 18, 137-146 (1963).
47. COPSON,E. T., On the Riemann-Green function. Arch. Rational Mech. Anal. 1,
324-348 (1968).
48. COTTRELL, A. H., and BILBY,B. A., Dislocation theory of yielding and strain ageing
of iron, Pvoc. Phys. SOC.62A, 49 (1949).
49. COTTRLLL,A. H.,Deformation of solids at high rates of strain, an ‘eProc.of t h e
Conference on the properties of materials at high rates of strain, London 1957,”
pp. 1-12, Inst. Mech. Engrs., London (1957).
50. COURANT, R.,and FRIDRICHS, K. O., “Supersonic Flow and Shock Waves.” Inter-
science, New York. 1948.
61. COURANT, R., and LAX, P. D., Nonlinear partial differential equations with two
independent variables, Comm. Pure A$Pl. Math. 2, 255-273 (1949).
52. COURANT, R., ISAACSON, E., and REES,M., On the solution of non-linear hyperbolic
differential equations by finite differences, Comm. Pure Appl. Math. 6, 243- 255
(1962).
63. COURANT, R., Cauchy’s problem for hyperbolic quasi-linear systems of first order
partial differential equations in two independent variables, Comm. Pure Appl.
Math. 14, 257-2285 (1961).
64. COURANT, R., and HILBLRT,D., “Methods of Mathematical Physics,” vol. 2.
Interscience, New York, 1962.
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 37 1

55. COWPER, G. R., and SYMONDS, P. S., Strain hardening and strain rate effects in
the impact loading of cantilever beams, Technical Report No. 28, Brown
University, September 1957.
56. CRISTESCU, N., Some problems of the mechanics of extensible strings, stress wave
in anelastic solids, i n “I.U.T.A.M. Symposium, Brown University, April 1963”
(H. Kolsky and W. Prager. eds.). Springer-Verlag, Berlin. 1964, pp. 118- 132.
57. CRISTESCU. N . . Some dynamic problems of one-dimensional elastic/visco/plastic
bodies, Eleventh International Congress of Applied Mechanics, Munchen,
August 30 to September 5. 1964, (forthcoming).
68. DHUTLER,H., Experimentelle Untersuchungen hbex die Abhangigkeit der Zug-
spannungen von der Verformungsgeschwindigkeit, Phys. Z . 81,247- 259 (1932).
69. DIAZ,J . B., On an analogue of the Euler-Cauchy polygon method for the numerical
solution of uxy = f ( x , y , u, ux, u,,), Avch. Rational Mech. Anal. 1, 357-390(1958).
60. MAC DONALD, R. J., CARLSON, R. Id,, and LANKFORD, W. T.. The effects of strain
rate and temperature on the stress-strain relations of deep-drawing steel, Proc.
A m . Soc. Test. Materials 86, 704-120 (1956).
61. DoucrIs.A.,Someexistencetheoremsforhyperbolicsystemsof partialdifferential equa-
tions in two independent variables, Comm. Puve A p p l . Math. 5 , 119- 154 (1952).
62. DRUCKER, D. C., A definition of stable inelastic material, J . A p p l . Mech. 86, 101 - 106
(1969).
63. DRUCKER, D. C., On the role of experiment in the development of theory, in Proc.
Fourth U.S. National Congress Appl. Mech.. Berkeley, June 1962, Am. Soc. Mech.
Engrs., New York. 1963, pp. 18-33.
64. DRUCKER,D. C.. Stress-straiq-time relations and irreversible thermodynamics,
in “Proc. Inter, Symposium on Second-Order Effects in Elasticity, Plasticity
and Fluid Dynamics, Haifa, April 1962.” Pergamon Press, pp. 331-381.
65. DRUCKER. D. C., A continuum approach to the fracture of metals, i n “Fracture
of Solids”, (D. C. Drucker and J . J . Gilman, eds.), Wiley, New York 1983.
pp. 3-60.
66. DRUCKER, D. C., Some remarks on flow and fracture, Technical Report, Brown
University, May 1983.
67. DUWEZ,P. E., and CLARK,D. S., An experimental study of the propagation of
plastic deformation under conditions of longitudinal impact, Proc. A m . SOC.
Testing Malerials 47, 502-532 (1947).
68. ERINGEN.A. C.. “Nonlinear Theory of Continuous Media,” McGraw-Hill, New
York, 1982.
88. FREUDENTHAL, A. M.. and GEIRINGER, H., The mathematical theories of the inelastic
continuum, in “Encyclopedia of Physics.” vol. V1. Springer-Verlag, Berlin,
1958, pp. 229-443.
70. FRIEDRICHS, K. 0.. Nonlinear hyperbolic differential equations for functions of two
independent variables, American Journal of Mathematics, 70, 555- 589 (1948).
71. GOLDSMITH, W., ”Impact. The Theory and Physical Behaviour of Colliding Solids.”
Edward Arnold, London, 1960.
72. MAC GREGOR,C. W., and FISHER,J . C.. A velocity-modified temperature for the
plastic flow of metals, J . A p p l . Tech. 11. (1946).
73. GURTIN,M., and STERNBERG, E., On the linear theory of visco-elasticity, Arch.
Rational Mecfr. Anal. 11, 291-3356 (1962).
74. GIZATULINA, G. M., On the problem of deflection of a circular plate of visco-plastic
material, (in Russian) Zsledovaniya Upv. Plast. 3 (1964) Leningrad.
75. HARDING, J., WOOD, E. 0.. and CAMPBELL.J. D.. Tensile testing of materials
at impact rates of strain, J . Mech. Engrg. Sca. 2. 88-96 (1960).
76. HARTMAN, P., and WfrimER, A., On hyperbolic partial differential equations.
American Jouvnat o/ Mathematics, 74, 836-843 (1962).
372 PIOTR PERZYNA

77. HAUSER,F. E., SIMMONS, J . A., and DORN.J . E., Strain rate effects in plastic
wave propagation, Technical Report No. 3, University of California. June
1960.
78. HAWSER,F. E.. and WINTER. C. A., An experimental method for determining
stress-strain relations a t high strain-rates, Technical Report No. 4, University
of California, December 1960.
79. HILL, R., “The Mathematical Theory of Plasticity,” Oxford, 1950.
80. HOHENEMSER. K., and PRAGER,W., uber die Ansltze der Mechanik isotroper
Kontinua. Z A M M 19, 216-226 (1932).
81. HOPKINS.H. G., and WANG,A. J., Load-carrying capacities for circular plates of
perfectly plastic material with arbitrary yield condition, J. Mech. Phys. Solids
8, 117-129 (1954).
82. HOPKINS,H. G . . Dynamic expansion of spherical cavities in metals, Progvess i n
Solid Mechanics 1, 83- 164 (1960).
83. HOPKINS,H. G.. Dynamic anelastic deformations of metals, A#@. Mech. Rev. 14,
417-431 (1961).
84. HOPKINS,H. G., Non-linear stress-wave propagation in metals, i n “Progress in
Applied Mechanics - The Prager Anniversary Volume,” Macmillan, New York,
1963, pp. 55-71,
85. HOPKINS,H. G., Mechanical waves and strain-rate effects in metals, stress wave
in anelastic solids, Proc. I.U.T.A.M. Symposium, Brown University, April 1963,
(H.Kolsky and W. Prager, eds.). Springer-Verlag, Berlin, 1964. pp. 133-148.
86. JEFFREY,A.. The development of jump discontinuities in nonlinear hyperbolic
systems of equations in two independent variables, Arch. Rational Mech. Anal. 14,
27-37 (1963).
87. JOHNSON, J . E., WOOD,D. S., and CLARK,D. S., Delayed yielding in annealed low-
carbon steel under compression impact, Proc. A m . Soc. Tesling Materials 53,
765-764 (1953).
88. JOHNSON, J. E., WOOD, D. S.. and CLARK,D. S., Dynamic stress-strain relation for
annealed 2 s aluminium under compression impact, J . Appl. Mech. 90, 523- 529
(1953).
89. KALISKI,S., On certain equations of dynamics of an elastic-visco-plastic body.
The strain-hardening properties and the influence of strain rate. Bull. Acad.
Polon. Sci.; Sev. Sci. Techn. 11, 239 (1963).
90. KALISKI,S., NOWACKI, W. K., and WLODARCZYK, E., Propagation and reflection of
a spherical wave in an elastic-visco-plastic strain-hardening body, Pvoc. Vibr.
Probkms 6. 31-56 (1964).
91. KANTOROVICH. L. V., and AKILOV,G . P., “Functional analysis in normalized
spaces” (in Russian). Moscow 1969.
92. VON KARMAN, T., and DWWEZ. P.. The propagation of plastic deformation in solids.
in “6th International Congress for Applied Mechanics, Paris 1946’; also J . Appl.
Pbys. 21, 987-994 (1950).
93. KELLER,H. B., On the solution of semi-linear hyperbolic systems by unconditionally
stable difference methods, Comm. Pure Appl. Ma& 14. 447-471 (1961).
94. KELLER,H. B., and T H O M ~V., E , Unconditionally stable difference methods for
mixed problems for quasi-linear hyperbolic systems in two dimensions, Comm.
Pure A p p l . Math. 16, 63-73 (1002).
95. KOLSKY,H., and DOUCH, L. S., Experimental studies in plastic wave propagation,
J . Mcch. Phys. Solids 10, 195-223 (1962).
96. KRAFFT,J. M.,SULLIVAN, A. M.,and TIPPER,C. F.. The effect of static and dynamic
loading and temperature on the yield stress of iron and mild steel in compression,
Proc. Royal Soc. B H A , 114-127 (1964).
97. KRAFFT,J . M., and SULLIVAN, A. M., Effects of grain size and carbon content
 FUNDAMENTAL PROBLEMS IN VISCOPLASTICITY 373

on the yield delay-time of mild steel. Trans. Amer. Soc. Mefak 61, 643-659
(1969) and discussion t o this paper by J. D. Campbell, pp. 669-6665.
98. KROYM.A.. Zur Ausbreitung von StoDwellen in Kreislochscheiben. Z A M M 38,
104 und 297 (1948).
99. KRZYZAfiSKI, M., “Second Order Partial Differential Equations” (in Polish).
Warsaw, vol. I (1957). vol. I1 (1962).
100. KUKUDJANOV, V. N., and NIKITIN,L. V., The propagation of waves in bars of
non-homogeneous elastic/visco/plastic material (in Russian), Isu. Akad. Nauk
S S S R , Mskh. Mash. 1, 53-59 (1960).
101. KUKUDJANOV, V. N., and NIKITIN,L. V., Impact of a bar with piecewise constant
yield limit on a rigid target (in Russian), mag. Zhur. 1, 177-183 (1961).
102. LANDAU, H. G., WHINER,J. H., and ZWICKY,E.E., Thermal stress in viscoelastic-
plastic plate with temperature-dependent yield stress. J . A p p l . Mech. 27.
297-302 (1960).
103. LAX, P. D., Nonlinear hyperbolic equations, Comm. Pure Appl. Math. 8, 231-258
(1963).
104. LEE, E. H., and WOLF,H., Plastic-wave propagation effects in high-speed testing,
J . A p p l . Mech. 18, 379-386 (1951).
105. LEE, E. H., Wave propagation in anelastic materials, Deformation and Flow of
Solids, i n “Colloquium Madrid 1955”, Springer-Verlag. Berlin, 1956, pp. 139- 136.
106. LUBLINER.J .. A generalized theory of strain-rate dependent plastic wave propaga-
tion in bars, J. Mech. Phys. Solids 12, 69-65 (1964).
107. LUDWIK,P., “Elemente der technologischen Mechanik.” Springer-Verlag, Berlin.
1909, pp. 44-47.
108. MAJCHER, G., Sur un problhme mixte pour 1’6quation du type hyperbolique,
Annales Polonici Mathematici, 6 , 12 1 - 139 (1958).
109. MAIDEN,C. J., and CAMPBELL, J. D., The static and dynamic strength of a carbon
steel at low temperature, Philosophical Magazine 8 , 872- 885 (1958).
110. MALVERN, L. E., The propagation of longitudinal waves of plastic deformation
in a bar of material exhibiting a strain-rate effect, J . Appl. Mech. 18, 203-208
(1951).
111. MALVBRN, L. E., Plastic wave propagation in a bar of material exhibiting a strain
rate effect, Qwr#.A w l . Math. 8, 406-411 (1951).
112. MANJOINE,M.. Influence of rate of strain and temperature on yield stresses of
mild steel, J. A M l . Mech. 11, 211-218 (1944).
113. MARSH,K. J., and CAMPBELL, J . D., The effect of strain rate on the delayed
yielding flow of mild steel, J . Mech. Phys. Solids 11, 48-69 (1963).
114. MENTEL,T. J,, The plastic deformation due to impact of cantilever beam with
an attached tip mass, J. Appl. Mech. B6, p. 515 (1958).
115. MIKLOWITZ, J., The initiation and propagation of the plastic zone in a tension
bar of mild steel as influenced by speed and stretching and rigidity of testing
machine, J. A p p l . Mech. 14, A-31-A-38 (1947).
116. NADAI,A. L., “Theory of Flow and Fracture of Solids.” McGraw-Hill, New York.
vol. 1, 1950, vol. 2, 1963.
117. NAGHDI,P. M., Stress-strain relations in plasticity and thermoplasticity, ar
“Plasticity” (E. H. Lee and P. S. Symonds, eds.), Pergamon Press, London,
1960, pp. 121-167.
118. NAGHDI,P. M., and MURCH,S. A., On the mechanical behavior of viscoelasticl
plastic solids, J. A$@. Mech. 80, 321-328 (1963).
119. NIKITIN,L. V., The propagation of elastic-visco-plastic waves in a thick-walled
tube (in Russian), Zzu. Vurov. Mash. 8, 14-23 (1958).
120. NIKITIN, L. V., The propagation of transversal elastic-visco-plastic waves in
beams and plates (in Russian), Zag. Sbw. 80, 31-46 (1960).
374 PIOTR PERZYNA

121. NIKITIN,L. W . , Elastic-visco-plastic shear uaves in a cylindrical bar (in Russian),

Issled. Fiz. Radiot. 2, 115-122 (1958).
122. NOLL,W., and TRUESDELL, C.. The non-linear field theories of mechanics,
i n “Encyclopedia of Physics” (S. Flugge ed.), vol. 11113, Springer, Berlin
1965.
123. OLSZAK, W., and PERZYNA, P., Propagation of spherical waves in a non-homoge-
neous elastic-visco-plastic medium, in “Colloque International C.N.R.S..
Marseille 1961, pp. 67-78”; Bull.Acad. Polon. Sci., S M e Sci. Tech. 9, 509-516
(1961).
124. OLSZAK, W . , On critical states in viscoelasticity, an “Progress in Applied Mechanics
- The Prager Anniversary Volume,” Macmillan, New York 1963.
125. OLSZAK,W., and BYCHAWSKI, 2.. Criterion of fracture for visco-elastic bodies
(forthcoming).
126. OLSZAK,W., and PERZYNA, P., The constitutive equations of the flow theory for a
nonstationary yield condition, in Eleventh International Congress of Applied
Mechanics, Munchen, August 30 to September 5, 1964, (forthcoming).
127. OLSZAK,W., and PERZYNA, P., On elastic/visco-plastic soil, in Proceedings of
the Symposium on Rheology and Mechanics of Soils held at Grenoble April
1964, (forthcoming).
128. OLSZAK, W., and PERZYNA, P., The constitutive equations for elastic/visco-plastic
soils, (forthcoming).
129. PARKES, E. W., The permanent deformation of a cantilever struck transversely
at its tip. Proc. Roy. Soc. B E A , 462-472 (1955).
130. PERZYNA, P.. Propagation of shock waves in non-homogeneous elastic-visco-
plastic bodies, Arch. Mech. Stos.. 18, 8.51-867 (1961).
131. PERZYNA, P., Propagation of shock waves in an elastic-visco-plastic medium of
a definite non-homogeneity type, Avch. Mech. Stos., 14, 93- 111 (1962).
132. PERZYNA, P., The constitutive equations for rate sensitive plastic materials,
Quavt. A p p l . Math., 30, 321-332 (1963).
133. PERZYNA, P., The study of the dynamical behaviour of rate sensitive plastic
materials, Arch. Mech. Stos. 15, 113-130 (1963): Bull. Acad. Polon. Sci.,
SLY. sci. tech. 12, 207-216 (1964).
134. PERZYNA, P., The constitutive equations for work-hardening and rate sensitive
plastic materials, Proc. Vabr. Pvobl. 4. 2R1-290 (1963); BUN. Acad. Polon.
Sci.. Sh. sci. tech. 12, 199-206 (1964).
135. PERZYNA. P., On the propagation of stress waves in a rate sensitive plastic medium,
Z A M P 14, 241-261 (1963).
136. PERZYNA, P., and BEJDA,J., The propagation of stress waves in a rate sensitive
and work-hardening plastic medium, Arch. Mech. Stos. 16, 1215- 1244
(1964).
137. PERZYNA, P., On a non-linear boundary-value problem for a linear hyperbolic
differential equation, Bull. Acad. Polon., Skrie des sci. tech. le, 589- 894
(1964).
138. PERZYNA, P., and WIERZBICKI, T., Temperature dependent and strain rate sen-
sitive plastic materials, Arch. Mech. Stos. 16, 135-143 (1964); Bull. Acad.
Polon. Sci., Slr. sci. tech. 1 2 , 225-232 (1964).
139. PERZYNA, P., The application of the iteration method to the solution of the problems
of propagation of stress waves in an inelastic medium, Arch. Mcch. Stos. 17,
87-107 (1965).
140. F‘LASS, H. J.. Theory of plastic bending in a bar of strain rate material, in “Proc.
Second Midwest Conference Solid Mech., Lafayette, Indiana, Purdue Univ..
1955,” pp. 109- 134.
141. PLASS. H. J.. and RIPPERGER, E. A., Current research on plastic wave propagation
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 376

at the University of Texas, Parts I and 11, in “Plasticity” (E. M. Lee and
P. S. Symonds, eds.). Pergamon Press, London, 1960, pp. 453-487.
142. PRAGER. W., Mkanique des solides isotropes au deli du domaine blastique, Mlmo-
rial Sci., Math. 8 7 , Paris, (1937).
143. PRAGER,W.. Discontinuous fields of plastic stress and flow, Proc. of the Second
U.S. National Congress of Applied Mechanics. pp. 21-32 (1954).
144. PRAGER, W., Non-isothermal plastic deformation, Proc. Kon. Ned. Akad. Wet. B81.
176- 182 (1958).
145. PRAGER,W., Linearization in visco-plasticity, &err. Ing. Avchiu, 16, 152- 167
(1961).
146. PRAGER,W., “Introduction to Mechanics of Continua.” Ginn and Company.
Boston, 1961.
147. PRANDTL. L., Ein Cedankenmodell zur kinetischen Theorie der festen Korper,
Z A M M N, 85-106 (1928).
148. ~ O U S EG., , Sulla risoluzione del problema misto per le equazioni iperboliche non
lineari mediante le differenze finite, A n n . di Mat. 46, 313-341 (1958).
149. PUGH.H. L. D., CUANG,S. S.. and HOPKINS,B. E., Tensile properties of a high-
purity iron from - 1H6 “C to 200 “C a t two rates of strain, Phil. Mag. %, 753-768
(1963).
150. K A J N A K ,S., HAUSER, F. E..and DORN,J . E., Theoretical prediction of strain
distribution under impact loading, Technical Report No. 1, University of
California, June 1961.
151. RAKHMATULIN, H. A , , and SHAPIRO, G . S.. Propagation of a disturbance in a non-
linear elastic and non-elastic medium (in Russian), I z v . Akad. Nauk S S S H . .
Otdelenie tech. nauk 8 , 68-88 (1955).
158. REINER,M., Plastic yielding in anelasticity. ./. Mech. Phys. Solids 8. 265-2261
(1960).
153. RUBIN,R. J., Propagation of longitudinal deformation waves in a pre-stressed
rod of material exhibiting a strain-rate effect, ./. A p p l . P h y s . 86.. 528-5636
(1954).
1.54. SCHAUDBR, J ., Cauchysches Problem fiir partiell: Differentialgleichungen erster
Ordnung. Anwendung einiger sich auf die Absolutbetrlge der Lesungen be-
ziehenden Abschatzungen, Comentaviz Math. Helu. H. 263-283 (1937).
.,
155. S C H A U ~ E JR , Das Anfangswertproblem einer quasi-linearen hyperbolischen
Differentialgleichung zweiter Ordnung in beliebiger Anzahl von unabhlngigen
Veranderlichen, Fundamenta Mathematicae Y4, 213- 246 (1935).
156. SIMMONS, J. A,, HAUSER, F. E.,and DORN,J . E., Mathematical theories of plastic
deformation under impulsive loading. University of California, Publications
an Engineering 5, 177-230, (1962).
157. SKBMFTON, A. W., and BISHOP,A. W., “Building materials: their elasticity and
inelasticity” (M. Reiner, ed.). North-Holland Publ., Amsterdam, 1964,
Chapter X, Soils.
158. SKRIPXIN, V. A , , On the model of dynamical work-hardening continuous media
in the Mises-Reuss theory of plasticity (in Russian). Prak. Mat. Mekh. 87,
109-115 (1983).
159. SOKOLOV, D. D., On the problem of determination of the resistance of metals to
the plastic deformation in dependence on the strain rate and temperature,
(in Russian) Doklady Akademii Nauk S S S R 6I (1949).
160. SOKOLOV, D. D., “Strength of Metals in Plastic Deformations.” (in Russian),
Moscow. 1963.
161. SOKOLOVSKY, V. V., Propagation of elastic-viscoplastic waves in bars (in Russian),
Prikl. Mat. Mekh. 18, 261-280 (1948).
376 PIOTR PERZYNA

182. SOKOLOVSKY, V. V., Propagation of elastic-viscoplastic waves in bars (in Russian),

Doklady Akad. Nauk SSSR 6U, 775-778 (1948).
183. SOKOLOVSKY, V. V., Propagation of cylindrical shear waves in elastic-viscoplastic
bodies (in Russian), Doklady Akad. Nauk S S S R 60, 1325-1328 (1948).
184. SOKOLOVSKY, V. V., Unidimensional non-steady motion of a viscoplastic medium,
Prikl. Mat. Mekh. 18, 823-6832 (1949).
165. STERNGLASS, E. J., and STUART, D. A,, .4n experimental study of the propagation
of transient longitudinal deformations in elastoplastic media, J . A p p l . Mech. 20.
427-434 (1953).
188. SZMYDT, Z., Sur un nouveau type de problkmes pour un systkme d’bquations
diffdrentielles hyperboliques du second ordre B deux variables independantes.
Bull. Acad. Polon., GI. 111 4, 67-72 (1958).
187. SZMYDT, Z., Sur une gkn6ralisation des problemes classiques concernant un systkme
d’dquations diffdrentielles hyperboliques du second ordre B deux variables
indhpendantes, Bull. Acad. Polon., Cl. I I I 4, 579-584 (1956).
188. SZMYDT, Z., Sur le problkme de Goursat concernant les dquations diffdrentielles
hyperboliques du second ordre, Bull. ,4cad. Polon., C1. I I I h, 571-b75
(1957).
189. SZMYDT, Z., Sur un problkme concernant un s y s t h e d’kquations diffdrentielles
hyperboliques d’ordre arbitraire B deux variables inddpendantes, Bull. Acad.
Polon., GI. I I I 6 , 577-582 (1957).
170. SZMYDT, Z.. Sur l’kxistence de solutions de certains problemes aux limites relatifs
B un systbme d’bquations diffkrentielles hyperboliques, Bull. Acad. Polon.,
St‘rie des sci. math., astr. et p h y s . 6, 31-36 (1958).
171. TAYLOR, G. I., The testing of materials at high rates of loading, J . Institution of
Civil Engineers 8 , 480-6619 (1945-46).
172. TAYLOR, G. I., The use of flat-ended projectiles for determining dynamic yield
stress, I : Theoretical considerations, Proc. Royal Soc. 104A, 289-299 (1948).
173. TAYLOR, D. B. C., The dynamic straining of metals having definite yield points,
J. Mech. Phys. Solids 8, 38-45 (1964).
174. T H O M ~ V.,
E , Difference methods for two-dimensional mixed problems for hyper-
bolic first order systems, Arch. Rational Mech. Anal. 8. 68-88 (1981).
175. T H O M ~ V.,
E , A mixed boundary-value problem for hyperbolic first-order systems
with derivatives in the boundary conditions, Avch. Rational Mech. Anal. t4.
435-443 (1981).
E , A difference method for a mixed boundary problem for symmetric
176. T H O M ~ V..
hyperbolic systems, Arch. Rational Mech. Anal. 2, 18, 122- 138 (1983).
177. TIETZ,T. E., and DORN, J. E., The effect of strain histories on the work ha-dening
of metals, Trans. Amer. SOE.lor Metals 41A, 183-179 (1949).
178. TING,T. C. T., and SYMONDS, P. S., Impact of a cantilever beam with strain
rate sensitivity, i n “Proc.Fourth U.S. National Congress Appl. Mech.. Berkeley
June 1982,” pp. 1153-1185.
179. TING,T. C. T., On the solution of a non-linear parabolic equation with a floating
boundary arising in a problem of plastic impact of a beam, Quart. Appl. Math.
21. 133-150 (1983).
180. TING,T. C. T., The maximum-minimum principles for a quasi-linear parabolic
finite difference equation, Quart. A p p l . Math. 92, 47-55 (1984).
181. TING.T. C. T., and SYMONDS, P. S., Longitudinal impact on visco-plastic rods
- linear stress-strain rate law, J. A p p l . Mech. 81, 199-207 (1964).
182. TOUPIN,R.. and BERNSTEIN,B., Sound waves in deformed perfectly elastic
materials. Acoustoelastic effect, J . Acoust. Soc. Amev. 88, 218-225 (1981).
183. TRUESDELL, C., and TOUPIN,R. A., The Classical Field Theories, i n “Handbuch
der Physik,” I I I / l , Springer-Verlag, Berlin, (1980), pp. 228- 793.
 FUNDAMENTAL PROBLEMS I N VISCOPLASTICITY 377

184. TRUESDELL, c., General and exact theory of waves in finite elastic strain, Arch.
Rational Mech. Anal. 8, 263-296 (1961).
185. TRUESDELL, C., and TOUPIN,R.. Static grounds for inequalities in finite strain
of elastic materials, Arch. Rational Mech. A n d . 18, 1-33 (1963).
186. VISHMAN, F. F., ZLATIN,N. A., and MODDE. V. S., Resistance to the deformation
of metals a t the rate of lo4 to 10-’ mm./sec., Zhurn. Tekhn. Fiz. 19 (1949).
187. VOLOSHENKO-KLITOVITSKY, Y. Y . , On the conformity of the law of yield limit
at low-temperatures loads, Izv. Akad. Nauk S S S R , Mekh. Mash. 1, 154-156
(1962).
188. WEHRLI,C.. und ZIEGLER,H., Einige mit dem Prinzip von der groBten Dissi-
-
pationsleistung vertrlgliche Stoffgleichungen, Z A M P 18, 372 393 (1962).
189. WHIFFIN,A. C.. The use of flat-ended projectiles for determining dynamic yield
stress, 11. Tests on various metallic materials, Proc. Royal Soc. 1R4A, 300-332
(1948).
190. WHITE,M. P., On the impact behavior of a material with a yield point, J . A p p l .
Mech. 16. 39-52 (1949).
191. WIERZBICKI, T., A thick-walled elasto-visco-plastic spherical container under
stress and displacement boundary conditions, Arch. Mech. Stos. 15. 297-308
(1963).
192. WIERZBICKI. T.. Impulsive loading of a spherical container with rigid-plastic and
strain-rate sensitive material, Avch. Mech. Slos. 15, 775- 790 (1963).
193. WIERZBICKI. T., Bending of rigid/visco-plastic circular plates, Arch. Mech. - 9 0 s .
16 (1964); Bull. Acad. Pol. Sci.. Se’r. sci. tech. 18, 611-618 (1964).
194. WIERZBICKI, T., On the impulsive loading of a spherical vessel, Bull. Acad. Pol.
Sci., Sbr. sci tech. 12, 217-224 (1964).
196. WIERZBICKI, T.. Dvnamics of visco/plastic circular plate, ‘4rch. Mech. Stos. 17,
851-869 (1965).
196. WINTNBR, A., On the conditions of validity of Riemann’s method of integration,
Quart. ApPl. Maths. 16, 94-98 (1957).
197. ZIEGLER,H., Uber ein Prinzip der groBten spezifischen Entropieproduktion und
seine Bedeutung fur die Rheologie. Rheologica Acta 2, 230-235 (1962).
198. ZIEGLER,H., Some extremum principles in irreversible thermodynamics with
application to continuum mechanics, Progress in Solid Mechanics 4. 91 - 193
(1963).
199. ZIEGLER,H.. Thermodynamic considerations in continuum mechanics. The 1964
Minta Martin Lecture, Massachusetts Institute of Technology.
200. ZVEREV,I. N., The propagation of perturbation in visco-elastic and visco-plastic
bars (in Russian), Prikl. Mat. Mekh. 14, 295-302 (1960).
201. DRUCKER, D. C.. and PRAGER,W., Soil mechanics and plastic analysis of limit
design, Quart. A p p L Math. 10, 157-165 (1952).
202. LEE, E. H., and TUPPER,S. J.. Plastic deformation in a steel cylinder striking
a rigid target, J. A p p l . Mech., el. 63-72 (1954).
 Author Index
Numbers in parenthesesare reference numbers and are included to assist in Loeating referencesin which auto&
name8 are not mentioned in the test. Numbers in italics refer to pages on which the complete referenceare listed.

A Campbell, W. R., 369

Akilov, G. P., 305, 372 Carlson, R. L., 252. 377
Alder, J . F., 252, 296, 296(1), 368 Carrier, G. F., 29. 42, 86
Alexeiev, N. A., 253, 368 Case, K. M., 29, 30, 31, 86
Alter, B. E. K.. 257, 368 Chadwick. P.. 253, 370
Alterman, 2.. 87 Chang, C. T.. 29, 86
Appleby, E. J., 352, 358, 368 Chang, S. S., 290(149), 296(149), 375
Argemi, J., 180, 241 Chiarulli, P., 42(47), 87
Atkinson, C. P., 191(29), 217(29), 241 Chiddister, J. L., 262, 290(34), 295(34).
Aziz, A. K., 313, 315, 316(5). 368 296(34), 370
Chu, S. C.. 308. 313, 314, 315, 370
B Cinquini-Cibrario, M., 303, 370
Barenblatt, G. I., 368, 368 Clark, D. S.. 245, 246, 248, 249, 261.
Batchelor, G. K., 82(90), 84, 85(90). 89 284, 288. 288. 370, 371. 372
Bejda. J., 316(136), 336(136), 340(136), Coddington, E. A.. 220(26), 241
361(8), 362(9), 368(7), 368, 369, 374 Coleman, B. D.. 260, 272, 370
Bell, J. F., 257, 258, 369 Conlan. J., 307. 370
Bergeron. T., 43, 47, 50(49), 51(48, 49), Copson, E. T., 308, 370
87 Cottrell, A. H.. 258. 370
Bernstein, B., 272. 376 Courant, R., 149(13), 754. 173(17), 241,
Baby, B. A., 258, 370 302, 303, 308, 340. 370
Bishop, A. W.. 253, 264, 375 Cowper, G. R.. 283, 366, 371
Bjerknes. J., 43, 47. 50(49). 51(48, a ) , Cox, A. D., 253, 370
87 Cristescu. N., 269, 371
Bjernes, V.. 43. 47, 51(48). 87 Curle. N., 40(42), 42(44), 87
Bland, D. R.. 260, 369 Curreri, J. R., 191(32), 193(32), 242
Bodner, S. R., 283, 286. 366, 367, 369 Curtis. C. W.. 267, 368
Bogoryad, I. B., 118(9), 753
Borisova, E. P., lOB(6). 153 D
Bragg. L. E.. 272, 369 Darboux, ti.? 167, 241
Budiansky. B., 92(3), 153 Davis. P. A., 68. 88
Bundgaard, R. C., 43, 50(49), 51(49), 87 Den Hartog, J. P., 204(37). 242
Bychawski. 2.. 264, 374 Deutler, H., 264, 371
Bykovcev. C. M., 361(22), 369 Diaz, J . B.. 307. 308, 313, 314, 316.
316(5), 368, 370, 377
C Dikii, L. A., 29, 30, 31. 86
Campbell, J. D.. 245. 246, 249, 250, 251. Dokuchaev, I,. V., 111(7). 117, 118(7), 153
262, 253, 256, 268, 259, 287, 289, Dorn. J . E., 246, 247, 248, 249, 266,
200(109). 296(108). 341. 36g. 370, 256, 257, 258. 269. 290, 368(160).
371, 373 369, 372, 375, 376

379
 380 AUTHOR INDEX

Doroshkin, N. Ya., 117(8), 118(8), 153 Hatanaka, H., 48, 87

Douch, L. S., 249. 250, 257. 268, 372 Haughton, K. E., 189, 241
Douglis. A., 303, 371 Haurwitz, B., 40, 43, 44, 47. 87
Drazin, P. C.. 21, 22(8), %(a), 39(8), Hauser, F. E., 246, 247, 255, 256, 257,
40(8). 42(8), 51(67), 52, 53(72), 268, 290, 368(150), 372, 375
67(67), 58, 65. 66. 69(83). 74. 76, Heisenberg. W.. 9(6), 85
76(83), 77, 85, 88 Helmholtz, H., 1, 3, 85
Drucker, D. C., 264, 2H2, 300, 301, Hilbert. D., 149(13), 302, 303, 308,
371, 372 340, 154, 173(17), 241. 370
Duby, J., 245. 249, 387, 369 Hill. K.. 273, 341(79). 372
Duwez, P., 260, 372 Hocking, L. M., 79(87, 8 8 ) , 88
Duwez, P. E., 246, 247, 248, 284. 286, Hohenemser, K., 272, 280, 372
370. 371 Holmboe, J., 60, 72, 88
Heiland, E., 18, 34(3h), 69, 86, 88
E Hollingdale. S., 40, 42, 87
Eckart. C.. 49. 79(64), 88 Hopkins, B. E., 290(149), 296(199),
Eliassen, A., 34(35), 69, 86 316, 318, 340, 375
Eringen, A. C., 261, 371 Hopkins, H. G., 246. 253, 354. 256,
Each, It. E., 78, 87, 88 258, 340(83), 352(81), 370, 372
Howard, L. N., 14(16), 17, 18, l9(16),
F 21, 22(8), 38(8), 39(8), 40(8), 42(8),
Faddeev, L). K., 148(12), 149(12), 154 51(67), 52. 57(67). 58, 61, 62(23),
Faddeeva, V. N., 148(12). 149(12), 154 63(25),64(25), 65(25), 66, 60(83), 71,
Fejer, J , A., 48(A8), 87 73, 74, 75, 76(83),85, 86, 88
Fjeldstacl, J . E., 88, 88 Hsu. C. S., 164(9). 171(9). 172(9), 174(9),
Fjertoft, R., 16, 85 188(9), 190(9), 221(9, 40), 224(9),
Foote, J . K., 14. 86 226(9),232(40), 241, 242
Pox, J. A,. 40, 43, 86
Freeman, J . C., 42(47),87
I
Freudenthal, A. M.. 283, 3W(69), 371
Frietlrichs, K. 0.. 12. 86. 303. 370. 371 Isaacuon, E., 303, 370
Ishlinsky, A. Y.. 368. 368
u
Garcia, H. V.. 42(43), 87 J
Geiringer. H., 283, 300(69), 371
Jeffrey, A.. 302, 372
Gill, A. E., 17, 82(90),84, 88(90). 86, 89
Johnson, J. A., 49, 51(65), 88
Gizatulina, G. M., 381(74), 371
Johnson, J. E.. 246. 249, 261. 288. 372
Godske, C. I,.. 43, 50(49),87
Goldsmith. W.. 371
Goldstein. S., 44. 70. 71, 87 H
Gurtin. M.,260, 261(73), 371
Kaliski. S.. 276. 316(90). 372
Kantorovich, L. V., 305, 372
H Kauderer, H.,165, 187, lUl(lO), 193(10),
Haag, J . , 167(15), 241 241
Hading, J., 249. 250, 255. 289, 341, Keller. H. B.. 303. 372
369, 371 Kelvin, W., 1, 3, 16(21), 36(2), 39(3),
Hartman, P., 303, 371 43(2), 44, 48(2). 82(2), 85, 86
 AUTHOR INDEX 381

Kinney. W. D., 166, 205(12). 207(12). Maiden, C. J.. 251. 252, 253. 290(109).
210(12, 13), 214(12), 216, 231, 232, 296(109), 369, 373
24 1 Malvern, L. E., 252, 254, 255, 273,
Klotter, K., 217(38), 242 277, 283, 290(34), 295(34), 296(39),
Kolsky, H., 249, 250, 2.57, 258, 372 370, 373
Krafft, J . M., 282, 258(97). 290, 296(96), Manjoine. M., 246, 247. 252. 296(112).
372 367, 373
Kromm, A., 340, 373 Marsh, K. J.. 246, 247. 249, 251(32),
Krzyzabski, M., 308, (99). 373 370, 373
Kuchemann, D., 49, 87 Menkes, J., 47, 87
Kukudjanov, V. N.. 368(100), 373 Mentel, T. J.. 365, 367, 373
Kuo, H. L.. 51(66), 52, 88 Michael, 1). H.. 52(60. 71). 53(71). 88
Kuo, J . K.. 163(8), 199(8, 18), 200. Michalke, A., 42, 87
201(8, 18), 202. 230(8, 43, 44). 233. Miishkes, A. D., 191(31), 193(31), 242
234, 241 Mikishev, G. N., 117(8), llE(8). 153
Miklowitz, J., 246, 373
I, Miles, J . W., 29, 48(68), 48, 62. 63
(77, 78), 71, 78, 86, 87, 88
Landau, H. G., 373
Minorski. N., 1.56. 179(1), 240
Landau, L., 48. 87, 191(33), 193(33).
Modde, V. S., 205(186), 377
242
Moiseev. N. N.. 91(1), 92(1), 93(l),
Lankford. W. T.. 252, 371
94(1), 95(1), 127(11), 753
Lax, P. I)., 303, 370, 373
Murch, S. A., 269, 261. 262, 267, 270,
Ledoux, P.. 158(.5), 240
373
Lee, E. ti., 245, 367, 368. 373, 377
Murphy, J. W.. 86
Lessen, M., 40, 42, 06
Levinson, N.. 220(26), 247
Liepmann, H. W.. 49, 57(74), 87. 88 N
Lifshitz, E. M., IYI(33). 193(33). 242
Nadai, A . L., 254, 373
Lighthill, M. J.. 15, 86
Naghdi. P. M.. 259. 261, 262. 267. 270.
Lin. C. C . , 6, 9(7, 9), 12, 14, 16, 20(7),
273, 281(117), 373
29. 36(7), 49, 58, 70(5), 71(5), 85. 86.
Nikitin, L. V., 318(110), 361, 368(100).
87
373, 374
Lipps, F. B.. 52, 88
Noll, W., 260, 261, 272, 370, 374
Lock, 1s. C., 53(73), 88
Nowacki, W. K., 316(90), 372
Loude, W.. 221. 222(41). 242
Lubliner. J., 250, 369(106), 373
Ludwik, P.. 264, 373 0
Olszak, W.. 264, 282(127, 128). 290(126),
M 300(128), 316(127), 374
MacDonald, K. J., 262, 377 Orr. W. M. F.. 28. 86
McDuff, J . N., 191(32). 193(32), 242
MacGregor. C. W., 371
MacLachlan, N. W., 223(42). 242
P
Maharem, N., 184(25), 241 Panofsky, H. A., 40, 87
Majcher, G., 313, 315, 337, 373 Parkes, E. W.. 365. 367. 374
Marwin, J., 158(6), 171, 176(6), 188, Patterson, A. M., 68, 88
189, 190, 240 Pearson, K., 193, 242
382 AUTHOR INDEX

Perzyna, P., 273, 278(132), 281(127, Rosenbluth. M. N., 14(17). 86

128), 283(132, 133). 288(133), 290 Roshko, A.. 49, 87
(126). 291(138), 297(132, 133, 134). Rubin, K. J., 368(153), 375
298(132),300(132),308(137), 316(127,
130, 131, 135, 136). 330(130, 131, S
135. 139). 331(130, 131, 135). 336
Sagamonian, A. Y., 253, 368
(130, 131, 135, 136) 340(136), 374
Sato, H., 40, 87
Petrov, A. A., 97(4), 100(5), 123(10),
Savic, P., 40(36. 37). 86
132(4). 138(4), 143(4), 148(4), 753
Schauder, J., 303, 375
Phillips, V. A., 252, 295, 296(1), 368
Semykina, P. D., 361(22), 369
Plass. H. J., 361, 374
Sideriadbs, L., 180, 247
popov, Yu. P., 97(4), 132(4), 138(4),
Simon, A,, 14(17),86
143(4), 148(4), 153
Simmons, J. A.. 246. 247, 248, 258,
Rager, W., 272, 280, 281(146), 282,
259. 290, 369, 372, 375
296, 297, 352, 358, 359. 368, 372,
Skempton, A . W., 253, 254, 375
375, 377
Skripkin, V. A.. 375
Prandtl, L., 254, 375
Sokolov, D. D., 252, 295. 375
Pretsch, J., 82(92). 89
Sokolovsky, V. V., 254, 255, 273, 330.
Prouse, G., 303, 375
h c k e t t . A. E.. 57(74), 88
335. 368. 375, 376
Solberg, H., 43, 47, 51(48), 87
Pugh, H. L. D., 290(149), 296(149),
Squire, H. B., 6, 68(4), 85
375
Sternberg, E., 260, 261(73), 377
Pukhnacher, Yu. V., 97(4), 132(4),
Sternglass, E. J., 257, 258(165). 376
138(4), 143(4), 148(4). 753
Stoker, J . J., 157(4), 219(4), 223(4),
It 224(4), 226(4). 240
Struble, R. A., 218(39), 231(39), 242
Rajnak, S., 255, 256, $57, 368(150),
Stuart, D. A., 257, 258(165), 376
375
Stuart, J . T., 10. 52(70), 53(70), 85, 88
Rakhmatulin, Kh. A., 253, 368, 375
Sullivan, A. M., 252, 258(97), 290,
Rauscher. M.. 157, 185, 240
296(96), 372
Rayleigh, J. W. S., 1, 3, 6, 10, 35(3),
Symonds, P. S., 283, 285, 366, 367, 368,
36(5). 37, 38(3), 44, 55(51). 70(3),
369, 371, 376
82, 84, 85, 87, 89
Synge, J. L., 61, 79(89), 88, 157, 167(2).
Rees, M., 303, 370
240
Reid, W. H., 10, 85
Szmydt. 2.. 3O3, 304, 306, 308. 376
Reiner, M., 264, 375
R h , E.. 34(35), 69, 86
Ripperger. E. A,, 361, 374 T
Rosenberg. R. M., 162(7), 163(8). 164(9), Taylor, D. B. C., 251. 376
166(ll), 166(13), l68(16), 171(9, l9), Taylor, C . I., 16, 44, 45, 68, 70, 86,
172(9). 174(11), 176(11), 176(19, 24), 87, 246, 247, 367, 376
179(19), 180(21). 182(24), 185(19), ThomBe, V.. 303, 372, 376
188(9), 190(9,28),191(20,29), 192(20), Tietz, T. E., 249, 376
193(20),195(20). 199(8),201(8),202(8). Timoshenko, S., 204(36), 242
203(35),205f35),208(35),210(13, 35), Ting, T. C . T., 368. 367. 368, 376
211(35). 213(36), 217(29), 220(21), Tipper, C. F.. 252, 290, 296(96), 372
221(7, a), 224(9), 225(9, 28). 230(8), Tollmien, W.,11, 35, 85
233(19), 240, 247, 242 Toupin, R. A., 260, 272, 376, 377
 AUTHOR INDEX 383

Troech. B. A.. 92(2), 153 366(193),357(193),368(193),359(193),

Truesdell, C., 260, 261, 272, 374, 376 360(193), 361(193, 195). 362(192,
Tupper. S. J.. 367, 377 104), 364(192, 194). 365(194), 374,
377
V Winter. C. A., 372
Vishman, F. F., 295(186), 296. 377 Wintner, A.. 303, 308. 371, 377
Voloshenko-Klitovitsky,Y . Y., 292(187). Wlodarczyk, E., 316(90), 372
295(187), 296. 377 Wolf, H.. 245, 368, 373
Von Kbrmbn, T., 260, 372 Wood, D. S., 246, 249, 251, 288, 370,
372
Wood, E. 0.. 240, 260, 280. 341. 371
W
Wang, A. I., 352(81), 372
Watson, G . N., 191(30). 262 Y
Wehrli, C., 291, 377 Yih, C. S.. 46(64), 87
Weiner. J. H.. 373
Whiffin, A. C., 246. 367, 377
White, M. P., 246, 367, 308(190), 377
z
Whittaker, E. T..191(30), 242 Ziegler. H., 291, 377
Wierzbicki. T., 291(138). 350(191, lop), Zlatin. N. A., 295(186), 296, 377
361(191. 194). 352(193), 355(193), Zverev. I. N., 368(200). 377
 Subject Index

A D
A-process def., 298 Delay-time, 246, 261
Admissible systems (NV), 158f.. 162 Difference equation, 321
Alfv6n waves, 63 Dislocation theory of crystalline material,
Autonomous system, admissible (NV). 258
160 Dissipation energy, 264
trajectories of, 170 Drucker’s postulate, 265
Amp-function, 196f. Duffing approximation, 228
Ateb-function, 191, I96 Dynamic test, 245
properties, influenced by temperature
and irradiation, 246, 262
B
B-process def., 299 E
Banach’s theorem, 305
Beta function, 193 E-curve def. (NV). 178
incomplete, 193 Elastic energy, 264
Bounding surface def. (NV), 169 Elastic-viscoplastic def., 269
Elastic/viscoplastic def., 259
Elastic/visco-(perfectly plastic) material,
C 277, 281
Elastic/viscoplastic region, boundary of,
Cam-function, 197, ff. 322
Cantilever-bepm specimens, 306f. e-tube (NV),217
Characteristics, 317, 321 Escalator method, 140, 147f.
Circular cylindrical container with apher- Exchange of stabilities, 19, 40, 65, 73
ical caps (FOL). 138
Coaxial cylinders, horizontal (FOL), 123
Configuration space (NV), 164 F
Conical container (FOL), 111 F-curve def. (NV), 184
Constitutive equations, 200ff. Finite differences, method of, 310
for elastic-viscoplastic material, 270 Flow surface, 202
for rate-sensitive plastic material, 272 convexity of, 266
Constitutive inequalities, 272 Fourier method (FOL), 97
Convexity of flow surface, 209 Fracture of time- and temperature-
Coordinate functions, choice of, 96 dependent materials, 300
Creep process, 264 Frank-Read dislocation source, 259
Critical layer (HS), 16f., 24, 03. 66 Free oscillations def. (FOL), 94
Critical state, 264 Frequency equation, generalized, (NV),
Cubical dilatation, plastic rate of, 282 21 1
Cylindrical container with horizontal axis approximate method of calculation,
(FOL). 132 102

384
 SUBJECT INDEX 386

0 Modal line, 173, 175

G-curve def. (NV), 184 Motion with rest point, 171, 176
“Geometrical method” (NV), 167
N
H
Natural boundary conditions, 97
Heisenberg series (HS), 20 Natural free vibrations in normal modes,
Hill equation, 223 162
Neumann operator, 94
I Neutral curve of stability (HS), 54
Impact process, 245 Neutral process (PV). 263
Inclined circular cylindrical container Non-isothermal processes, 272
(FOL), 106 Non-linear boundary-value problem, 303
Inertial instability of plane parallel flow, 3 for linear hyperbolic equation, 308
of axisymmetric jets, 80 Nonlinear instability (FOL), 7, 25
Inertial waves, 50 Nonlinearizable (NV), 161, 180
Initial-value problem of stability, 2, 6f., Nonsimilar normal-mode, 199, 216
22. 25, 28, 67, 69 Normal-mode of nonlinear system, 173,
Instability of jets, 15, 22. 36ff.. 56ff., 66f.. 176
69f.. 74, 76
of half-jets, 21 0
of shear layers, 21f.. 35, 37, 39. 42.
47, 49, 51, 56ff.. 65, 70, 73, 76 Orr-Sommerfeld equation, 7, 9
Internal gravity waves, 44ff.. 55, 62, 65ff. Out-of-phase mode, 225
Inviscid plasticity, 270
P
I(
Parallelepipedical container (FOL). 100
Kbrmh-Taylor-Rakhmatulin theory, Permanent strain distribution, 246, 250
268f. Perturbation of mass parameters, 202
Kelvin’s cat’s-eye diagram, 16 of spring parameters, 202
Kelvin-Helmholtz instability, 1. 39, 45, Physical relation for one-dimensional
56, 66 problem, 254f.
Picard problem. linear generalized. 316
L generalized, 337
L-surface (NV), 217 Plane waves in half space, 328
Liapunov stability. 218 Planetary waves, 49f.
Linearizable (NV), 161 Plastic corner, 270
Linearization of constitutive equation, Plastic strain-rate, vector, direction of,
297 269
Loading process, 263 Plastic waves in rods, longitudinal. 361
Loading surface, 271 Poincar6 stability, 218
instantaneous, 268 Polygon method, 307
Longitudinal impact on viscoplastic rods, Principle of least action, restricted, 166
367 “Pseudo-system’’ (NV), 164, 205

M
Magnetohydrodynamic waves, 52f. (1
Mathieu equation, 223 Quasi-static solutions. 36Off.
386 SUBJECT INDEX

R in r o t d n g system, 49
Rapid load path, 267 with variable Coriolis parameter, 61.
Rate-sensitive materials, 273 57
Rayleigh’s necessary condition for in- Stability of normal-mode vibration (NV),
stability, 2, lOf., 15, 26f., 36, 61, 66, 82 230
Rayleigh stability equation, 7, 9f., 12, 14, Stable elastic/viscoplastic material, dy-
17, 20, 31, 42. 44, 47, 50 namical condition for, 282
Rayleigh-Taylor instability, 46 Stable inelastic material, 264f.
Relaxation equation, 300 State depending on loading path, 244
Relaxation process for general states of on load history and on time, 246
stress, 298 Static yield function, 273
Reynolds stress, 14f., 42, 64, 83 Static properties, changed by previous
Riemann function def., 308 dynamic loading, 246, 251
right circular-cylindricalcontainer (FOL), Steady-state forced vibrations, 162, 204
101 Strain acceleration, influence of, 268
with annular base (FOL), 101 Strain-hardening effect, reduction of, 249
Ritz method, 93, 96f. Strain-rate sensitivity, 246, 266
Rossby waves, Slf., 67 of beams under impact, 366f.
of metals, 276
of plastic materials, 290, 291
S of yield limit. 246
Sam -f unction, 197ff. Strain rate and temperature, simultane-
Schauder’a theorem, 306 ous influence of, 291
Semicircle theorem, 17, 49, 63, 68, 79, 83 Stress space, 262
Semi-linear hyperbolic partial differential Stress-strain curve, dynamic, 248
equation, 303 Stress-strain relation. static 277
Shear waves, cylindrical, 316, 326ff. dynamic, 277
a,&-curves, dynamic, 248 Stress waves, numerical examples. 340
Simple trajectories, existence of, (NL), in elastic/visco-plastic soil, 316
233ff. propagation, 302
Soils, experimental results for, 263 solution in the elastic region, 336ff.
dilatation rate. of, 281 Strong discontinuity, 318
Sokolovsky-Malvern theory. 269 Strongly nonlinear (NV).161
Sound waves, 43. 47, 49 Strubel. method of, 231
Spherical container (FOL),118 Strutt chart, 224, 227
Spherical container, impulsive loading of
(PV),362 T
Spherical problem, quasistatic sol., 350 T-curves def. (NV).186f.
Squire’s theorem, 2, 6, 43, 46, 48. Mm.,82 Temperature-dependent materials, 29Of.
Stability boundary, 64, 63f., 69f.. 73. 76ff. Thermostatic and thermodynamic in-
Stability in the large (PV),266 equalities, 272
in the small, 265 Thick-walled spherical container, quasi-
Stability of compressible fluid, 47, 67 static sol., 360
of electrically-conducting fluid in a Time-dependence of stress and strain, 244
magnetic field, 62 Toroidally shaped container (FOL),143
of fluid of variable density, 44, 60, Transversal motion, influence of, 250
64, 68. 60 Transversals (= P-curves) def. (NV).
of non-parallel flow, 22 184f.
 SUBJECT INDEX 387

V Viscosity constant, 214

Viscous strain, separation of, 261
Variational method for liquid oscillations,
93 W
Varicose disturbances, 19, 22, 36ff.. .66, Wave problems for rods and beams, 361ff.
14f., 82 Waves, cylindrical radial, 316, 322
Vibration absorber, tuned, 213 spherical, 316
Vibrations in unison, 162 plane, 316
Viscoelastic material, criterion for, 246, Weakly nonlinear (NV), 181
262 Work-hardening, isotropic, 276
Viscoelasticity, linear, 260 Y
Viscoplastic flow of a circular plate, quasi- Yield condition, dynamical, 276
static sol.. 352 Yield criterion, 259, 262, 264
Viscoplastic theory, 268 Yield limit, 249
Viscoplasticity, 244 upper, 249'-