UNCLASSIFIED

AD NUMBER
AD143756

NEW LIMITATION CHANGE

TO
Approved for public release, distribution
unlimited

FROM
Distribution authorized to U.S. Gov't.
agencies and their contractors;
Administrative/Operational Use; SEP 1957.
Other requests shall be referred to Office
of Naval Research, Washington, DC.

AUTHORITY

onr ltr 13 sep 1977

THIS PAGE IS UNCLASSIFIED

THIS REPORT HAS BEEN DELIMITED

AND CLEAREL FOR PUBLIC RELEASE

UNDER DOD DIRECTIVE 5200.20 AND
NO RESTRICTIONS ARE IMPOSED UPON
ITS USE AND DISCLOSURE,

DiSTRIBUTION STATEMENT A

APPROVED FOR PUBLIC RELEASEj

DISTRIBUTION UNLIMITED.
 iti

0~ m~ed e ca alu
M4M I

! DOCUBEN T SEI|V CE CEN TER

-A AMII

as :IIL JAVITi.'lli
mff

FOOr,

"{-,
STI -
-J
------
- .- W
sumS
cammATWn
NorX-362(2o)/2

A DEFINITION OF STABLE INEASTIC MATERIAL

bV

E6 C. kueker

Technical Report N. 2
Division of WEngiring
Brown University
Providenoe, Rhode Islad

September 1957

Sponsored by
Departmsnt of the Navy
OffTic of Naval Research
Under catrect Nonr-562(20)
Task Order NR 4

Reproducrtion in whole or in part is permitted for

ru paose of the United States Ooverrnent.
I
Nonr-562(20)/2

A DEFINITION OF STABLE INIASTIC MATERIAl]

S*
D. C. Drucker (Brown University)

IAbstract
i The definitions given previously for work-hardening and perfect

plasticity are broadened to cover viscous effects. As before, the system

considered includes both the body and the time-dependent set of forces
acting upon it. An external agency is supposed to apply an additional

I set of forces to the already loaded body., It is now postulated that

the work done by the external agency on the addition displacements it

produces is positive. Some of the restrictions thus imposed on per-

i missible stress-strain relations are explored. Especial attention is

paid to simple laws of creep* Uniqueness of solution also is studied.

I
I
I
=1
1
I* The romults presented in thi-s p pr vere obtained in the course of
research ponsored by the Office of Naval Research under Contract
fonr.*56f(20) with Rrow tlnisAiVty.

I Choir"=# Division Of bV Ing.

 lon,--562(20)/2 2-

VIn Introdutton
the description of material, terms such as elastic, viscous,

and plastic are employed separately or in combination as in elastic-plastic,

visco-elastic, and visco-plastic. Their meaning Is quite clear in a
qualitative sense. The difficulty lies in obtaining a broad and pbsically

valid definition of terms which will lead to a useful mathematical

Sformulation of general stress-strain relations. Any restrictions imposed

limit the class of materials which are described and so are arbitrary

to a considerable extent, In this paper and Its predecessors (I) (2)*

the restriction is to stability of material in the strict sense*

An elastic body is reversible and non-dissipative and time

independent under isotheral conditions. All work done on the body is

stored as strain energy and can be recovered on unloading. The stress-

strain curve in simple tension may have a wide variety of shapes including

the unstable portions shown in Figuze 1. If stable elastic material is

specified the drop in stress with increase in strain must be ruled out

explicitly.
Plasticity denotes irreversibility and permanent or residual

L strain upon unloading. In the narrow but convenient terminology of the

Smathematical theory of plasticity, time independence of material properties
is understood despite the frequent use of terms such a,3 strain rate aM

1' velocitys Time alvays enters homogeneously in all equAtions of statics.

A given loading path producei the same deforations rmzardless of the

I actval time history of the loadings Figure 2 gives possible sizple tension
curves for a material combi linear elasticity and workihardening

plasticity. Tho phenomenon of the upper yield point in mild steel is

I -mbers in parentbeas refer to the abiograpty at the er of the

p*aor.
 Nonr'-562(2o)/2 --

a convincing demawtration of the fact that instability of material in the

mechanical sense is encounter ed in the physical world. A gradually falling P
stress-strain curve likewise is a physical possibility although not found

in structural metals when actual rather than nominal stress is plotted.

If stable elastic-plastic material is specified, therefore, the drop in 7

stress must be ruled out explicitly.

Time effects are present to soe extent in all real materials.

When the stress-strain curve is affected significantly by the rate of

loading or more generally by the time history of the loading, where tim

enters in an essential manner, the iscosity or the modifier visco-

is used. The idealization known as linear visco-elasticity may be

visualized in terms of combinations of linearly elastic springs and

Newtonian dashpots, Figure 3. Non-linearity of the springs and the

dashpots poses no conceptual difficulty nor does the inclusion of elastic-

plastic, "spring" elements. Falling stress-strain curves are easily pro- i

duced by a simple time history in the most elementary Kelvin or Maxbwll

models. Instability now has a more obscure meaning and certainly must
if involve time. For example, a falling curve on a plot of stress versus
strain-rate for a Maxwell type raterial would denote viscous instability, I

* Figure 3.

Previous Postulate_ for-Elastic and flastiao Materials.

A tundalnta defiion has been advanced previously for a

- stable elastic or Vork-huidenin t m i1 ndfnt materia (i) (2)0

Comider a body of uh ater at re stated upon by a set ofbuay

tractions and b* forces p Fre 4*- An external soncy is

J .mpoafl to add slowly a set of martte tractiats ATj &M-d bo4 fore
fl~The dislacaauts of theo b tfld chutwoe "as or*i #1
Nonr-.52.20)/2 -4-

and equilibrium is maintained* It was postulated that the work done by

the external agency during the application of force ust be positive and

over a cycle of application and removal# positive or zero, zero only when

purely elastic changes take place. In such a cycle, work cannot be

extracted from the system of the body and the set of forces acting

upon it. Materials following this postulated behavior are st;able. A

falling stress-strain curve in simple tension is ruled out as is far more

complicated behavior which normally might not be considered as unstable.

Despite the severity of the definition, the consequerces do seem in accord

with all xistirg intuitive concepts.

The implications or consequences of the fundamental definition

are remarkably strong for the mathematical theory of plasticity. The

yield or subsequent loading surface f(oy) - 2 which separates states
of stress which can be joined by a purely elastic path from those requirin

elastic-plastic action, Figure 5, must be convex. Furthermore, as shown,

the plastic strain increment or strain "rate" mist be normal to the surface

at a smooth point eij - Xaf/*ij or lie between the outward normals to

adjacent points at a corner.

Both convexity and normality had been postulated long before

(3) (4) and it is certainly valid to question the value of replacing

one set of postulates by another. One answer is that it is desirable to

have the saallest number of postulates. A better statement, perhaps, is

that ftdwavtal definitions have any specific implications beyorA those

of imadiate use, Dixrect, eperiental study and theoretic4al Ooosderation

of a baic postuIate form the basis of unlited tuture applicatimo

On th other hand, it "ould be ren bred that the do, of stAl
t-ninds~pe~nt mteria is a de~fion a#M not a lsv of natur, %tW#~b
uat~~~~~rJ~~~~ls~
i. id o~I lo ro~i1 -~ syatema of te-
 Nonr.562(20)/2

Coulomb type do not follow the postulate (5). A sub-class of materials,

therefore, is specified ar the problem reduces to how well particular

solids are described.

An Extended Postulate or Definition

The absence of time effects in the mathematical theory of

I elasticity and plasticity is not matched in the physical world (6).

Bodies under constant load will creep, bodies under constant deformation

J will relax. For many structural metals at working temperatures the

time effects are small enough to be ignored in practical applications

and the previous postulate is adequate. Howver, all metals at elevated

temperatures and most materials at room temperature display appreciable

time effects. A modified and broadened postulate, therefore, is required.

I ~When time or viscous effects are appreciable, it is no longer

true that the work done by an external agency need be positive. Spring

elements as in Figure 3 generally will contain energy which can be extracted

j from the system. A body under constant load usually does not maintain

an unchanged configuration. Application of additional force to an already

Jloaded body by an external agency is accompanied by a change in the dis-

placement history, partly due to the existing force system and partly

produced by the external agency. Figure 4 may now be reinterpreted.

Figure 4m shows the surface tractions Ti and body forces Fj. ich

are functions of time. The displacements ui, the stresses cij, and
the strains cij are likewise functions of time. Figure 4b shows the

existine system plus the xternal agency and the ccbined effeet. The
added surface tractions ATi and the added be forces AFi also vary
I with time as do the chage~s in dipUcent, stress, anA strain auj,

6Aaip Asijf wioh M fro the lotion of U. externAl W-noy

 onrumS(2o)12 -6.

superposed on the existing set of forces*' Except for the very special

fully linear case, it will not be true that, ATi and AF1 acting alone

(Ti = 0, Fi = 0) would cause Aui, Aaij, A6 1jo It is Ti + ATi and

Fi +tAF wbioh procuce u1 + U , aij + A a lj , J +A i j. all of which

i

are functions of time.

i A fundamental definition of a stable inelastic (elaste-visc-

plastic) witerial can now be stated:

The work done by the external- -ancy on the chne

in_diSlacements it roduces must be positive or zeros,

It is seen itmediately that the previous postulate is included in this

broader one because no change in displacement takes place in an elastic-

11 workhardening material at constant load.

The definition may be put in mathematical form as

Ptc Ifr
If {AT AiuidA + JAF AuidV~ dt >0[I
ilwhere the dot indicates derivative with respect to time$ and t -0 is
the instant of time at wbich the external agency begins to apply force.

The sumuation convention is followed for repeated subscripts in the same

term, for example AT:1,A = AT. Au, + ATV A + AT, and, for later
uses Clj tij OX -X + ay 87 + az St + Ix "fY + , Yz + or Yz

in the usual notation.

It is sometimes more inst-awtive to think of two alternative

patbsofioad.ig (T) an1 F:2) which diverge at t o,

Figure 6, Each w produce us ,

7Its a ij* T de ti than

has the xathmatical fore

I
 Nonr-562(2O)/2

tt {) (2) T l)] (2) dA)

fj ] ui -U dA

+ 4 F 1(I f 1] dV dt > 0

which is really identical to [L]. It has the advantage of focussing

attentL ,, on the meaning of the incremental quantities and the type of

stability which is postulated.

Stability in the -Large and S1all - Permissibility of Path

i l : : . m..
... . ',,, - :- EA MU
MMII

Sitability in the large and stability in the small correspond

respectively to t. unrestricted and to tc limited to the immediate

neighborhood of t = 0. As has been stated the requirement of stability

is an arbitrary one in a sense, The requirement that the tensile stress-

strain curve for an elastic-plastic body be a continuously rising one
(complete stability) is not of real consequence when the stress point

is on a portion of the curve which rises and small (infinitesimal) changes

in stress are examined (stability in the small).

There is a further choice involved within each classification

of stability. Any path labelled (1) in Figure 6 may be permitted or

instead Ti ) and F(l) may be restricted to constant values, their
J values at t = 0, It must be kept in mind always that inelastic
materials are irreversible and not all paths are permissible paths in

general. For example, at t = to the system will be at A if it follows

j path (1) and at B if it follows path (2)v With few exceptions. a straigbt
line path from A to B,. Tjnj w (f) M1), F 1 F (1 +

~j2 41>,with a gdingfrom 0 to 1$ wiflnotchbange the

State At (1) to the cUte at (2)* PsUrthenmre, no nelchborins path to

 Nonra..562(ZO)/2 -8-

the straightlizwj will exist which is capable of giving the same final state

B. In fact for the most general materials there Is no available path what-

soever from (1) to (2); path (2) is the only waty of prxducing the state

of the material at B.

In view of the irreversibility there is no clear decision to

be made on how much stability to require. Stability in the small with

path (1) restricted to the point corresponding to t = 0 wuld seem

essential. The generalized definition without restriction, or stability

of material in the complete sense, seems desirable. Some of its consequences

will be explored in what follows.

Stress-strain Relations

The conaequences of the basic postulate with respect to stress-

strain relations are strongest and seen most easily for a homogeneous system

whose mass or kinetic energy can be neglectedo Surface tractions, body

forces, and velocities in Inequality [ 2] then can be replaced by stresses

ij and strain rates tij througb the theorem of virtual work. For any

continuous u

ii Ti1. dA + J(,1 Vd 31
(S I

f where Ti, F', crJ are any set of equilibrium values and u j, ja

any compatible set (7). There need be no relation between the equilibrium
(2 )
i ~ ~~2) 1)frT
Ti
and the comptible sets. Substitution of T - Ti :'or '

for Ui, etc., in the virtual work expressiom [3] thus is permissible
(2- (2) (- alone will not prouce

"i - u i Integration over the volume can be omitted for a homogeneous

I ) "z
Nonr-56Z9.(20)/29-

state of stress and strain with the result

J f=0
taj (2) " ")]j
(2)" ( j iJ)]% t am [4

in the neighborhood of t = 0 the stresses and strain rates

can be expanded in a power series for each of the paths (i) anM (2), the

subscript zero denoting the value at t = o.

] t +a t 2/ oo
a0

ij]
=BJ 0 +
SiJ ajjajj 0 ij /2 +,e
ij t 2j]+
/2 00
0 00
ei j)0+ jt+0 &jt 4 /+

The two paths are identical up to so that stress and strain are
t=0
the same at t=0 (2) and 6 ]o o Time deriva-
ij 0
h a tO i, i 0
tives of these quantities need not be the same so that strain rates and

stress rates may differ. For to very close to t = 0, Inequality (4]

gives

*(2)(2) (l)
+(a. () .(2) () (210(2

+ i
aj-c0+ij 10 ai a ij -o iij 0iJ %o

+ ..0. >0 [61

As to may be made varishingly small, it is necessary that,

(2 ((2) 0(1)
[ij - ij ii ij . o
Nonr-5962 (20 )/2 _C

*(2) "(l)
If, however, "iJj
-iJ then it is necessary that

(2) "() '(2) 00(l)

] e
(a -a Ii:I >0 (8]
ij ij 0 ij j

This process may be continued to higher and higher time derivatives of

strain and of stress as weii if ;(2) ;(1), etc. The general requi:re-
ij iJ
ment in the sense of [7] and [8] is
(m(2) m() n (2) n (1)
8tm - - 0 tn 8atn o -9]
at at

with m and n each > 1. Inequality [9] insures stability in the

small in the broad sense.

If a j is restricted to remain constant at a ] as time

goes or, for the more restricted definition of stability in the small [7]

and (8] are replaced by

*(2) .(2) .( )Jo

aj
ii JO ij " eij _.0 [0

10
. (2)(10

i. j 0o[]

For time independent materials the (1) terms disappear conipletely leaving
. .
the earlier result ijeij > 0.

Return now to stability in the large and to tbe dependent

mterials. If the path OA of Figure 6 is traversed very rapidly and

then the system remains at A, and a similar time history holds for

path OB, the fundamental postulate in the form- [14] requires

B A B .A
(i ")
ij )(ij ii -
j Nonr-.,2(20)/2 -11-

In general, the strain rates ei and ij will depend, of course,

upon this specially selected time history. Here the distinction between

j the broad and narrow delinition of stability does not arise because the

path OAB traversed with extreme rapidity gives the same state point B

f as path OB, for materials which need time to deform.

Hwny special paths may be considered but their treatment will

be deferred• Specific results for plasticity, linear visco-elasticity,

and creep will be discussed next.

Consequences for Plasticity

As has been mentioned, the previous postulate for time-

~(1.)
independent material, with a held constant has strong implications
in plasticity. All yield and subsequent loading surfaces f(&3j. cj

I history of loading) = k 2 must be convex. The plastic strair rate vector

"p must be normal to the surface in stress space, Figure 5, in the

extended sense. At a smooth point of f =I, for A 0 or--

I an outward pointing a~ which produces plastic deformation*

Gji au Oum
mn [13]

3 where G may be a function of stress, strain, and the prior history

.
of loading. It may also depend upon the stress rate aij. Time is

not truly relevant so that G must be homogeneous of order zero in (,

e~,,G [1+a r )2
9en

IThe broader definition in this paper, of stability in the amall,

restricts G still more. It is permissible to deal with plastic strain

I
 No r-562(2o)/2 -12-

rates only and substitution in Inequality [7] gives

[ f ") ... .(1)] G()

,, (2). o(1) .(O (14
aa l al 'J
Baaf
[ o 2 a - [
i - kJ6u--

The magnritude of the stress rate is arbitrary and can be chosen to give

a constant value of the coaponent ij normal to the yield or

oa
loading surface. Suppose this is done for each pair (i) and (2)

so that the left bracket of (14] is zero. The bracketed expression

containing G(l) and G(2) then must likewise be zero. If not, a very
small increase or decrease in the magnitude of ij or the opposite
"ot op t
for cij will violate Inequality [14]. Therefore, (2) G( ) or

G is independent of aij* The stress-strain relation [13] thus is

restricted to linearity in the increments or rates of stress and strai

All incremental stress-strain relations now in use are linear or are

combinations of linear forms as a matter of convenience. Although no

'IT nre-conceived notions are upset, it is of great interest and possibly

of value to know that a single postu.1;te or definition leads inevitably

IT to all the main features of plastic stress-strain relations.

Stability in the small alone was employed in this application

Iof the postulate, but stability in the large is needed for convexity.

As stated before, there is always the possibility that stabill.ty in the

large with the new postulate may be more restrictive than is reasonable

1 on physical grounds for all paths of loading in all materials.

Consequences for Linear Visco-elsticity

A linear visco-elastic material may be defined in terms of

Smodels, Figure 3, with linear springs and dashpots in combinatioa (8).

An equivalet deftx tion my be Civen by ow or more differential equations

 -onr-562(2o)/2 - -

in which time derivatives of arq order may appear but stress and strain

enter linearly. The difference between two solutions of such equations

Is a solution as well. Therefore, if stress state and history a(1)

9(l) (2) "(2) ij
produce and stress state and history aij produce '(J
4(2) . (1) *(2) *(l)
iJ 0, will give e - Sij A stable linmar viscoelastic material
according to definition as expressed by the basic Inequality (4] must
obey

• =0 ci j e i j dt > 015

This is no surprise at all as it is simply a statement that

work must be done on an unstressed and unstrained viscoelastia material

in o-der to deform it. Any combination of linear springs and dashpots

t thus provides a stable material. Although the result is not unexpected,

it is comforting to know that the implications of the broadened definition

are in accord with physical intuition.

ConseqLuences for Cree

ie sNon-linearity is the rule rather than the exception in inelastic

materialse The creep of metals in particular is markedly non-linear.

Consider now a Maxwell type material, Figure %which deforms indefinitely

under constant load. The elastic response can be ignored in discussing

the viscous. Suppose that, for convenience and for reasonable agree-

I
ment with actual behavior, the viscous strain rate

of stress alone.
*iJ
ii
is a function

As the existing state of stress determL-es the strain

0(2 (2)
rate, the stress rate has no imnediate influence; [ci) i(
izi
in Inequality (7] is automatically zero at sny instant of time when

two paths of loading diverge. Inequalty (8] for stability in the snf,

hovevr, does provid, a nonatrivial nstriottn& on thbe oewV strain ationo

 Nonrs562(20)/2 el4e

Stability in the large gives far more spectacular and useful results.
Referring to Inequality [12], the restriction on the path or time

stn4 4 14 '+amA o V >0 appli4es to

any two stress states at ar time because c is supposed to depend upon

A
a alone and so is not path dependent. In particular, aA may be chosen

as the stress free state and the result is

B 3B
S J =>i) - 0 (16]
ij ij

or the rate of doing work is always positive. As the strain rate is

viscous, and, therefore, dissipative, this too is an obvious result where-

as (12] is not.
The restrictions on stress-strain relations in creep imposed

by [12] and (16] are strong and may be extored for each special

case. In pictorial terms as shown in Figures 7, 8, the strain rat vector

B ,B
makes an acute angle with the stress vector, 04 4 4 > O; the incrememt

in strain rate for any stress path is in the direction of the change in
stress (OBj atj )(sij _> 0.

jSuppose,for example, that it is possible to write

eta=aij
7" where (p is a function of the existing state of stress only. The surfaoe

= conrtant in stress space then bears some pictorial resemblance to the

yield suface in plasticity. The strain rate vector is normal to the

surface, As illustrated in Figure 8a, a stress path tangential to the
surface at one point cannot intersect the surface at another point. If

r it did, equality (12] vold be viaate4 vbmalty therefore lead*

to conveity of the surfa* m tant, Norality in the extended

 Nonr-562(20)/2 -15-

sense, as at a corner of such a surface likewise requires convexity if

[12] is to te satisfied, Figure 81b

,kcomnon assumption for creep in simple tension (not necessarily

in accord with the actual physical behavior) is that the strain rate is

proportional to some power n of the stress (9). A generalization of

this hypothesis is to choose p in [17] as a homogeneous function of

stress of degree n + l. Substituting in (16], the rate of doing work

or rate of dissipation is from Euler's theorem on homogeneous functions

W(aij) - £n =l = (lp (18]

Under these conditions p = constant is a surface in stress space of

constant rate of dissipation of energy. TLerefore W constant defines

a convex surface in stress space. 0

The additional assumption of isotropy requires W to be a

ftnction of the stress invariants, for example; the sum of the principal
stresses J and the invariants J2j = 1 sjk si of
21 sij sivJ :5 JsksiO
the stress deviation tensor sij, (10). Considerable discussion of convex

homogeneous functions of stress with these invariants has been given in

papers on plasticity (10)(11) and need. not be repeated here. As an

example, the requirement of convexity restricts the constant c in

(n+l)2 = W -= W3 - to no more than 2.25*

Uniaueness of Solution to Equilibriui Broundar Value Problems (12)(13)

Consider a body at time t - 0 whose state at each point can

be specified completely as, for example, the initial unstressed and

unstrained condition. Suppose surface tractions Ti are specified on

the portion of the surface ii, diaplacements u on the remaining

I
Nonr-562(20)/2 -j6-

portion Au, and body forces Fj throubhout the volume V, each given

as a fumction of time. The question of uniqueness concerns the possibility

of tio (or more) solutions 1 and n for the interior stresses, strains,

and displacements or any of their time derivatives.

Assume that two equilibrium solutions do exist which satisfy the

boundary conditions. Substitute their difference in the equation of virtual

work [3] Just as in the description following Equation [3]. The general

result is that tae left hand side is always zero and for any order of tim

derivative of stress and strain

0-J [a S m j M
a~Oj
i 2) " C3' M GiJ
1)
] [ jan &t
"MM
ii(2) (1) ]
an. 6j dV [19]1
[9

which is reminiscent of Inequality '[9] except that here the notation

is meant to include the zero order as i.e.

,ell#

(2) (1) (2) (2) " () " ] ' etc.

[cit J " ij ( , [ () (). (2) am(1)I

ij o
(2)= =in napai
Uniqueness the sense of ~ .or= or
(2.()~ 8 tm ii at m

at nn (1) is assured if the integrand of [19] is positive.

8t
_n ij
Just as in the special case of plasticity, a discussion of uniqueness in j
ouch great generality cannot be more specific (13). Any lack of uniqueness

which may be present in the stress-strain relation appears here as well*

With this understanding the term uniqueness will be used without qualification

in the following.

Reference to Inequality [7] shows that stability in the small

guarantees uniqueness of ai, or 4 j at each instant of time. If e

i3 dependent on a and not is nequalty [8] takes over and rj

or u:j are shki~n directly to be uiqueo This type of consideration is

Nonr-562 (20)/2 -l?-

of importance in the study of dynaic problems in which mass or inertia

cannot be neglected. If needed, this process can be continued to any term

of Inequality ( 9.. The initial state and the complete history of loading,

there.fore, determine the stress and the strain uiquely throughout any

body composed of inelastic material obeying the definition of stability

in the small.

It might be expected that stability in the large guarantees

uniqueness of stress independent of the path of loading* This is true

for any material undergoing the rapid loading type of path described for

Inequality [ 12]. It also is true for those stable materials following

any creep law in which the strain rate depends upon the stress only because

they too most obey Inequality ( 12 1. It is not true for most materials.

Certainly the fundamental definition does require that if a

path exists from A to B, Figure 6, the work done by the external agency

on the change in displacements it produces must be positive or zero,

j (aij-oA)(sij-iA) dt> 0. However, in general the integral bears little
A
or no relevance to the virtual work term (a i - A) whose positive-

ness assures uniqueness. Prthermore, no path AB is a permissible one

in the general case.

All of the emphasis so far has been on equilibrium problems. Mass

and kinetic energy have been ignored, although velocity and acceleration

terms are finite. Aa wil be seen, the inclusion of dynamic terms does not

affeet the conclusions to be draun and, in fact, makes les demands on the

stress-strain relations, Hoiogeneous states f strers and strain me not

appropriate and tha steps uhich lead to Inequaty (4] Cive instead the
 ffonr-562(20)/2 -18-

more inclusive

fto (r2) (1)3 ;(2)C (2b d2 d+

where the second term is the value of the v~olume integral at t o minus

the value at the time of divergence of paths, t = O. If the external

agency applies load slowly, the acceleration ui is the same for the

two paths (1) and (2), Figure 6. Necessarily, in all cases the velocity

(2) throughout the volume is the same as ( at tO andthe

1

second term of [20] is positive or zero at all times t 0 The Inequality

I [ 4] is much more significant, therefore, than (20].

! , Conclusions

A single postulate or definition is ad-vanced 'or 01 stable in-

elastic -ateriase Its im.licatiiis are far reachings In plasticity

V theory, which is a particii.arly well-explored subjec#t, the conrcepts of

convexity, normality, and incremental linearity all are consequences of

the definition. An interesting result also is obtained for a generalization
I@
of the simple tension creep law 6 = KA The mirface of -onstant rate of

dissipation for the special case is convex and e is nozial to it just as

r for the yield surface in plasticity. Hwever, non-liyar= viscous behavior

coupled with elastic and plastic action is an almcst unknown field. .ich

F more must be done before the value of the postulate cw be determined.

Stability in the large is an especially troublesome point. Nevertheless,

Kthe uniqueness of solution assured by the definition is a good indication

of its acceptability as a startin point.

 Nom, 562 (20)/2 -

Bibliogrsh,
(1) "Soe Implications of Work Hardening and Ideal Plasticity" by
D. C. Drucker, Quarterly of Applied Mathematics, vol. 7, no. 4.,
January 1950, pp. 411-418.
(2) "AMore Fundamental Approach to Plastic Stress-Strain Relations"
by D, C, Drucker, Proceedings of the First National Congress of
Applied Mechanics, ASME, 1951, pp 487-491.
(3) 'echanik der plastischen Formanderung von Kristallen" by R. v. Mises,
ZAMZ, vol. 8, 1928, pp. 161-185.
(4) "Recent Developments in the Mathematical TheorI of Plasticity" by
W. Prager, Journal of Applied Physics, vol. 20, 1949, pp. 235-241.
(5) "Coulomb Friction, Plasticity, and Limit Loads" by D. C. Drtuker, Jounal
of Applied Mechanics, v0o 21, Trem. ASME, vol. 76, 1954, PP. 71-74.
(6) "Rheology Theory and Applications, Volume I", Edited by F. R. Eirich,
Academic Press, New York, 1956.

(7) "Theory of Perfectly Plastic Solids" by W. Prager and P. G, Hodge, Jr.,

John Wiley and Sons, New York, 1951.
(8) "Mechanical Properties of High Polymers" by T. Alfrey, interscience,
New York, 1948.

(9) "Recent Advances in Theories of Creep of Engineering Materials" by

F. K. G. Odquist, Applied Mechanics Reviews, vol. 7, no. 12, 1954,
Pp. 517-519. Refers to work of Bailey, Hoff and Main among others.

(10) "Strain Hardening under Combined Stresses" by W, Prager, Journal of

Applied Phsics, vol. 16, 1945, PP. 837-840.

(11) "Some Extensions of Elementary Plasticity Theory" by F. Edelman and

D. C. Drucker, Journal of the Franklin Institute, vol. 251, 1951,
pP* 581-6c5.
(12) "The Mathematical Theory of Plasticity" by R. Hill, Oxford University
Press (1950).

(of (13) "On Uniqueness in the Theory of Plasticity" by Dc C. Drucker, Quarterly

Applied Y~thematics; vol. 14, 1956, pp. 35-42.
ApleImhmtc p
I
I
 Nonr 362(20)/2

440

FIGURE 1. AN ELASTIC MATERIAL (REVERSIBLE): CURVE

RISING TO THE RIGHT O'E>O

UNSTABLE

FIGURE 2. AN ELASTIC-PLASTIC MATERIAL (TIME-

I INDEPENDENT): RISING CURVE ol>0
Nonr 62(20)/2

iii- rT
KELVIN MAXWELL
OR
VOIGT

FIGURE 3. VISCOELASTIC MODELS

Ti Ti + AT i

Fi + AF;

u i , O'ij eij ui+l' u i , Orij+acfij , eij+Af.ij

ta) (b)

FIGURE 4. EXISTING SYSTEM (a) AND EXISTING SYSTEM

PLUS EXTERNAL AGENCY (b)
Nowr 562(20)/2 4022-

YIELD OR

SURFACE II

FIGURE 5. THE FUNDAME~tA7L DEFINITION REQUIRES

CONVEXITY AND NORMALITY IN
PLASTICITY

2)

(AVpF

FIGURE 6. PATHS OF LOADING. (Arrows join points

of the some time)
Nonr 562(20)12 0023-

OF PATH

(I PAT H

FIGURE 7. [(' a.. 0 0 0

Ii ei
Ai

PERMMISSIBLE

(~) (b)

FIGURE 8. (a)NORMALITY, ao:/a~rip AND INEQUALITY [121

REQUIRES CONVEXITY. (b) A CORNER ALSO IS PERMISSIBLE

 DISTRIB TION LIST FOML UNLASSIFIED
TECHNICAL REPORTS ISSUED UNDER
Contract Kon-562(2O), Task Nr 064-424

Chief of Naval Research Ar-med Services Technical Information

Department of the Navy Agency
washington 25, D. C. Doment Service Center
Attn- Code 438 (2) Knott Bilding
Code 463 (1) Dayton 2, Ohio (5)
C an Officer Office of Technical Services
Office of Naval Research Dev'i nt of C- rce
Branch Office ashington 25, P. 0 (1)
495 Summer Street
Boston 10, Massachusetts (1) Office of the Secretary of Defense
Research and Development Division
Commanding Officer The Pentagon
Office of Naval Research Washington 25, I C.
Branch Office Attn: Technical Library (I)
John Crerar Library Building
86 E. Randolph Street Chief
Chicago 11, Illinois (1) Armed Forces Special Weapons Project
The Pentagon
Commanding Officer Washington 25, Do C.
Office of Naval Research Attn: Technical Information Div. (2)
Branch Office Weapons Effects Division (1)
346 Broadway Special Field Projects (1)
New York 13, New York (1) Blast and Shock Branch (1)
C-rndin Officer Office of the Secretary of the ArEW
Office of Naval Research The Pentagon
Branch Office Washington 25, Do C.
1030 E. Green Street Attn: Army Library (1)
Pasadena, California () Chief of Staff

Coamanding Officer Departmnt of the Arvy

Office of Naval Research Washington 25v D C.
Branch Office Attn: Development Branch
1000 Geary Street Rese and Dev. Division (I)
San Francisco, California (1) Research Branch
Res. and Dev. Division (I)
Commanding Officer Special Weapons Branch
Office of Naval Research Res. and De. Division (1)
Navy #100, Fletet Post Office
New York, New York (25) Coaua Officer
Engineer Research Development
Director Laboratory
Naval Research Laboratory Fort Belvoir, Virginia (I)
Wiashington 23$ Do C.
Attnt Tech* Info Officer (6) Office of the Chief of Ordnance
Code 6200 (1) Department of the Anu
Code 6205
Code 6250
(1)
(1)
Washington 25, Do C.
Attn: Research and Hater als Branch

Code 6260 (1) COrd. Res, and Devo Div.) 11)

Nonr-562(20) Distribution List -2u

Office of the Chief of Engineers Comn-ant, Marine Corps

Department of the Army Headquarters, U. So Mari ne Corps
Washington 25, D. C. Washington 25, D 0. (1)
Attng EI.PA-HL ib, Bro, Adnu Set.
Div. (1) Chief, Bureau of Ships
ENG-WE Eng. Div., Civil Department of the Navy
Works (1) Washington 25, D* C.
ENG-EB Prot. Constr. Br., Attn: Code 312 (2)
Eng. Div., Mil. Code 376 (1)
Constr (1) Code 377 (1)
ENG-WD Planning Div. Civil Code 420 (I)
'Wrks (1) Code 423 (2)
ENG-EA Struc. Br., Eng. Div., Code 442 (2)
Nil. Constr. (1)
ENG-NB Special Engr. Br, Chief, Bureau of Aeronautics
Engo Res. and Dev. Department of the Navy
Division (1) Washington 25, Do C.
Attn: AE-4 (1)
Office of the Chief Signal Officer AV-34 (1)
Department of the Armny AD (1)
Washington 25, D. C. AD-2 (i)
Attn: Engineering and Technical GM (i)
Division (1) FR-7 (1)
RS-8()
Commanding Officer SI (I)
Watertown Arsenal TD-42 (1)
Watertown, Massachusetts
Attn: Laboratory Division (, % -"na
Department of the Navy
Commanding Officer Wrashington 25, Dt C,
Frankford Arsenal Attn: Ad3 (I)
Bridesburg Station Re (I)
Philadelphia 37, Pernsylvarda Res (i)
Attn: Laboratory Division (1) Reu (I)
ReS5 (1)
Office of Ordnance Research ReSl (1)
2127 Myrtle Drive Ren (1)
Duke Station
Durham, North Carolina Chief, Bureau of Yards and Docks
Attn: Division of Engineering Department of the Navy
Sciences (1) Washington 25, Do C.
Attn: Code D-202 (1)
Commanding Officer Code D-202,3 (1)
Squier Signal Laboratory Code D-220 (1)
Fort Monmouth, New Jersey Code D-222 (1)
Attn: Components and Materials Code D-410C (1)
Branch (i) Code D-440 (1)
Code D-500 (1)
Chief of Naval Operations
Department of the Navy Commanding Officer and Director
Washington 25, Do C. David Taylor Model Basin
Attn; Op 37 (1) Washington 7, D. C.
Attn: Code 700 (6)
Code 720 (2)
Code 740 (1)
 onr-562(20) Distribution List -3-

Commander
Uo S. Naval Ordnance Laboratory
k ihte Oak, layland
Attn: Technical Library
Technical Evaluation
14 Department
Ccvuander
Naval Ordnance Test Station
Underwater Ordnance Division
(2) 3202 E. Foothill BouLevard
Pasadena 8, California
(1) Attn: Structures Division (1)

Director Superintendent
Materials Laboratory Naval Post Graduate School
~~New Yor.. w_!-2 Siyalrd --- - -- ,-"
()
Brooklyn 1; New York
Comandit
Commanding Officer and Director Marine Corps Schools
U. S. Naval Electronics Laboratory Quantico, Virginia
San Diego 52, California (1) Attn: Director, .Marine Corps
t Officer-in-Charge
Naval Civil Engineering Research Coz 4dng General
and Evaluation Laboratory U. S. Air Force
U. S. Naval Construction Washington 25, D. C.
Battalion Center Attn: Research and Developiment
17 Port 1ueneme, California (2) Division (1)
Director Comannder
Naval Air Experental Station Air Material Comnd
Naval Air Mterial Cepter WrIght-Patterson Air Force Base
Naval Base Dayton, Ohio
Philadelphia 12, Pennsylvamia Attn: MREX-B (1)
Attn: Materials Laboratory (1) Structures Division (1)
Structures Laboratory (1)
Cczander
r Officer-in-Charge U. S. Air Force Institute of
Underzater Eplosion Research Division Technology
Norfolk Naval Shipyard Wright-,Patterson Air Force Base
Portsmouth, Virginia Dayton, Ohio
Attn: Dr. A. H. Keil (2) Attn: Chief, Applied Yechanics
Group (I)
Commander
. U. S. Naval Proving Grounds Director of Intelligence
Dahigren, Virginia (1) Headquarters, U. S. Air Force
Washington 25, 1. C.
Superintendent Attn: P. V. Branh
Naval Gun Factory (Air Targets Division) (1)
Washington 25, D. C.
4WCommier
Commander Air Force Office of Scientific Research
Naval Ordnance Test Station Washington 25, Do C.
Inyokern, China Lake, California Attn: 3echanics Division (1)
Attn: Physics Division (i)
Mechanics
i Branch () U. S. Atomic Energy Coaission
tbshington 25, Do C.
Commanding Officer and Director Attn: Director of Research (2)
Naval Engineering Experiment Station
Annapolis, 1aryland (1)
 Nonr-562(20) Distribution List
Director
National Bureau of Standards
Washington 25, D. C.
Attn: Division of Mechanics
Engineering Mdechan ics
Section
Aircraft Structures
Commandant
U. S. Coast Guard
1300 E. Street, N.'1.
Washington 25, D. C.
Attn: Chief, Testing and
Development Division
U. S. Maritime Administration
General Administration Office
Building
Washington 2 , D C.
Attn: Chief, Division of
Preliminary Design
National Advisory Committee
for Aeronautics
1512 H Street, N.W.
Washington 25, D. C.
L Attn: Loads and Structures
Division
Director
Langley Aeronautical Laboratory
SAttn: Langley Structures
Field, Virginia
Division
Director
Forest Products Laboratory
Madison, Wisconsin
Civil Aeronautics Administration
Department of Commerce
Washington 25, D. C.
Attn: Chief, Aircraft Engineering
Division
Chief, Airframe and Equipment
Branch
IDepartment
National Sciences Foundation
1520 H Street,, N.th
Washington, D. C.
Attn: Engineering Sciences
Division
-4-
National Academy of Sciences
2101 Constitution Avenue
Washington 25, D. C.
(1) Attn: Technical Director, Committee
on Ships' Structural Design (1)
(1) Executive Secretary, Comvittee
(1) on Undersea Warfare (1)
Professor Lynn S. Beedle
Fritz Engineering Laboratory
Lehigh University
Bethlehem, Pennsylvania (1)
(1) Professor R. L. Bisplinghoff
Department of Aeronautical Engineering
Massachusetts Institute of Technology
Cambridge 39, Massachusetts (3)
Professor F. H. Bleich
Department of Civil Engineering
(1) Columbia University
New York 27, New York CL)
Professor B. A. Boley
Department of Civ.il Engineering
Columbia University
New Yorc 27, New York (1)
(2)
Professor G. F. Carrier
Pierce Hall
Harvard University
(2) Cambridge 38, Massachusetts (1)
Professor Herbert Deresiewicz
Department of Civil Engineering
Columbia University
(1) 632 W. 125th Street
New York 27, N. Y. (1)
Professor D. C. Drucker, Chairman
Division of Engineering
Brows University
(1) Providence 12, IRhode Island (1)
(1) Professor A. C. Eringen
of Aeronautical Engineering
Purdue University
Lafayette, Indiana (I)
Professor W. Flugge
(I Department of Hechanical Engineering
Stanford University
Stanford, California (1)
 Professor Jo No, Goodier
Departmient of YMecmaical
En~ineerlng
Stanford University
Stanford, California
Professor ". E. Goodman
Engineering Experiment Station
University of 11innesota.
Minnepolis, Minnesota
Professor 1-U Hetenyi
The Technological Institute
Northwestern University
Evanston, Illinois
Professor P. G. Hlodge
Department of eaeronautical
Engineering
Illinois Institute of Technoloyekly4
Chicago 16, Illinois
TiProfessor N. J. Hoff
Department of Aeronautical
&ngineering
Stanf ord University
Stanford, California
P1rofess or 4 He Hoppmann, T
Department of 1-1echanics
Rensselaer Pol.ytechnic Ihstitute
Troyq New York
Professor Bruce G. Johniston.
University of Michigan
Ann Arbors Michigan
Profsso J.KemnerDepartment of Engineerizg
ProfssorJ.
empnr
Department of Aeronautical
-gineering and Applied Mechanics Anm Arbor, M~ichigan()
Polytechnic Institute of Brooklyn
99 Livingston Streiet
Brooklyn 29 New York
Professor H. L. Langhaar
Department of Theoretical and
Applied Hechanics
University of Illinois
Urbana, Illinois
Professor B. Jo Lazan, Di rector
1University
Engineering Experiment Station
of "Kinnesota.
11inneapolis 14,* Minesota
Ii~b
Professor IS.Hi. Lee, Chairman
Division of Applied Yathematics
Browi University
Provide-nee 12, Rhode Island (1)
(1) Professor George He lee
Director of R~esearch
Rensselaer Polytechnic Institute
Troy, 1~ew York (1
(1)
Sotwst Research Institute
8500 Culebra Road
S:an Antonio 6, Texas (1)
(1)
Professor Paul Lieber
Geology Department
Untiversity of California
Bzcly4,Clfri
Clfri 1
(1) Professor Hsu Lo
School. of Engineering
Purdae UniJ.versity
Lafayette, Indiana (1)
Prof essor F4 D. Mkxini
(1) Dspartment of Civil Fngineering
Columbia University
632 . 5th street
New York 27, New York(1
() Dr. As INA -I
136 Cherry Valley Road
Pittsburgh 21s Pennsylvania()
() Professor ?al 1-1. N#aghdi
Mechanics
University of Mttehgan
Professor Wil1liam ANash
() Departmnt of Rrgineertng Hechamics
University of Florida,
Gainesville, Florida()
Professor -4. 14. Newmark,, Heed
Despartm~ent of Civil bigizeering
() Univem. ity of ]llinois
UbnIlni
()
Professor Aris Phillips
Departmezit of Civil &gineering
15 Promwect Street
=al University
Nov rn, Corzacti0ttl
 Nonr-562 (20) Distribution List -6-

Professor Wo Prager, Chairman Dr. John F. Brahtz

Physical Sciences Council Department of Engineering
Brown University University of California
Providence 12, Rhode Island (1) Los Angeles, California (i)

Professor E. Reissner Mr. Martin Goland, Vice President

Department of Mathematics Southwest Research Institute
Massachusetts Institute of Technology 8500 Culebra Road
Cambridge 39, .assachusetts (1) San Antonio, Texas (i)

Professor Mq A* Sadowsky Mrs S Lev-y

Department of -lechnics General Electric Research Laboratory
Rensselaer Polytechnic Institute 6901 Elmwood Avenue
Troy, New York (1) Philadelphia 42, Pennsylvania (1)
Professor Jo Stallmeyer Professor B. Budianaky
Department of Civil Engineering Department of Mechanical Engineering
University of Illinois School of Applied Sciences
Urbana, Illinois (1) Harvard University
Cambridge 38, Iassachusetts (i)
Professor Eli Sternberg
Department of Applied Mathematics Professor H. Kolsky
Brown University Division of Engineering
Providence 12, Rhode Island (1) Brown University
Providence 12, Rhode Island (1)
Professor S. P. Timoshenko
School of Engineering Professor Et Orowan
Stanford University Department of 1leohanicr1. Engineering
Stanford, California (1) Massachusetts Institute of
Technology
Professor A* So Velestos Cambridge 39. Massachusetts (1)
Department of Civil Engineering
I University of Illinois
Urbana, Illinois (1)
ji Professor Enrico Volterra
Department of Engineering Mechanics
University of Texas
j Austin, Texas (I)
Professor Dana Young
Yale University
New Haven, Connecticut (1)

I
!
I
I