AVERAGE

STRESS

IN MATRIX AND AVERAGE ELASTIC ENERGY

WITH MISFITTING
INCLUSIONS*
T.

MORIt

and

K.

OF MATERIALS

TANAKA7 $

Hrtving noted an important role of image stress in work hardening of dispersion hardened materials,*.3)
the present paper discusses 8 method of calculating the average internal stress in the matrix of a material
containing inclusions with transformation strain. It is shown that the average stress in the matrix is uniform throughout the material snd independent of the position of the domain where the average treatment
is carried out. It is also shown that the 8ctual stress in the matrix is the average stress plus the locally
fluctuating stress, the average of which vanishes in the matrix. Average elastic energy is also considered
by taking into account the effects of the interaction among ths inclusions and of the presence of the free
boundary.
COSTRAIXTE

MOYENNE
XATERIAUZi

DANS LA MATRICE
ET ENERGIE
CONTEKATUT DES INCLUSIONS

ELASTIQUE
IMPARFA~S

MOYENNE

DES

Ayant remarque que la force image joue un &le important dans la consolidation
des mat&iaux
durcls par dispersion, les auteurs proposent ici une m6thode de calcul de la contrainte inteme moyenne
dam la metrice dun mat&iau conbnant des inclusions pr&entant des d6formations dues $ une transformation, et montrent, que 18 contrainte moyenne dans la matrice est uniforme 8. travera le matbriau
et indbpendsnte de 18 position de la zone dans laquelle le traitement moyen est effect&.
11s montrent
aussi que ia contra&e
r&&e dans 18 matrice est &8le $, 18 somme de 18 contrainte moyenne et de la
contfsinte locale variable dont 18 moyenne pour toute 18 metrice tend vers z&o. L&ergie &stique
moyenne est Bgalement calculb en tenant compte des effets dinteraction entre les inclusions et de 18
presence du joint libre.
DIE

JlITTLERE

SPANNUNG
IN DER MATRIX
ENERGIE
VON MATERIALIEN

UND DIE MITTLERE

MIT EINSCHLUSSEN

ELASTISCHE

rvschdem die gro&. Bedeutung der Bildkraft fiir die Verfestigu~

von dispersions geh~e~n
Materi81iens+SL beront wurde, diskutiert die vorliegende Arbeit eine Method8 zur Beschreibung der mittleren inneren Spannung in der Matrix eines Materials, das Einschliisse mit UmW8ndlUngSVen3p8nnUngen
enthitlt, Es wird gezeigt, d8B die mittlere Spannung in der Matrix im ganzen Material gleichfiirmig und
unabhitngig van der Lage des Bemichs ist; fiir den die Behandlung durchgefiihrt wurde. AuRerdem wird
gezeigt, d8B die aktuelle Spennung in der Matrix gleich der mittleren Spannung plus einer lokat
fluktuierenden Spannung ist, deren Mittelwert iiher die gesamte Matrix verschwindet.
Die mittlere
eIastische Energie wird ebenf8lls diskutiert
unter ~~c~~ehti~ng
der W~hseIwjrkung~~ekte
znischen den Einschliissen und der Gegenwart der freien Oberfliiche.

1. INTRODUCTION

Brown pointed out in describing the present

situation,(f) some papers have recently appeared which
discuss work-hardening of dispersion hardened material~.(*-~) All these papers deal with internal stress
which developed as a result of plastic deformation
occurring only in the matrix. References (2) and (4)
pursue energy consideration and agree with each other
with respect to the hardening rate, while references
(3) and (5) discuss the role of int,ernal stress in the
matrix in work hardeni~~~. Although Brown presented
a comprehensive vipn- of what was discussed in these
papers, (1) we would like nonetheless to report our
own understanding. for Browns explanation and
derivation of some stresses and strains were, in details,
not completely acceptable from our point of view and
we feel that an alternative treatment is possible.
Therefore, the results of some of our calculations
of the internal stress and of elastic energy will be
presented here. Since the internal stress developed
-4s

by uniform plastic deformation occurring only in the

matrix can be duplica~d by giving uniform tranaformation strain to inclusions alone, internal stress will
be charscterized in the following sections by inclusions
with uniform transformation strain. Thus, some ofthe
results will be directly applicable to a material with
misfitting precipitates.
2. AVERAGE

INTERNAL

STRESS

IN

MATRIX

For simplicity, the case where elastic constants

Ciinl are uniform throughout a specimen V, will be
considered. Suppose the specimen has N inclusions
which are, on the macroscopic scale, uniformly distributed in the matrix. When an inclusion V acquires
uniform transformation strain EijT, total strain eij is
introduced into the specimen. E<~is divided into
two parts; cijm and eijim. Eij * represents constrained
strain* (total strain) when the inclusion is in an infinite
body, and E~: represents image strain introduced
into the actual specimen by the presence of the free
boundary 1V,l of the specimen. &ijim is elastic in
is elastic
nature throughout the specimen.
&ii *
outside the inclusion, but ~~~~~- cijT) is elastic

* Received July 22, 1972: revised September 8, 1972.

t Department of Metallurgic81 Engineering, Tokyo Institute
of Technology, Ookayama, Meguro, Tokyo, Japan.
$ Now at: National Research Institute for Metals, Nakameguro, Meguro, Tokyo, Japan.
-.
571
ACTA METALL~RG~~A,
VOL. 21, MAY 1973

* In Eshelbys notation eija is expressed 8s size.

ACTA

572

inside the inclusion.

Epj* = E,,
+

NETALLURGICA,

cij m is given by

4j, li(xa~)1/2}
dDtx)>

(1)

where Gki is Greens function for an infinite homogeneous body with elastic constants Cijkl.@) It can be
shown t,hat when aU inclusions acquire identical
transformation strain pijT, the specimen as a whole
undergoes a shape change and the average strain,
(Q~~)~~, that descibes the specimen shape change is
given by
(EijF)gTb

fe*jTt

(2)

where f is a volume fraction of the inclusions. From

this, let us assume the following : (1) the average of
total strains, in domain V,, from all the inclusions,
is also equal t,o fciiTif VR is a representative domain
for the specimen. This implies that V, has a sufficient
number of inclusions, say, Y inclusions, MV/V,
= f,
and VR is not in a neighborhood close to the boundary
1V,,l. Assumption (I) is realistic. As an example, let
us consider the lattice constants of martensite. The
macroscopically determined ltverage lattice constants
of a martensite single crystal (if it were present) are
believed to be equal to those measured by a narrow
beam X-ray technique if interstitial atoms are uniformly distributed. Assumption (1) was justified by
Eshelby when &ijT = &aij (bij is the Kronecker
delta).?)
Let us consider the case where V, has a shape
similar to that of the inclusions. Then total strain
cijF is expressed as
M
P>M

s
-j- 2
I=1

e,,iyx.xPf,

(3)

where .zijm(x, xp) is the quant,ity defined by equation

(1) when the P-th inclusion is at xp, ~~~~~(5,
sp) is t.he
image strain due to the P-th inclusion, and VR is
assumed to contrtin inclusions 1 to i@. Apparently,
the second and the third sums are eIast,ic and do
not fluctuate much in VR. Therefore, it, is meaningful
to define the average of the second and the third
sums in equation (3) in V, as t,he average elastic strain
(%j)Y,3

VOL.

21,

1973

As shown in a previous paper,(*) when a single inclusion

V with uniform transformation strain is within an
infinite homogeneous body, volume integrals of total
stmin and stress vanish if the integration is carried
out, in the region V - I, where V is a domain
surrounding V and is of shape similar to that, of I.
Applying this conclusion, the average of the first sum
of equation (3) in V, is calculated as
ilf

where XijKlare Eshelbys tensors and are equal to the

integrals in equation (1) when x is within I in equation (l).(6) Assumption (1) and equat.ions (3) through
(5) give
(&ij)yn

-f

K%jmnemnT

ejjT)-

(6)

Thus, the average elastic strain defined above is

independent of the posit,ion and the size of I,.
The average internal &ress, (~i~~o)~-~,
defined by
(Oi$O>va= ~i$&&-R

(7)

is also independent of the position and the size of

V,, insofar as VR can be regarded as a representative
volume. (aijo) v, is the average in V, of the sum of
the image stresses of all the inclusions and the stresses
of the inchtsions outside V, when they are in an
infinite body. However, it should be noted that. this
sum itself is nearly constant in V,.
Next, let us consider the following sum,

where aij m(x, zp) is internal stress due to the P-th

inclusion at xp within V, when it, is in an infinite
body. This sum, ai,* (1, M) cannot be assumed nearly
instead, it fluctuates and t,he wave
constant in V,;
length of the fluctuation is apparent.ly of the order of
the inter-inclusion spacing. cijm (1, X) is called 1ocaIIy
fluctuating stress, to which nearby inclusions obviously
contribute predominantly. However, the average of
this locally fluctuating st,ress in the matrix of Vn can
be shown to be zero. Instead of averaging equation
(8) directly, let us take the following approach.
Consider a fixed point in V,.
First imagine V,
without inclusions. Next introduce an inclusion into
V, such that the fixed point is always outside the
inclusions and record (T$~~~V. Here, uilm is the stress
due to the inclusion when it is in an infinite body snd
SV is a small volume element around the fixed point.
If the introduction of the inclusion is repeated many
times in a random manner, Caij6V becomes proportional to the integral, Sv,_ v aiim6 V, where the center
of the inclusion is conversely fixed at the opposite

NOR1

AND

TAKAKA:

MATERIALS

position of the above fixed point from the center of

Ti,. Because of the statement following equation (4),
this integral vanishes. Since this is t,rue for every
inclusion, the average of ~iirrnfrom ali the actually
present, inclusions in VR at a point within the matrix in
V, becomes zero. That is,
@ij(ll M)), = 0.
(9)
The total average stress, (CT&~, in the matrix within
1B is the sum of {G,~O}~~and (~,~(l, M)jnl. From
equations (6), (7) and (9),

WITH

MIISFITTISG

INCLUSTOSS

673

when it is flexible. Brown and St,obbs also considered

t,he role of the locally fluctuating stress in work
hardening in a somewhat di~erent manner.(3~ It is
important to note that, hardening due to this factsor
operates as an energy dissipation mechanism.*5)
Because of its fluctuating and position-dependent
character, estimation of hardening due to the locally
~uctuating stress involves certain approximations and
seems to depend on the choice of particular situations
to be considered; thus, it would not seem to be as
uniquely and simply performable as estimation of
hardening due to the average stress in the matrix,
taij)M*

with
Csijr* = %?JL&nnT

4%

(11)

3. ~NTE~CTIO~
ENERGY

AMONG
INCLUSIONS
CONSIDERATION

AND

That the average internal stress in the matrix is

where oiila? is stress within a single inclusion when it is
equal to -fa..I naturally indicates that the actual
in an infinitely extended body. {aJIM is independent
stress within cn inclusion is, on the average, equal to
of the size and position of V, and of the shape of the
aijlm - foijIco. Thus, elastic energy per unit. volume
specimen.
of the specimen, EeEft,is given by
In summary, the internal stress at any point in the
E,1 = -f(l
- f)oii%ilT/2.
(12)
matrix is the uniform average stress (aJM plus the
In Reference (2), the term (1 - f) was omitt,ed on the
locally fluctuating stress from nearby inclusions.
This locally fluctuating stress is averaged to be zero assumption that the interaction among inclusions can
in the matrix. It is important to note that (a&X is be ignored. However, equation (12) is a correct
expression which takes into account interaction among
the average, within the matrix, of the sums of the
stresses of the inclusions when they are present in an inclusions together with the effect of the presence of
infinite body (Cbijm) and of the image stresses of all the free boundary of the specimen. If f is small,
(1 - f) in equation (12) can be, for all practical
the inclusions (X(rijim). The relative contributions of
purposes, replaced by unity. However, whenf is large,
Cuijg and of Ceijim are not generally determined.
ignoring the effect,s of the interaction among inclusions
However, as shown in the appendix, the contribution
and the presence of the free boundary, as was done
from each term can be calculated, if the specimen is of
ellipsoidal shape. Especially when the specimen is previously,@) may lead to a significant error.
As already mentioned, Hart clarified,t5) in the case
similar in shape to the inclusions, (u~~);,~is solely due
of n-ork hardening of dispersion hardened materials,
to LT,,.
that a. hardening rate derived from consideration of
Bs Brown mentioned,) when it moves in the matenergy balancet2) is essentially equivalent to that
rix as a straight line, a dislocation feds, as a whole,
only t,he average stress-(oij)4,f, since the average of the derived from consideration of the average internal
stress in the matrix.cxv3 If equat,ion (12) is used for
locally fluctuating stress in the matrix is zero. In
such a case. an extra applied stress equal to - (u,~)~~ an elastic energy calculation, exact agreement of the
ia needed to move a dislocation from the case where hardening rates oalculat,ed by the two approaches is
verified.
there is no internal stress. If &ijTis regarded as proACKNOWLEDGEMENTS
duced by plastic deforma~,ion, t,he above hardening is
We
appreciate
discussion with Professor T. Mura of
work hardening which Brown and Stobbs identified
Northwestern
Universit,y
and the encouragement, he
as due to the image stre#)
and to which energy
gave
us.
Dr.
L.
M.
Brown
not only supplied us with a
balance consideration was applied.f2) The equivalency
copy
of
his
paper
before
publication but also corn.
of both treat,ment>shas been discussed by Hart.t5)
mented
on
our
&udy,
and
we greatly appreciat,e his
From the present understanding of internal st.ress,
kindness.
Harts treatment of work hardening of dispersion
REFERENCES
hardened materiaW5) is identified as that which
1. L. M. BROWN, Private communication (to be publidwd in
emphasizes and estimates the role of the locally
Acta Met.).
0. K. TAKAKA and T. MORI, Acta Net. X8, 931 (1970).
fluctuating stress. As treated by Hart, a dislocation
3. L. M. BROWN
rend W. M. STOBBS, Phil. Mw. eS, 1185
feels obstruct,ion from t,he locally fluctuating stress
(1971).

ACTA

574

METALLURGICA,

4. T. B&IRA,&tat.Conf.
Me&.
Pmpetiies
of Materials
at K.yoto,
Vol. 2. (1971).
5. E. W. HART,
Aeta Met. 80,275
(1972).
6. J. D. ESEIELBY, Proc. R. Sot. A!%& 376 (1957).
7. J. D. ESHELBY, J. appl.
Phys.
25,255 (1954).
8. K. TANAKA and T. MORI, J. Blmticity.
To be published.
9. Ii. TANAKA and T. MORI, Phil. Mag. 25,737
(1971).
10. N. KINOSRITA and T. MURA, Whys.
Status Solidi. 5, 759
(1971).
APPENDIX
AVERAGE
STRESS
AND IMAGE
STRESS
IN A
UNIFORM
ELLIPSOIDAL
BODY CONTAINING
ELLIPSOIDAL
INCLUSIONS
WITH
TRANSFORMATION
STRAIN

Suppose that an ellipsoidal body, VO, contains an

ellipsoidal inclusion I with transformation
strain
~~~~~Elastic constants Cijkl are uniform throughout
the body. Let dijDobe stress, assuming the inclusion
to be in an infinite body, and let rrijim be image stress
to reflect the presence of the boundary IV,,l in the
actual body. The total internal stress in the actual
body, oij, is uijm + aijim. Since for static equilibrium,
s I0

crij dD =

tir r-0

(dijm + CT~;~)dD = 0,

qjin dD = - jjj.rF-o

dD -s,_pj

=dD.

(Al)
As u, m in V is uniform,@ the first term in equation
(Al) is expressed as

VOL.

21,

1973

Equations (A3) and (A4) do not depend on the position

of the inclusion. From equations (Al) and (A3),
UiiimdD = -VCijKI[Xktllnn(V~Vo)EmnT- &klT]. (A5)
s v0
Consequently, the average of the image stress in the
body, V,, is calculated as
(air>r-a= -(vlvotc,j~,Es*,,,(v~cl,)e,,T

&*?I*

Equations (A5) and (A6) are independent of the shape

and the position of the inclusion. What is used in bhe
above derivation is uniformity of sbress and straiu
within an ellipsoidal inclusion with uniform transformation strain when the inclusion is in an infinite
body. This uniformity is assured in an anisot,ropic
case.(rj Thus, the above conclusions are valid when
the body is anisotropic.
Finally, the average stress in the matrix will be
considered when the body has many inclusions, the
number of which is N, all of which are identical in
shape and volume. When the body is similar in shape
to the inclusions, the result of equation (A4) becomes
zero and the average stress in the matrix becomes
identical to the sum of the average image stress
given by equation (A5). However, when the body has
a different shape, the average stress in the matrix is,
from equations (A4) and (Ati),
(ofj>Jf = -(Nv/v*)C;.jkl[Sklmn(VO)EmnT -

where &,,,(I)
are Eshelbys tensors defined for the
shape of the inclusion. Reference (8) show that the
second term in equation (Al) is independent of the
size of the body V, and of the position of the inclusion
and is given by

(A31
- 4&nn(v)l~?nnTt
where ~~~~~(Va) are Eshelbys tensors appropriate
for the shape of the body I,. Thus, the average, in
the matrix, of clijac of a single inclusion is written as
laii4).lI

(aij>~o-v

{v/(vCl
[4c,,,CVo,

~mAWmnT.

(NV~~~)( I -

x 1h%n(VcJ

&klTl

V/VJCii,,

%%mAwmnT.

Here, NV/ V0 is equal tof, the volume fraction of the

inclusion, and is finite. Since N is a large number,
V/V, must be negligibly small compared to unity,
Consequently, the above value becomes
((3ij)izI

-~c~j*~[~*~~~(v)&~~T

&*lTI*

In terms of the stress, defined in equation (I I), which

is theinternal stress within a single inclusion in an
infinitely extended body, (ni,) is written as
Xaij)+lI = -fiflm,

V)}Cijkl
(W

(A6)

which is, of course, ident8ical to equation (10).