Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

J. Mech. Phys. Sotidr Vol.27, pp.

315-330
lbrgamon PressLtd. 1979.Printedin Great Britain

SOLUTIONS FOR EFFECTIVE SHEAR PROPERTIES IN


THREE PHASE SPHERE AND CYLINDER MODELS

R. M. CHRISTENSEN
Chcmiitry and Materials Science Department,
Lawnmce LiWrmorc Laboratory,
University of California, P.O. Box 808, Livermore, CA 94550, U.S.A.

and K. N. Lo
Shell Development Company, P.O. Box 1380, Houston, TX 77081, US&

(Received9 April 1979)

SOLUTIONS arc presented for the effective shear modulus of two types of composite material models. The
first type is that of a macroscopically isotropic composite medium containing spherical inclusions. The
corresponding model employed is that involving three phases: the spherical inch&m, a spherical annulus
of matrix material and an outer region of quivaknt homogeneous material of unlimited extent. *
corresponding tw~~~~onal, polar model is used to represent a transvcrsdy isotropic, fiber reinforced
medium. In the latter case only the transverse &ectivc shear modulus is obtainad. The rdativc volumes of
ths inclusionphaseto the matrixannt&s phase in the three phase models am taken to be the given vokrme
fractions of the inclusion phases in the composite materials at large. The msults am found to di&r from
those of the well-known Kcmer and Hermans formulae for the same models. The fatter works are now
understood to violate a continuity condition at the matrix to quivalqt homofgmamts medium interface.
The present results arc compared extqsivcly with results from other related models. Conditions of linear
elasticity arc assumed.

NOMENCLATURE

spherical polar coordinates


polar coordinates
normal and shear stress contponents, respectively
surface traction components
displacerncnt ~~~~~
elastic properties: Lam& con&ant, shear modulus, Poisson’s ratio and bulk
modulus, respectiveIy
as subscripts refer, respectively, to equivalent homogeneous phase, matrix phase,
spherical inclusion phase, and fiber phase
strain energy
volume fraction of spherical inclusion or fiber phase
radius at spherical inclusion or fiber to matrix interface
radius at matrix to equivalent homogeneous medium interface.

1. IN~oDUCH~N
ONE OF THE most basic problems in composite-materials theory is the prediction of
effective or average macro~opic properties in terms of the properties and relative
315
316 R. M. CHR~~~NSEN
and K. Ii. Lo

amounts of the individual materials or phases. While such predictions are difficult to
obtain, they are very useful in assessing the performance that can be expected from
homesite-mate~al fo~ulations. The present work presents two exact (elasticity-
theory) solutions for such effective properties, one for the effective stiffness of a
material containing spherical inclusions and one for the effective stiffness of a material
with cylindrical (fiber) inclusions.
The complications involved in obtaining soiutions are evident when one considers
the typical geometry of the phases. For example, for a manufactured system of aligned
fibers with a common diameter size, the positions of the fibers-in a given case-are
never known, and also vary from one case to another. Since the fibers do not occupy
periodic positions in space, the resulting problems for determining the effective
properties would be extremely complicated. For this reason geometric idealizations
are often employed. Two of the most common idealized geometric models are the
composite-spheres and com~~te~linders models, devised and employed by
Professor 2. Hashin and his associates (e.g. HAZX~N and ROSEN,1964). Intere~in~y,
even for these models not all the exact solutions for effective properties are available.
The composite-spheres model consists of a continuous phase containing a
particular size distribution of spherical inclusions such that the resulting distribution
of composite spheres forms a vohtme-filling configuration. The solution for the
ef%ctive bulk modulus can easily be obtained, but the exact solution for the effective
shear modulus has not been found, although there is considerable information
available on its bounds.
Previous workers have, purportedly, presented exact shear-modulus results for
other models. Specifically, KERNER(1956) presented such a solution, but an
examination of his work reveals it to be brief and sketchy. We are thus sympathetic
with the s~~rn~t of HASHIN(1970, p. 222) concerning Kern&s work : “The writer
has unfortunately not been able to understand this paper”. However, it is not difficult
to discern the existence of an error in Kemer’s derivation. This matter will be
discussed in Section 5. An entirely similar situation exists in the case of fibers, where a
result corresponding to that of Kerner’s for spheres has been deduced by Hermans. In
this result an unsubstantiated assumption also can be pinpointed, as discussed briefly
later in this section and in Section 5.
At this point, it is appropriate to mention the information available on the bounds
of the shear moduli. Care must be taken to distinguish between the two different
“bounds approaches’*, both of which are applicable to the composite-spheres model.
HASHIN(1962) has given bounds on the effective shear modulus as determined directly
from the composite-spheres model and the dassical minimum theorems. A more
general bounds approach applicabIe to any rna~o~opi~~y isotropic composite
material has been derived by HASHIN and SHTRIKMAN(1963) and rederived and
generalized by WALKBLE(E&56). The most restrictive combination of these bounds is
that obtained by taking the upper bound from the composite-spheres model and the
lower bound from the more general approach. Our solutions for effective shear
moduli are found to lie within these most restrictive bounds. Interestingly, KERNER’S
(1956, equation (8)) formula, obtained by uncertain means, coincides with the lower
bound from the general approach.
SMIIM (1974, 1975) has derived results that correspond to ours for the spherical-
model case. However, his work uses an unjustified assumption that ours does not.
Effective shear properties of composite material models 317

This assumption is detailed at the end of our analysis of the spherical model, Section
3, and further discussed in Section 5.
A similar situation exists with regard to fiber-reinforced media. In the composite-
cylinders model there are five effective elastic properties to be obtained, consistent
with the transversely isotropic symmetry property of the composite media. HASHIN
and ROSEN (1964) and HILL (1964) gave exact solutions for four of the five properties.
The fifth property, the transverse shear modulus, is often referenced to have the exact
solution given by HERMANS (1967). However, an examination of the latter reveals it to
be based on an assumption that our work invalidates ; namely that in determining the
effective transverse shear modulus it is incorrectly assumed that the state of stress in
the fiber phase is uniform shear.
Bounds for the transverse shear modulus that are appropriate to the composite-
cylinders model were given by HASHIN and ROSEN (1964). Later, HASHIN (1965)
obtained bounds for the transverse shear modulus with no restriction on the
geometry of the composite-cylinders model. The solution given here for the transverse
shear modulus is consistent with these bounding results. Our result lies within but
does not coincide with any of the bounds.

2. THE MODEL
To calculate the effective properties, we must have a geometric model of the
composite material. Interest here centers around macroscopically homogeneous,
isotropic and transversely isotropic materials. The model for which we determine the
effective properties is that of KERNER (1956). The model consists of the single
composite sphere embedded in the infinite medium of unknown effective properties
(Fig. 1). It will be required that the effective homogeneous medium possesses the
same average conditions of stress and strain as does the spherical model of Fig. 1. The
procedure here will be applied to the case of the shear property. Again, it is important

jgg sRoIIcv,&g--L

I MATRIX

tm EounuuNl
M&+y--
FIG. 1. Model for solution.
318 R. M. CHRBTENSEN and K. H. Lo

to observe that we are solving the same problem as that posed by KERNER(1956), and
we should, of course, expect to recover his results. However, we do not get the same
final forms. The similar model will be solved in the case of fiber reinforcement. Our
results will be found to differ from those of HERMANS(1967). The relationship of the
present models to the composite-spheres and cylinders models will be discussed in
Section 5.

3. SPHERICAL-MODEL
EFFEC~VESHEARMODULUS
Introduce a spherical coordinate system with the origin at the center of the
composite sphere shown in Fig. 1. The ratio of radii in Fig. 1 is taken such that
u3/b3 = c, the volume fraction of the inclusion phase. Let the displacement u have the
form
u, = U, sin2 0 cos 24,
uo= U,sin8wsews2& (3.1)
u+ = U, sin e sin 24, 1
with
U, = - U,, (3.2)
where U,, U, and U, are functions of r only and r, 8 and 4 are the usual spherical
polar coordinates. The form in (3.1) will be shown to be consistent with the basic state
of shear deformation, which is of interest here. As seen in Fig. 1, there are three
separate regions in which the solutions must be obtained. The general forms of the
elasticity solutions of the equations of equilibrium (from Love (1927)) appropriate to
the three separate regions of Fig. 1 are given by, (i)
(5 - 4v) D*
u, = D,r+F+--
(1-2~) r2’
r B b, (3.3)
U,=D,r-F+q,

in the equivalent homogeneous medium, (ii)

km
U, = B,r- B2r3+F + iI>))Bz,
(1-2&n) m
a,<r<b, (3.4)
(7- 4%) B r3 -- 283 + 28,
U, = B,r-
(1-2v,) 2 P r2
in the matrix phase, and (iii)
6v.
U, = A,r- (1 A2r3,
OGr<a, (3.5)
Uoi = A,r-w A2r3,
- v,
in the inclusion phase, where v, vi and v, designate values of Poisson’s ratio. In these
equations, D,, D,, D4, B,, B,, B,, B4, AI and A, are constants to be determined from
Effective shear propertiesof composite material models 319

the continuity conditions at the two interfaces. These conditions involve the
continuity of the stresses CT,,TVand t&, and the displacements u,, ue and u,+.Only eight
conditions result from the initial twelve continuity equations due to the redundancy
of many of the forms. The continuity conditions give the following equations:

6~. 6~ (5-4&I) B* I
Ala- (1 AS3 = Bla- (l_2v,) (1-2v,) az’ (displacement

1~
at r = a),?

1,
2(5 - 4v,) B4 (stress at

(l-2&) as I = a),

(7+2v,)
Pi AI - (1 A2c2
1
E
B4-
I,
(7+2v,) 2(1 +v,)
883+
= L BI - t1 _2vm) B2a2+ a3
(l-Z&) T_

Brb-
6vm
(1 _2v,) B2b3+
388+
b4
(5-4v,) B4
(l-2&) F
,(displacement
=D,b+E$+~$, at r = b),
-v

Bib-
('-4VdB b3 -- 2B3+szD b -- m3+m4
(1 - 2v,) ’ b4 b2 1 b4 bZ

3&, [‘B,b’-
91
1
18v, 2(5 - 4v,) B4
_--1283
+2k B1- (1 _2v,) B2b2 b5 (l-2&) bJ

1’
2(5 - 4v) D4 (stress at
(l-2v) bJ r = b).

2(1 +v) D4
(13 1 I
t
(3.61)
Here, and elsewhere, remarks in parentheses refer to continuity conditions.
320 R. M. CHRISTENSENand K. H. Lo

Note that although there are nine constants, only eight of them are basic unknowns.
The constant D1 is specified directly by the condition of uniform shear strain at
infinite distance from the origin. Properties (&,pi, vi) and (4,,,~,,,, v,) are,
respectively, the Lame, shear, and Poisson’s ratio properties for the spherical
inclusions and the matrix phase, and (J., p, v) are the unknown effective properties of
the equivalent homogeneous medium.
To determine the effective shear modulus ~1we make use of a basic result obtained
by ESHELBY (1956, equation (5.1)): namely that, for a homogeneous medium
containing an inclusion, the strain energy U, under applied displacement conditions,
is determined by

where S is the surface of the inclusion, U,-,is the strain energy in the same medium
when it contains no inclusion, p and up are the tractions and displacements in the
medium when it contains no inclusion and q, and uie are the corresponding quantities
at the same point in the medium when it does contain the inclusion.
The energy term U in (3.7) represents the total strain energy stored in the model of
Fig. 1. The objective is to solve for the stiffness properties of an equivalent
homogeneous medium, such that under the same applied displacement conditions
this equivalent homogeneous medium stores the same energy U as the configuration
of Fig. 1. Thus, the problem, for which we present a rigorous solution, is that of
obtaining the relationship between the configuration shown in Fig. 1 and the
equivalent homogeneous medium.
The next step in the procedure, then, is to write the criteria for the energy equality
of the model of Fig. 1 and the equivalent homogeneous medium. As already stated,
the energy of the Fig. 1 model is just Cr. The energy of the equivalent homogeneous
medium, UesuiV,is clearly the same asthe energy of the model in Fig. 1 if there is no
inclusion. But from Eshelby’s derivation and the corresponding definition of terms
stated with (3.7), this energy is just given by U,, the energy of the medium with no
inclusion, and thus U = U,. Combining the basic energy-equivalence statement
U&v = U with the id%& UcqtiV= U,, gives the following energy statement to be
satisfied :

u= u,. (3.8)

Combining (3.7) and (3.8) finally gives the form

i [~~uU,+T.S)gUg,+T~U~~-~~US)-T~g.ugO-T~,U~]
dS = 0, (3.9)
with
dS = b2 sin 0 d4 de. (3.10)
Effective shear properties of composite material models 321

Now, the present problem having conditions of simple shear at infinity, it follows
that
0,” = 2& sin’ 8 cos 24,
T$ = 2~0, sin 8 cos 8 cos 24,
ror&= - 2@, sin 8 sin 24,

up = D1rsin’ 8 cos 24,


~8”= D,rsin 8 cos 0 cos 24,

US= - D1 r sin 8 sin 24,


and

I (3.12)

Substitution of (3.1), (3.2), (3.10), (3.11) and (3.12) into (3.9) gives the simple result
D, = 0. (3.13)
The problem now reduces to solving for D4 from (3.6) and setting it equal to zero
in accordance with (3.13). This gives the condition for determining the effective shear
modulus p:
($+(&)B+D=O, (3.14)
where
A = 8 [~i/lu,- 1](4_5V,)rl, c”‘~ - 2[63(ki/lc,-
1)ttz + 2rl, t/j]c”’
+252~i/C(m-l]~2C5’3-25~i/~,-l](7-12V”+8V~)r12C
+4(7- lOV,)t/Zrtj, (3.15)
B= -41jl~/~-1](1-5~~~~~‘~‘~+4[63(~~/~~-l)rt~+2~r~~]~“~
- 504G4/Pm- I]t/2C5”+ lSO~i/&- l](S--V,)V,?f2C
(3.16)
and
D = 4[~i/~~-l](5V,-7)rl~C’o’3-2[63(~~/~,-l)tt,+2~,~3]~7’3
+ 252 Cd&n - l]~zCs’3+25[~i/~,-I](V~-7h,C

-(7+%h~~, (3.17)
with
‘11 = IMP, - l](49-50Vrv,)+35(&//&)(Vi-2V~)+35(2Vi-V,),
V2 = 5viCPi/&n- 8l+IA+Pm+41, (3.18)

~3 = W~M8- lOvm)+
(7-5Vrn)v
and with c being the volume fraction, viz. c = (c1/6)~.
322 R. M. CHRBTENSEN
and K. H. Lo
Relation (3.14) is the final form sought. Using (3.14) and the quadratic formula,
one cm determine the exact solution for the equivalent shear modulus of the spherical
model. It is of interest to extract the dilute-suspension result from (3.14). This is done
directly by letting the volume fraction c be small and using the binomial expansion
in (3.14). It is found that

Ir: l-, W -4X1 -P&l&


--“zz +O(cZ) for c + 0. (3.19)
Pm 7- 5v,+2(4- w(P~/cc,)
This is the same dilute-suspension result as deduced by ESHELBY (1957, p. 390)
It should be mentioned that the proof we give of the coefficient D4 vanishing (see
(3.13)) corresponds exactly to the assumption made by Sharon (1974, 1975) at the
outset in his analysis of the same problem. However, in the context of his work, the
~sumpti5~ t>s = 0 had no ~~ti~~a~un whatsoever. This situation will be discussed
further in Section 5. The final results 5f his and our separate analyses are identical.

4. CYLINDRICAL-MODEL TRANSVBRSESHEAR
EFFECI?M! MODULUS

A two-dimensional idealization will now be taken, corresponding to the preceding


three-dimensional model. The model is thus applicable to an aligned fiber system
which possesses transverse isotropy, with five independent effective properties. Using
the composite-cylinders model, HASHIN and ROSEN(1964) have determined four of the
five properties. The fifth property, the transverse shear modulus, poses a more
difficult prebiem, and Hasin and Rosen only obtained bounds on it. Tfie probiem
was further discussed by HA~EEIN (1972).
The effective trasverse shear modulus for a fiber system will be obtained here by
using the same three phase model just considered, Specifically, take the model shown
in Fig. 1 as having polar symmetry. The solutions of the equation of equilibrium
(from SAVIN (1961)) have the forms, (i)

(!I
b
2r b b”
tl, = - - +(tl+l)a, ; +e, ; cos 28,
4~ C b
r a b, (4.1)
sin 28,

in the equivalent homogeneous medium, (ii)


b
%n= +czOlm+
Gi
+b2 cos 28,

b b
4hn -C2hll- 1’;
bJ
-f-b2 sin 28,
r
(_-)I
EE=tive &ear properties of coqx3sitematerial models 323

in the matrix phase, and (iii)

in: the fiber phase, where a3, c3, a2, d2, c2, b2, al, and d, are constants to be
determined, and
? = f3-4vf, qn = f3-4v,), F&= (3 -4&f. (4.4)
The appropriate subscripts designate the fiber and znatrixphases,whitequantities
with no subscript are those of the equivalent homogeneous medium.
The constants are determined from the conditions of continuity of b,, TV, u, and ug
across the two interfaces in Fig. 2, The resulting eight equations are

(stress at
t = a),
az~ttm+3)-~c,(~~++)l-c, i 4111X9m- ll-Pln(ttr- 111 (4.5)
0

@b-&J+b2 f
0L+)lrCm)
= 0

,(stress at
b).
r 15$.

(1956, equation (5.1)) result, equation (3.7), is used to


As in Section 3, ESHELBY’S
obtain the requirement that

(4.61
324 R. M. CHRISTENSEN
and K. H. Lo

23f
/
- Pmwnt solution

21 -
19 i
--- Lower bound arbitmry phase gdomstry
Hashin (1965)

15 - Hashin (19721

E 13-
i
= ll-

S-

7-

6-

FIG. 2. Efktive shear modulus for spkical model, with .u&,, = 2346, v, = 022 and v, = 0.35
(glass/epoxy composite).

where
a,” = cos 28, z; = - sin 28, (4.7)
-r
0
=Los20, u$ =E - sin 28, (4.8)
uF 2/l 2P
and

14.9)

The notation in (4.6) through (4.8) is the same as used in Section 3.


Substitution of relations (4.1), (4.7) and (4.9) into (4.6) yields the simple result
a3 = 0 (4.10)
(cf. (3.13)).
If we proceed in the same manner as in Section 3, the eight equations (4.5) are solved
for a,, which is then substituted into (4.10). After a lengthy process of algebraic
reduction, the governing equation for the effective transverse shear modulus ~1is
found to be
Effective shear properties of composite material models 325

where

D = 3c(l -c)~CMP,- 1lC~/~n,+~rl


+ CPf/Pnlhnl
+ WP, - 1k + 1X/4/P,+ tlf + Olrhlnhnl- Vf)c31, (4.12)
B = -6c(l -~)~[jlr/~,- lIG&,,+rlrl
+C~/~,)tl,+(~f/~,-lI+llC(rt,-l)(~f/~,+ttf)-2~3~/~,)tf,-~f)l
+ (r7m+ 1)cCPf/Pnl
- llCPf/Pnl+)Ir+ (Pf//.W?m-)lr)C31~ (4.13)

and
A = 3c(l- c)~CMP,,,- llC&, + qrl
+ r.Pf/WLll
+ rlfrlm- (./4/Pn& - rtfk31
xC~,c~/~,--1)-_/~,tf,+111, (4.14)

with c being the volume fraction of the fiber phase, viz. c = (~/b)~.
Before turning to the discussion, it is relevant to note that in the dilute case
(c 4 1 ), (4.11) reduces to the form

_=
P l+bf-CcmMl+hl) c= 1+
C
+O(c2) for c + 0,(4.15)
P” oLt%n+P,) P* km+ i&n
Pf-c(nl + 2L + b,

where k, is the bulk modulus of the matrix phase.

5. DBCUSION

Relations (3.14) and (4.11) provide the solutions for the effective shear moduli of
the spherical model and the cylindrical model of Fig. 1, the latter being the transverse
shear modulus.
It is of interest to compare the present results with other theoretical predictions.
Consider first the spherical model. Figures 2 and 3 show results for systems having
two different ratios of inclusion shear modulus to matrix shear modulus. Also shown
for comparison are the upper bound derived by HASHIN(1962, equation (53)) for the
composite-spheres model and the lower bound predicted by HASHINand SHTRIKMAN
(1963, equation (4.3)) and WALP~LE (1966, equation (28)) for general, or arbitrary,
phase geometry. Only these two bounds are shown because they are the most
restrictive combination of all four possible bounds for this model. KERNER’S(1956,
equation (8)) formula coincides with the lower bounds shown in Figs 2 and 3. The
main observation to be made is that none of the bounds coincide with the result
derived herein. Clearly, Kerner’s result cannot be exact, as claimed.
326 R. M. GWXEN.SEN and K. H. Lo

25

23

21 -
- Pmmt solution
19 - ---- Lomr boundvbitary phm
geometryHmhin& Shuikmn
17 - (19531:Wdpoka(19961

15 -
--- &qnrbound
CMpooi~ m&
HathinflW2)
2 13-
i
11

FIG. 3. Efktive shear modulus for spherical model with M,/& = 135.14, v, = 0.20 and v,,, = 0.35
(boron/epoxy composite).

FIG. 4, Effective transverse shear modulus for cylindrical modeI with k/k = 23.46, vf = 0.22 and v, = 0.35
(glass/epoxy composite).
Effective shear properties of composite material models 327

23

21

19

17

FIG. 5. Effectivetransverse shear modulus for cylindrical model with k/p,,, = 135.14, v, = 0.20 and
v, = O-35(boron/epoxy composite).

For some purposes the lower-bound prediction can be used as a reasonable


approximation to the solution. From Figs 2 and 3, as long as the volume
~n~ntra~on of inclusions c does not exceed 05, this replacement would seem to be
reasonable and would be advantageous, since the lower bound formula given by
WALF+OLE(1966, equation (28)) is vastly simpler than the exact formula obtained from
(3.20). However, such replacements must be made with caution, i.e. they are not
generally valid.
Figures 4 and 5 show the transverse shear modulus results for the fiber ease and
glass/epoxy and boron/epoxy systems. Also shown are the upper bound derived by
HASNIN (1972, equation 3.4.113) for the composite-cylinders model and the arbitrary-
phase-geometry lower bound derived by HASHIN (1965, equation (4.27)). Again,
observe that none of the bounds coincides with the result presented here. It is also
pertinent to note that in some ranges of properties and volume fraction the lower
bounds shown in Figs 4 and 5 may be reasonable approximations to our results.
The lower bounds displayed in Figs 4 and 5 coincide with the predictions of
HERMANS(1967, equation (26)) for fiber-reinforazd media. The approach of Hermans
for the fiber case is the counterpart of that by Kerner for spherical inclusions. The
basic model used by Hermans was that of Fig. 1. However, Hermans’ analysis is not
exact; he assumed that the state of strain in the fiber phase is that of uniform shear,
and as a consequence he violated continuity conditions at the matrix/equivalent-
medium interface. As already noted, the deficiency in Kerner’s and Hermans’ work
was observed and discussed by Hashin and later by BEHRENS(1971). Our present
work satisfies all the continuity conditions.
It is worth mentioning the comparison of our work with the predictions of the
“self-consistent” model which involves embedding the spherical or cylindrical
inclusion directly in a homogeneous medium that has the effective properties to be
328 R. M. CHRISTENSEN
and K. H. Lo

determined. As noted by HILL (1965a, b) and BUDIANSKY (1965), there are physical
inconsistencies in the predictions of this model and it does not admit the limiting case
c = 1. The present models can easily be shown to admit the proper behavior in the
limiting case c = 1. We have already noted that the present solutions yield, as special
cases, the correct forms for dilute suspensions.
A further check on the consistency of our method can be deduced as follows. With
regard to the composite-spheres model, we found by using Eshelby’s formula (3.7)
that the constant D., in (3.13) must vanish. An examination of the displacement forms
(3.3) for the equivalent homogeneous phase reveals that were D4 not required to
vanish, then the resulting solution would have involved not only the effective shear
modulus p but also the effective Poisson’s ratio v. A similar situation exists with
respect to stresses in the effective-medium phase: D4 must vanish to eliminate a
dependence on a second property. Thus, had D, not been required to vanish, the
method would not have given p independently of the second effective property. In the
case of the fiber solution, an entirely similar situation arose whereby the vanishing of
a3 in (4.10) eliminates the effect of a second effective property in the displacement
form (4.1).
In the case of the spherical model, it is helpful to finalize the relationship of the
present method relative to that of SMITH(1974, 1975). In the present approach it was
proved that constant D4 must vanish. In SMITH’S(1975, p. 419) approach the basic
assumption upon which the work is based is stated as : “If the same boundary stresses
are applied to each of the spheres, it is assumed that the displacements in the two
spheres will be the same at a distance r ti 1, except for terms of a high order in l/r”.
The term “high order” is arbitrarily defined by SMITH (l/r)3. Thus, since the
displacement term involving D, is of the order (l/r)‘, it follows from Smith’s basic
assumption that D, = 0. No such assumption is involved in the present work and it is
independently proved that D4 = 0.
Finally, it is useful to clarify the status of the present work relative to that of the
composite-spheres and cylinders models. The composite-spheres model is as shown
in Fig. 6. It involves various sizes of spherical inclusions, such that a volume-filling

FIG. 6. Composite-spheres/composite-cylinders model.


Effective shear properties of composite material models 329

configuration is obtained with each individual composite sphere having the same
ratio of radii, u/b. In determining the properties of the equivalent homogeneous
medium, the criterion to be used is that the repeating cells be replaced by the
equivalent homogeneous material without changing the conditions of average stress
and average strain. In considering an infinite medium of the composite-spheres
model, one could replace all but one of the cells by the equivalent homogeneous
medium, to arrive at the model used herein, viz. Fig. 1. While we cannot prove that
the mode1 of Fig. 1 is rigorously equivalent to the composite-spheres mode1 of Fig. 6,
they are known to give the same result in the case of dilatation. This entire procedure
was extensively studied and discussed by HASHIN (1968) in the thermal/electrical
conductivity context. However, as it stands, all that can be said with certainty is that
the solutions found here are rigorous results for the mode1 of Fig. 1, and it only may
be conjectured that they have application to the model of Fig. 6.

ACKNOWLEDGEMENT

This work was performed under the auspices of the U.S. Department of Energy by the
Lawrence Livermore Laboratory under Contract No. W-7405ENG48. This work was also
supported by a contract from the U.S. Army Research Office.

REFERENCES

BEH~~NS,E. 1971 Trans. ASiUE 93, Ser. E, J. appl. Mech. 38, 1062.
BUDUNSKY, B. 1965 J. Mech. Phys. Solids 13,223.
ESHELBY,J. D. 1956 Prog. Sol. St. Phys. (edited by F. Seitz and
D. Turnbull) 3, pp. 79-144. Academic Press,
New York.
1957 Proc. R. Sot. A 241, 376.
HASHIN. Z. 1962 Trans. ASME 84, Ser. E, J. appl. Mech. W, 143.
1965 J. Mech. Phys. Solids 13, 119.
1968 J. Comp. Mater. 2, 284.
1970 Mechanics of Composite Materials (Proceedings of
the Fifth Symposium on Naval Structural
Mechanics, Philadelphia, May 1%7) (edited
by F. W. Wendt, H. Liebowitz and N.
Perrone), pp. 201-224. Pergamon Press,
Oxford.
1972 Theory of Fiber Reinforced Materials,
NASA CR 1974.
HASHIN, Z. and ROSEN,W. 1964 Trans. ASME, 86, Ser. E, J. appl. Mech. 31, 223.
HASHIN,Z. and SHTRIKMAN, S. 1963 J. Mech. Phys. Soli& 11, 127.
HERMANS,J. J. 1967 Proc. K. Ned. Akad. Wet. B 70, 1.
HILL, R. 1964 J. Mech. Phys. Solids 12, 199.
1965a Ibid. 13, 189.
1965b Ibid. 13, 213.
KERNER, E. H. 1956 Proc. Phys. Sot. Land. 69B, 808.
LOVE, A. E. H. 1927 A Treatise on the Mathematical Theory of
Elasticity (4th edn), p. 141. Cambridge
University Press. (Also: 1944 Dover, New
York.)
330 R. M. CHRISTENSEN
and K. H. Lo

SAWN, G. N. 1961 Stress Concentrations Around Holes, p. 234.


Pergamon Press, Oxford.
SMITH, J. C. 1974 J. Res. Nat. Bur. Stand. 78A, 355.
1975 Ibid. 79A, 419.
WALPOLE, L. J. 1966 J. Mech. Phys. Solids 14, 151.

You might also like