Vous êtes sur la page 1sur 10

Acta metall, mater. Vol. 39, No. 7, pp. 1617-1626, 1991 0956-7151/91 $3.00+ 0.

00
Printed in Great Britain. All rights reserved Copyright © 1991 PergamonPress plc

FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS


J. S. H U A N G and L. J. GIBSON
Department of Civil Engineering, Massachusetts Institute of Technology,
Cambridge, MA 02139, U.S.A.

(Received 8 October 1990)

Abstract--Existing models for the fracture toughness of brittle cellular materials are based on the
assumptions that the crack is large relative to the cell size and that the modulus of rupture of the cell
wall material is constant: both may, in practice, be invalid. Here, we account for the effect of short cracks
by using a finite element analysis and for that of the variability in the cell wall modulus of rupture by
describing it by a Weibull distribution. The results of the analysis indicate that if the crack half length/cell
size is less than 7 the fracture toughness is reduced and that the way in which the fracture toughness varies
with cell size depends on the Weibull modulus of the cell wall material.

R6smn6---Les mod61es existant pour la t6nacit6 de rupture des mat6riaux cellulaires fragiles sont fond6s
sur les hypoth6ses suivantes: la fissure est grande devant la taille de la cellule et le module de rupture de
la paroi cellulaire du mat6riau est constant. Les deux peuvent, en pratique, ne pas ~tre valables. Ici nous
rendons compete de l'effet de courtes fissures en utilisant une analyse par 616ments finis et de l'effet de
la variabilit6 du module de la paroi cellulaire en le d6crivant par une distribution de Weibull. Les r6sultats
de cette analyse indiquent que si le rapport de la demi-longueur de la fissure ~i la mille de la cellule est
inf6rieur ~ 7, la tenacit6 de rupture est r6duite et que la fafon dont la t6nacit6 de rupture varie avec la
taille de la cellule d6pend du module de Weibull de la paroi cellulaire du mat6riau.

Zusammenfassung--Die vorhandenen Modelle der Bruchz~ihigkeit spr6der zellul/irer Materialien beruhen


auf den Annahmen, dab der RiB gro6 im Vergleich zur Zellgr6Be ist und dab der Bruchmodul des
Zellwandmaterials konstant ist. Beide Annahmen k6nnen aber in der Praxis ungiiltig sein. Hier analysieren
wir den Einflul3 kurzer Risse mittels der Methode der Elemente und den Einflug eines variierenden
Bruchmoduls der Zellw/inde mittels einer beschreibenden Weibull-Verteilung. Die Ergebnisse der Analyse
weisen darauf hin, dab die Bruchz/ihigkeit verringert wird, solange das Verh/iltnis von Rig-Halbl~.nge zur
Zellgr6Be kleiner als 7 ist. Die Art der Variation der Bruchzfihigkeit mit der Zellgr6ge hfingt von dem
Veibull-Modul des Zellwandmaterials ab.

1. INTRODUCTION toughness of brittle honeycombs using finite element


analysis. The results indicate that short cracks reduce
Brittle cellular materials are increasingly being used the fracture toughness; the reduction factor is insen-
in engineering applications. Ceramic honeycombs are sitive to cell geometry. We then reanalyze the fracture
used in catalytic converters in automobile exhaust toughness assuming that the moduli of rupture of the
systems, open-cell ceramic foams are used as filters cell walls follow a Weibull distribution. The results
in metal casting operations and glass foams are used indicate that Ktc increases with cell size if the Weibull
as thermal insulation. In each of these applications modulus, m, is greater than 4; it decreases with cell
the thermal shock resistance of the material, which size if m is less than 4; and is unaffected by cell size
depends on its fracture toughness, is important. if m is equal to 4. The analysis is extended to foams
This paper describes the fracture of brittle honey- in the following companion paper.
combs; the following companion paper, that of brittle
foams.
Previous attempts to model the fracture toughness 2. LITERATURE REVIEW
of brittle honeycombs and foams have applied well-
known results for a c o n t i n u u m by assuming that the Gibson and Ashby [3] derive an expression for
crack length is large relative to the cell size. In the fracture toughness of a brittle honeycomb with
addition, the tensile strength of a single cell wall has hexagonal cells using a c o n t i n u u m approach, assum-
been assumed to be constant [1, 2]. Both assumptions ing that the crack length is large relative to the cell
are, in practice, unrealistic: short cracks, roughly size. Consider the honeycomb shown in Fig. 1 with
three to four cell diameters, are c o m m o n in brittle a crack of length, 2a, under a remote tensile stress, tr,
honeycombs and foams; and brittle cell walls have a normal to the crack surface. The inclined cell edges
Weibull distribution of strengths. In this paper, we have a length, l, the vertical edges, a length, h. The
first evaluate the effect of a short crack on the fracture cell edges have a thickness, t, and the honeycomb has

AM 39/7--p 16 1 7
1618 HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS

This model has assumed that the crack length is


t t t large relative to the cell size and that the modulus of
rupture of the cell wall is constant. In practice, the
crack length is often small relative to the cell size.
~ "~:' ~"'i" " In addition, the cell wall itself has internal cracks
of varying lengths; as a result, its modulus of rupture
is not constant. In this paper, we first use a finite
element analysis to determine the effect of short
cracks on the effective fracture toughness. We then
recalculate the fracture toughness assuming that
the cell wall modulus of rupture follows a Weibull
distribution.

3. FINITE ELEMENT ANALYSIS FOR


SHORT CRACKS

To study the limiting crack size for which the


continuum model for fracture is valid, the frac-
ture toughness given by the continuum model is
Fig. 1. Crack front in brittle honeycomb under uniform first compared with that calculated using a finite
tension [after Gibson and Ashby (1988)]. element analysis of a centrally cracked plate of a
honeycomb-like material with square cells (Fig. 2).
The finite element analysis uses a grid of 50 × 30
a width, b. A singular stress field, trr, develops ahead cells (3080 nodes and 9240 degrees of freedom)
of the crack tip giving a local stress of approximating an infinite array of cells: the stress
distribution along the outermost layer of cells is
(1) uniform and the stress field far away from the crack
tip is undisturbed. The analysis gives the forces and
where r is the distance ahead of the crack tip. moments at both ends of each member from which
The force, F, on the first unbroken cell wall ahead the critical skin stress of the first unbroken cell edge
of the crack tip is ahead of the crack tip can be calculated; failure
occurs when the critical skin stress reaches the modu-
F oZ trrbl lus of rupture of the cell wall which, in this part
F (2) of the study, we assume to be constant. The effect
of three parameters, relative density (t /l), cell size (l)
assuming that the first unbroken cell wall is a distance and semi-crack length (a), on fracture toughness is
proportional to l, ahead of the crack tip. The moment evaluated.
on this cell wall is then
3.1. Relative density
M oc a b x / ~ . (3)
The relative density of a honeycomb (the density
If the honeycomb is of low relative density (t/l < 0.3) of the honeycomb divided by that of the solid from
bending dominates the cell deformation and the which it is made) is given by
critical stress at the first unbroken cell wall is
p* t/l(h/l + 2)
6c~~-~~'
M
p~ 2 c o s 0 ( h / l +sin 0)"

or For honeycombs of the same cell shape (h/l =


constant and 0 = constant), the relative density is
directly proportional to the ratio of the cell wall
,4,
thickness to length, t/l. In this study, five different
ratios of t/l are considered: t/l = 0.035, 0.050, 0.071,
Failure occurs when the critical skin stress reaches the
0.141, and 0.283. In each case the cell size l and the
modulus of rupture of the cell wall, qfs. Hence, the
half-crack length, a, are held constant at l = 1.8 mm
applied stress at failure, af, is
and a = 15.2mm.
3.2. Cell size
The fracture toughness of honeycombs with four
Noting that the fracture toughness, K~c, is equal
different cell sizes (l = 0.9, 1.8, 4.5, 9.0ram) are
to ~f~/n-a and that the relative density, P*/P~, is
evaluated. The relative density, t/l, and the half-
proportional to t/l gives
crack length, a, are held constant at t/l = 0.141 and
K,c = C, at, x / ~ ( P */O~)~. (6) a = 12.7 ram.
HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS 1619

t~

Fig. 2. Loading geometry for finite element analysis of a brittle honeycomb with a central crack of
length 2a.

3.3. Crack length where F, the axial force, and M, the bending moment
Six different crack lengths are used in this study: in the member, are found from the FEM analysis.
a = 2.5, 5.1, 10.2, 12.7, 15.2 and 25.4mm. The rela- Fracture occurs when the internal stress in the
tive density, t/1, and the cell size, l, are held constant cell wall ahead of the crack tip reaches the modulus
at t/l = 0.141 and l = 1.8 mm. of rupture of the cell wall, ~rfs, at an applied stress
The honeycomb geometries and crack lengths ana- of cr = ~f. The fracture toughness is then calculated
as
lyzed are listed in Table I. The fracture toughness
of each honeycomb, assuming that the cell wall Kxc = ~ . (7)
material is linear elastic to fracture and that it has
a constant modulus of rupture, ~fs, of 34.45 MPa, The resulting fracture toughness for each case is listed
is calculated as follows. A known uniform tensile in Table 1 and plotted in Figs 3-5.
stress, tr, is applied to the honeycomb. This generates
3.4. Comparison of F E M results and the continuum
an internal stress in the cell wall ahead of the crack
model
tip of
The continuum model suggests that the fracture
F 6M toughness depends on the square of density and on
ai =-~ + bt----
f the square root of cell size [equations (6)]. Figures 3
1620 H U A N G and G I B S O N : FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS

Table I. Finite element results for the fracture toughness of brittle


100
honeycombs (0 = 45 ° and h/I = O)
\ ----- Continuum model
a / KIc
(mm) (mm) tit all (KN-m -3/2) \ \ o FEM result
Density effect 15.2 1.8 0.035 8.49 4.12 \
15.2 1.8 0.050 8.49 7.5 ~,
15.2 1.8 0.071 8.49 14.2 I::
15.2 1.8 0.141 8.49 45.4 Z
15.2 1.8 0.283 8.49 133 v
v
Cell size effect 12.7 0.9 0.141 14.14 31.9 o
12.7 1.8 0.141 7.07 44.4 ,¢
12.7 4.5 0.141 2.83 57.9
12.7 9.0 0.141 1.41 66.0 t~
(tJ \
Crack length effect 2.5
5.1
10.2
12.7
1.8
1.8
1.8
1.8
0.141
0.141
0.141
0.141
1.41
2.83
5.66
7.07
29.5
36.7
43.1
44.5
e-
r"
O')
"s
O
'• -0.5
\
\
15.2 1.8 0.141 8.49 45.5 \
25.4 1.8 0.141 14.14 45.3 "n

ii

and 4 show that the FEM results for the fracture


toughness of honeycombs confirm with the con-
10 I I I I I I Ill I I I [ i I Ill
tinuum model when the relative density is low
1 10 100
(t/l < 0.2) and the ratio of crack length to cell size
is large enough (a/l > 7). At ratios of t/l higher Semi-crack length / cell size, all (-)
than 0.2, axial stresses in the cell walls, neglected Fig, 4. Dependence o f fracture toughness o n semi-crack
length/cell size (a = 12.7 ram, t/l = 0,141).
in the continuum model described in Section 2,
become significant, decreasing the slope of the
log K ] c - log(t/l) plot below 2. At ratios of a/l < 7, corresponding to two broken cell nodes) the fracture
the fracture toughness is lower than that suggested toughness is two-thirds that for a "long" crack
by the continuum model. This effect is shown (a/l > 7).
more clearly in Fig. 5, for the fracture toughness of The effective fracture toughness for short cracks in
honeycombs with varying crack lengths at constant a brittle honeycomb of one particular cell geometry
t/l and cell size. Again, for a/l greater than 7, has been evaluated. To investigate the effect of cell
the fracture toughness is constant, independent of shape on the effective fracture toughness for short
crack length. But for all less than 7, the fracture cracks we repeated the finite element analysis on
toughness decreases rapidly. For a/l = 1.4 (a crack honeycombs of different cell geometries.

60
1000 --
------ Continuum model
------ Continuum model //
/
a FEM result
[] F E M result /
/ 50
O4 /
/ . . . . .

/ EJ
Ei
z
/p z
~" 40
100
v
2.0~/ (3

/
0m ,,J

d ¢n 30
(n
iI e-
t- e-
.c=
O)
o)

o 10
o 2o

_=
0

U. I " 10
I
I
I
1 I I I i I I II~ I I I i i I I II I I
0.01 0,1 1 10 20
Cell wall thickness I l e n g t h , t/1 (-) S e m i - c r a c k length / cell size, a/I (-)
Fig. 3. Dependence of fracture toughness on the cell watt Fig. 5. Dependence o f fracture toughness on semi-crack
thickness/length ratio (a ffi 15.2 ram, 1 ffi 1.8 ram). length/cell size (I = 1.8 ram, t/l = 0.141).
HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS 1621

Table2. Finiteelementresultsfor the effectivefracturetoughnessof 1.2 -


honeycombsof differentcell geometrieswith short cracks (0 = 45°,
l = 1.8 mm and t/l = 0.141)
Kit
u a// Kic K~c.~.~
(ram) (--) (KN-m- 3a) (__) 1.0
h/l = 0 2.5 1.41 29.5 0.65
5.1 2.83 36.7 0.81 "7"
10.2 5.66 43.1 0.95 ~3
12.7 7.07 44.5 0.98
15.2 8.49 45.5 1.00 ~ 0.8
25.4 14.14 45.3 1.00
t~
h/I = 0.5 2.5 1.41 26.2 0.68
5.1 2.83 31.9 0.83 t-
O
10.2 5.66 36.8 0.96 ~ 0.6
12.7 7.07 37.8 0.99 I o h/If0.5, FEM result
15.2 8.49 38.4 1.00 "m
0)
h/l = 1.0 2.5 1.41 20.8 0.64 tr & hll=l.0, FEM result
5.1 2.83 26.6 0.82
10.2 5.66 30.8 0.95 0.4 ----- Appropriate function
12.7 7.07 31.9 0.98
15.2 8.49 32.6 1.00

o.2 I I I
3.5. Cell shape effect for small all 0 5 10 15

Three different cell geometries are analyzed: Semi-crack length / cell size, a/I (-)
h/l=O and 0 = 4 5 ~ (square cells); h/l =0.5 and
Fig. 7. Reduction factors for honeycombs of different cell
0 = 45 ° (hexagonal cells); and h/l = 1.0 and 0 = 45 ° geometries.
(hexagonal cells). The ratio of a/l varies from 1.41
to 14.14. The cell dimensions and the fracture
Fig. 7, are almost equal; they can be approximated by
toughnesses given by the FEM analysis are listed in
the equation (Fig. 7)
Table 2 and plotted in Fig. 6. The results for all three
cell geometries are similar: the fracture toughness ot = l - exp[-O.8(a/1)°7s]. (8)
is constant for all greater than 7 and drops sharply
In summary, the continuum model describes the
as a/l is reduced below 7. The reduction factor for
dependence of the fracture toughness on relative
the fracture toughness, :t, is given by dividing the
density and cell size well for relatively long cracks
fracture toughness at small all by the constant value
(aft > 7). The effective fracture toughness for short
for large aft. The reduction factors for the three cell
cracks is described well by a modified continuum
geometries analyzed, listed in Table 2 and plotted in
model, incorporating the reduction factor, ~t, found
from the finite element analysis
100 - -
Klc = Ccttrfsx/Ql(t /l): (9)
where C is a constant related to the cell geometry.

E= 4. WEIBULL ANALYSIS FOR VARIABLE


z STRUT STRENGTH
v
v

0
The continuum model also needs to be modified to
take account of the variability in the modulus of
,d rupture of the cell wall material. The Weibull distri-
03
r-
bution is often used to characterize the variability in
e-
tensile strength of brittle materials; here, we assume
--i

o that it can be used to describe the variation in the


modulus of rupture of the cell wall. The Weibuli
analysis predicts that brittle solids exhibit a "size
o hll=0.5
effect" (larger specimens are weaker in tension) which
LL •', hll-l.0 depends on the Weibull modulus of the material.
Here we demonstrate that the use of the Weibull
distribution to account for variability in the cell
10 I I I wall "strength leads to a "cell size effect" which,
5 10 15
too, depends on the Weibull modulus of the cell wall
Semi-crack length / cell size, all (-) material. We first summarize results of the Weibull
Fig. 6. Fracture toughness given by finite element analysis analysis for the probability of failure and the mean
for honeycombs of different cell geometries. failure stress for a brittle material under uniform
1622 HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS

tension and under bending. We then apply the analy- where


sis to the fracture of brittle honeycombs.
The Weibull distribution for the probability of h= - g"dV,
dV=ao"
failure, Pf, in a brittle material of volume, V, loaded
with a uniformtensile stress is [4] depends on the volume of solid and material proper-
ties (1/, a0 and m). The mean failure stress can be
Pf = 1 - exp_[- V(-----~-0
] - aa. ) " for a > Gu expressed in terms of the material properties, the
volume of solid and the loading configuration
and
Pf = 0 for a < au (10) Of=~P, daf=f;exp[-a'['h]dat
where a,, ao and m are material properties character-
izing the inherent flaw population in the material.
The Weibull modulus m, an empirical constant, is
related to the flaw size distribution of the brittle
material [5]. Because fracture becomes significant
whenever brittle materials are in tension, it is often
assumed that au = 0. Evaluating the mean failure stress for several typical
The mean tensile strength can be calculated from loading configurations we find
the failure probability distribution [equation (10)] to
be [6] #f=~F(l + 1 ) [2(m + 1)2]'/"

3"=
; a dPf = j0r,, - Pf) da = f0+P, da (three-point bending) (16a)

~, = ~ - ~ F (1 + I ) [ 2 ( m + l ) ] '/"

(pure bending) (16b)


+o(m'-)
=V~/.F 1 + (11) _ oo

af=~--~F ~]L ~~2- _J


where P, is the survival probability and F(1 + m -l)
(four-point bending with equal spans). (16c)
is the gamma function.
Fracture of brittle honeycombs is controlled by Comparison of the above results indicates that for
the tensile failure of the cell walls in bending.The any loading system:
Weibull analysis for the failure probability of a beam
in bending, subject to a non-uniform,uniaxial stress af=-~t/ F(l+l)f(m) (17)
system, gives, following Jayatilaka [6]
G _Gu rn where f(m) is a function which depends only on
the loading configuration. Beams of different volumes
under the same loading configuration have equal
for a > au
values of f(m). For two specimens of different
Pr=0 for a<a.. (12) volumes, V1 and I/2, made from a material with
a Weibull modulus, m, loaded in the same configur-
The stress at any point of the beam can be written ation, the ratios of their mean failure strengths in
in terms of the failure stress, a r bending is given by

a = afg (13) +,,,, (,,,ql,,,.. (18)


#f,,= k.Vl]
where g depends on the geometry and the loading
system. The survival probability when a, = 0 is This equation describes the size effect for brittle
materials in bending; it is the same as that for brittle
materials in uniform tension.
We now apply these results to the fracture of a
brittle honeycomb. The tensile strength of the cell
+.°xp[_d+\ ao ] J
wall material is assumed to follow a Weibull distri-
bution. For honeycombs with the same cell shape
(equal h/l and 0) each inclined member of length l
=oxp[ g "
,l+, (Fig. 1) is under the same loading configuration,
independent of the relative density of the honeycomb
HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS 1623

and the cell size. For two honeycombs of different cell 100 --
sizes, the ratio of the mean moduli of rupture of the
cell walls, aft, l/aft.2, is [equation (18)]
fffs,, (V2~ l/m (b12t2~ '/"

r ll" I']"
O=fs,2 ~VI/] ~" ~ b / - - ~ ]
E
S,
12 (12~ 2`m ~ (19)
O

d
When t;/l 2 =fitl I (the two honeycomb specimens e-
¢_.
have the same relative density), then
2
~fS' 1 ~--" ( / 2 ~ 2/m (20)
m=3
~,,2 \llJ "
m=4
Assuming that two honeycombs of the same cell
shape have equal relative density and volume, the one m=5
with the smaller cells will have a larger cell wall
modulus of rupture. 10 I I I , u ,111 I a , , i ntnl
The fracture toughness can be described by 0.1 1 10
[equation (9)]
Cell size (1 turn)
Ktc = C3~x~'fs(xl)1/2 ~ " Fig. 8. The effect of cell size on the fracture toughness of
brittle honeycombs made from cell wall materials with
Weibull moduli of 3, 4 and 5 (t/l, h/l, O, specimen size all
The ratio of the fracture toughnesses of two speci- constant).
mens of honeycombs of equal volume and identical
cell shape but with different cell sizes is (assuming 5. EXPERIMENTAL
a/1 > 7)
Brittle, cordierite (2MgO-2AI203-5SiO2) honey-
K,c, , C3'l#fs'l(gll)l/2(~) 2
combs with square cells (h/l = 0, 0 = 45 °) are made
(21) by Coming Glass Works (Corning, New York).
Limitations in the cell sizes, relative densities and
specimen sizes available restricted the scope of the
Substituting equation (19) into (21) gives experimental program. Nevertheless, two sets of
tests were possible. The failure loads of notched
honeycombs with three different cell geometries,
KIc, 1 = (22)
loaded in three-point bending, were measured. The
Klc, 2 t2
fracture toughness of each honeycomb was then
calculated from the failure load and compared with
If the densities of the two honeycombs are equal those given by the finite element program for bend-
(tl/ll = t2/12), then ing. In addition, single cells were excised from the
KIC,, = ( ll ~ 1/2- 2/m honeycombs and tested in compression to measure
Ktc,2 \~2// (23) the modulus of rupture of the cell wall material. The
results were used to confirm the Weibull analysis
Noting that the Weibull modulus is larger than for a single cell wall.
zero, we find that: The failure load of notched cordierite honeycombs
were measured in three-point bending in an Instron
(1) F o r a cell wall material with a Weibull
testing machine using the configuration shown in
modulus greater than 4, increasing the
Fig. 9 (a/w = 0.5). Load and crosshead displacement
cell size increases the fracture toughness.
were recorded on an X - Y plotter. Twenty specimens
(2) For a cell wall material with a Weibull
of each of three different honeycomb geometries
modulus less than 4, increasing the cell
were tested; the dimensions of each geometry are
size decreases the fracture toughness.
listed in Table 3. Bending tests were preferred over
(3) For a cell wall material with a Weibull
tensile tests due to the limited size of specimens
modulus equal to 4, the cell size has no
available.
effect on the fracture toughness.
Fifty-unit square cells were cut out of each
The dependence of the fracture toughness on cell geometry of honeycomb and loaded in compression
size for materials with Weibull moduli of 3, 4 and 5 along their diagonal in a deformation stage in a
is shown in Fig. 8. scanning electron microscope; the failure load of each
1624 HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS
P

"F
I

L ./
I
O

Fig. 9. Notched three-point bend test on a brittle honeycomb (W = 38 mm, S = 140 mm, a = 19 mm,
b = 13 ram).

unit cell was recorded. Micrographs of the loading Since the ratio of a/l is larger than 7 for all
configuration and failed specimen are shown in of the honeycomb beams tested, we assume that
Fig. 10; uniaxial compression of the cell induces this equation for a fully dense solid can be used
bending in each cell wall. The dimensions of each cell to calculate the fracture toughness for the honey-
wall were measured using the cursor in the SEM. combs tested; the average results are listed in
The modulus of rupture of the broken cell wall was Table 3. Using the measured cell wall modulus of
then calculated from the failure load and the cell rupture from Table 3 and the cell dimensions, the
dimensions. fracture toughness of the honeycomb can also be
calculated using the finite element analysis, modified
6. RESULTS AND DISCUSSION to account for the bending loading configuration.
The FEM results are also listed in Table 3; agree-
For a solid brittle material, the fracture toughness ment is good, giving confidence in the finite element
can be calculated from the measured failure load, P, analysis.
in a three-point bend using the equation [7] The modulus of rupture of the cell wall material,
PS ars, is calculated from
Kic = ~w3/2f (a/w ) (24)
-0.354P 1.06Pl
where S and B are the span and thickness of the
trfs - b~ + -----T-
bt
specimen respectively, and f(a/w) is a factor which
where P is the applied compressive load on the unit
depends on the semi-crack length to specimen depth
cell at failure, t, I and b are the thickness, length and
ratio, a/w

f=3(a)°'5{1.99 - a ( l - a ) [ 2 . 1 5 - 3 . 9 3 a + 2 . 7 ( a ) 2 1 }
(25)

Table 3. Experimentalresults for honeycombs


width of the cell wall, respectively. The first term gives
Geometry 1 Geometry2 Geometry3
the normal stress arising from the axial load in the
t (ram) 0.458 0.282 0.167
l (ram) 2.220 1.428 1.187 cell wall while the second gives that arising from the
t/I (--) 0.206 0.198 0.141 bending moment. The average cell wall modulus of
rupture calculated for each of the three honeycomb
Fracture toughnessof notched beams
geometries is listed in Table 3.
all (--) 8.56 13.3 16.0
The Weibull parameters tr0 and m for the cell wall
(Kr~-ra- i 82.8 74.9 52.5 material of each honeycomb geometry are deter-
KIC. FEM mined from the experimental results using the follow-
(KN-ra-3/2) 85.5 82.7 48.8
ing statistical analysis. Assuming that failure of any
Note: 20 beams of each geometrytested; a/w = 0.5 for all beams.
cell wall will result in failure of the unit cell, Rosen's
Cell wall modulusof rupture weakest link theory [8] can be used. There are four
8f, (MPa) (measured) 18.5 25.0 28.0 cell walls, connected at both ends, in the unit cell; a
ao (MPa) [equation(28)] 19 21 21
m (equation(28)] 6 6 6 unit cell [Fig. 1l(a)] can be regarded as a solid, made
af~(MPa) [equation(29)] 19.6 25.3 28.5 up of four connected beams under the same loading
Note: 50-unitcells tested for each honeycombgeometry. condition [Fig. 1l([b,c)].
HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS 1625

D
(b)
(o)

(a)

(c)
Fig. 11. (a) Square unit cell; (b) four individual members in
square unit cell; (c) tension regions in an individual member
of the unit cell.

procedure; the resulting values are listed in Table 3.


The Weibull modulus of 6 suggests that the fracture
(b) toughness increases with cell size for all three honey-
comb geometries.
Fig. 10. (a) Uniaxial compression test along the diagonal of
a square unit cell, inducing bending in the cell walls. (b) A
failed unit cell, showing cracks at the cross-sections of n G e o m e t r y 1, m=6,
maximum moment. o 0 =19 M P a
o G e o m e t r y 2, m - 6 ,
The failure probability for the non-uniform stress A G e o m e t r y 3, m=6,
field can be written in an integral form (tr u = 0) o o -21 M P a

Pf=l-exp[--;,(~o)mdV]. (26)
0.8

Integrating the failure probability over the volume of "7"


v
the beam in tension [shaded area, Fig. 1 l(c)], neglect- a.-
ing the stress due to the axial force, gives the failure >~ 0.6
probability for each cell wall [Fig. 1 l(c)]

Pf = 1 - exp
[ V (tTfs~m1
2(m + 1)2 \¢r0} 1
(27) ¢0
JO
o
Q.
0.4
where V = blt is the volume of a single cell wall. A
unit cell is composed of four cell walls (V = 4blt); the
14.
failure probability for a unit cell can be obtained
directly

Pf = 1 -- exp
[ 4bit
2(m + 1): \ a 0 / _J"
(28)
0.2

The measured moduli of rupture are ordered to


I ta dh.O""5'
give a failure probability for each ars; the failure 0 10 20 30 40
probability is plotted against modulus of rupture in Cell wall modulus of rupture, ors (MPa)
Fig. 12. a0 and m for the cell wall material of each Fig. 12. Probability of failure plotted against cell wall
honeycomb geometry are found by fitting equation modulus of rupture for unit ceils tested as shown in Fig. 10.
(28) to the data of Fig. 12 using a trial and error Lines indicate the Weibull distributions fitted to the data.
1626 HUANG and GIBSON: FRACTURE TOUGHNESS OF BRITTLE HONEYCOMBS

Once the values of tr0 and m are known, the mean For honeycombs with a/l < 7, the effective fracture
modulus of rupture of the unit cell can be easily toughness is reduced, by a factor given in Fig. 7; this
calculated [equation (16a)] factor is insensitive to changes in the cell geometry
(h/l and 0).
The variability in the cell wall modulus of rupture
has been accounted for by assuming that it follows
Equation (29) can be used to check the goodness a Weibull distribution. Tests on unit cells excised
of the previous procedure of determining the material from brittle cordierite honeycombs indicate that this
properties. Using the values of cr0 and m listed in assumption is valid. The Weibull analysis suggests
Table 3, the predicted mean moduli of rupture that the cell wall modulus of rupture depends on the
are 19.6, 25.3 and 28.5MPa [equation (29)]. The volume of the cell wall, leading to a cell size effect for
experimental values are 18.5, 25.0 and 28.0MPa; the fracture toughness. If the Weibull modulus, m, of
agreement is good. the cell wall material is greater than 4, the fracture
The Weibull modulus also can be determined from toughness increases with increasing cell size. If m
experimental data by using equation (28), which can is less than 4, the fracture toughness decreases with
be written in the following form increasing cell size. And if m = 4, there is no cell
size effect. For most brittle materials, m is typically
In [ln (1- ~ r ) ] larger than 4, suggesting that the fracture toughness
increases with cell size. As m approaches infinity, the
fracture toughness increases with the square root of
= ln(~om)+m ln(af~)-ln[2(m+l)2]. (30) cell size [equation (23)], as suggested by the existing
model.
Trustrum and Jayatilaka [9] recommend estimating
the Weibull modulus using the method of least Acknowledgements--We gratefully acknowledge the techni-
squares if the sample size is about 40 and both cal assistance of Dr J. T. Germaine and the financial sup-
the accuracy and ease of calculation are of concern. port of the Army Research Office Program in Advanced
Construction Technology (Grant Number DAAL 03-87-
Using this method, the calculated Weibull moduli
K-0005).
are 5.25 (geometry 1), 6.69 (geometry 2) and 4.53
(geometry 3); all of them are larger than 4, implying
that the fracture toughness will increase with cell size REFERENCES
for specimens of constant t/l.
1. M. F. Ashby, Metall. Trans. 14A, 1755 (1983).
2. S. K. Maiti, M. F. Ashby and L. J. Gibson, Scripta
7. CONCLUSIONS metall. 18, 213 (1984).
3. L. J. Gibson and M. F. Ashby, Cellular Solids: Structure
The existing model for the fracture toughness of and Properties. Pergamon Press, Oxford 0988).
4. W. Weibull, J. appl. Mech. 18, 293 (1951).
brittle honeycombs has been modified to take into 5. A. De S. Jayatilaka and K. Trustrum, J. Mater. Sci. 12,
account the effect of short cracks and variability in 1426 (1977).
the cell wall modulus of rupture. The finite element 6. A. De S. Jayatilaka, Fracture of Engineering Brittle
study indicates that the existing model, based on Materials. Applied Science, New York (1979).
7. J. E. Srawley, Int. J. Fract. 12, 475 (1976).
a continuum approach, is satisfactory for honey- 8. B. W. Rosen, AIAA J. 2, 1985 (1964).
combs of low relative density (t/l < 0.2) and with 9. K. Trustrum and A. De S. Jayatilaka, J. Mater. Sci. 14,
semi-crack lengths greater than 7 times the cell size. 1080 (1979).

Vous aimerez peut-être aussi