Heinrich J.G., Aldinger F. (Eds.) Ceramic Materials and Components for Engines
Подождите немного. Документ загружается.
samples. In this study, parallel plates with the same
dimensions were used as electrodes, forming an almost
uniform electric field between the electrodes. In such a
case, the effect of a microstructural weak point is
equivalent, regardless of where it is positioned between
the electrodes.
On the other hand, in mechanical bending tests, it is
widely known that there is a stress gradient in the
thickness direction; tensile stress is generated on the
bottom face and compressive stress is generated on the
top face. A large crack on the top face might therefore
not be recognized
as
a weak point in the mechanical
bending test, whereas the same crack would act as a
serious electric flaw in the parallel plates electrode
system. In such a case, another weak spot, such as an
internal void near the bottom face could act as a stress
concentrator in a bending fracture.
In conclusion, the results of high voltage screening
are similar to those of stress screening when surface
flaws are preferentially oriented perpendicular to the
tensile direction, but surface flaws that parallel the
direction of tension are much less apparent
in
measurements of bending strength.
In
other words,
high voltage screening of ceramic parts can detect
significant flaws that are parallel to the tensile direction,
which cannot be detected by mechanical screening using
flexure or tensile methods.
REFERENCES
1)A. Kishimoto, K. Endo, Y. Nakamura, N. Motohira,
H.
Yanagida, and K. Sugai, Effect of high voltage
screening on strength distribution for Ti02 ceramics,
J.
Am. Ceram. SOC., 78
[8]
(1
995) 2248-2250.
2)
K.Endo, A.Kishimoto, N. Motohira, Y. Nakamura and
H.
Yanagida, High Voltage Screening on Ceramics
Parts with Bimodal Strength Distribution, J. Ceram.
SOC. Jpn.,
103
[lo] (1995) 1090-1092.
3)
A.Kishimoto, K.Endo, N. Motohira, Y. Nakamura H.
Yanagida, and
M.
Miyayama, Strength Distribution of
Titania Ceramics after High Voltage Screening,
J.
Mater. Sci.,a (1996) 3419-3425.
4)
T. Tanaka and A. Kishimoto, "Reliability
improvement of titania ceramics with surface flaw
through high voltage screening, Kor.
J.
Ceram. SOC.,
5)
A.Kishimoto,
K.
Koumoto and
H.
Yanagida,
Mechanical and Dielectric Failure of BaTi03
Ceramics, J. Mater. Sci.,
24
[2] (1989) 698-702.
6)
A. Kishimoto, K. Koumoto and
H.
Yanagida,
Comparison of Mechanical and Dielectric Strength
Distributions for Variously Surface-Finished Titanium
Dioxide Ceramics,
J.
Am. Ceram. SOC.,~ [8]
7)
Y. Nakamura, M. Suzuki, N. Motohira, A. Kishimoto
and H. Yanagida, Comparison Between the
Mechanical and Dielectric Strength Distributions for
Hardened Gypsum, J. Mater. Sci.,
32
[I] (1997)
8)
A. Kishimoto, M. Nameki, K. Koumoto and
H.
Yanagida, Microstructure dependence of mechanical
and dielectric strength distributions: I
,
porosity, Eng.
Fracture Mechanics,
40 [4/5]
(1
99
I)
927-930.
9)
N. Yoshimura,
A.
[to,
J.
Funaki, and T. Ogasawara,
Effect
of Pore on Electrical Conduction and Dielectric
Breakdown for Dielectric Ceramics, Trans of. IEE.
Japan-A
108 [4] (1988) 155-161.
10)
Y.Matsuo, T. Ogasawara, S.Kimura and E. Yasuda,
Statistical Analysis of the Effect of Surface Grinding
on the Strength
of Alumina Using Weibull's
Multi-Modal Function,
J.
Mater. Sci.,
22
(1987)
11)
T. Asokan and T.
S.
Sudarshan, Dependence
of
the
Surface Flashover Properties of Alumina on Polishing
Abrasive Parameters, IEEE, Trans, Elect, Insl.,
28
[4]
12)
E.
K.
Beauchamp, Effect on Microstructure on Pulse
Electrical Strength of MgO,
J.
Am. Ceram. SOC.,~
13)A.
G.
Evans and
S.
M.
Wiederhorn, Proof Testing
of
Ceramic Materials. Analytical Basis for Failure
Prediction, Int.
J.
Fract. Mech., 10[3] (1974) 379-392.
5[4](1999)386-389.
(1989) 1373-1376.
11
5-1
18.
1482-88.
(1993) 3419-3425.
[lo]
(1971) 484-487.
232
IN SITU OBSERVATION OF TENSION AND CYCLIC FATIGUE DAMAGE
IN Hi-Nicalon FIBEWSiC COMPOSITE
Y.
Kaneko", T. Mamiya**, M. Mizuno***,
SJ.
Zhu*,
Y.
Kagawa**,
Y.
Ochi*.
(*)
The University
of
Electro-Communications, Tokyo, Japan
(**)
Institute
of
Industrial Sciences, University
of
Tokyo, Tokyo, Japan
(***)
Japan Fine Ceramics Center, Nagoya, Japan
ABSTRACT
In
situ observation of cyclic fatigue crack initiation
and propagation of Hi-NicalonTM fibers reinforced Sic
composite at room temperature has been carried out by
scanning electron microscopy. Fatigue crack initiates at
the pores between fiber bundles. Delamination at the
interfaces between longitudinal fiber bundles and
transverse fiber bundles is the main damage process.
This is because the inconsistent deformation causes
shear stress. Due
to
delamination, the transverse crack
deflects to the interfaces. As a result, crack propagation
is retarded. There
is
no
single crack propagation.
Therefore,
it
is impossible
to
use fracture mechanics
parameter to characterize crack propagation.
INTRODUCTION
Ceramic matrix composites (CMC) have been
designed for use at high temperature because they have
good fracture toughness and other performances.
In
high
performance turbine engines, high temperature
components such as after-burner vanes, combustion
chambers and even turbine blades may eventually be
manufactured from CMCs. CMCs may also take an
important role as fascinating devices for hot fractures
NicalonTM fiber/SiC composite has many good
mechanical properties, but it has the low creep
resistance owing to a viscous flow at temperatures as
low as 1000-1200 "C. The presence of SiCxOy
amorphous phase in NicalonTM fibers
is
responsible for
Young's modulus degradation and the low creep
resistance owing to a viscous flow at temperatures as
low as 1000-1200 "C. The elimination of SiCxOy from
the fibers by electron irradiation in vacuum instead of
curing in air can improve creep resistance. The modified
NicalonTM fibers are called Hi-NicalonTM fibers (Nippon
Carbon, Tokyo, Japan). Hi-NicalonTM fibers with a very
low oxygen content (c1 wt%.) exhibit improved thermal
stability [4-71.
Carbon coating layer in SiC/SiC leads to low
oxidation resistance at high temperatures in air. A glass-
forming, boron-based particulate can be added
to
the
matrix that reacts with oxygen
to
produce a sealant glass
that inhibits oxidation. This technology
is
applied
to
SiC/SiC composites. The modified SiC/SiC is called
enhanced SiC/SiC composite.
To
increase the
[l-31.
mechanical properties of SiC/SiC, Hi-NicalonTM fibers
are used
to
reinforce the enhanced
Sic
matrix.
Fatigue
is
responsible for the majority of failure
of
structural components. The fatigue mechanisms
in
composite materials are more complex and involved a
multitude of spatially distributed and interacting
mechanisms[8-11]. SiC/SiC composites exhibited a
definite fatigue limit in classical S-N curves, hysteresis
loops and the reduction of Young's modulus with
number of loading cycles[12-131. It was reported that
the fatigue limit was
65-80%
of tensile strength.
Although
cyclic
fatigue of CMCs was investigated and
possible fatigue mechanisms were proposed, in situ
observation of fatigue crack initiation and propagation
in Sic-fiber/SiC composites
is
little [9].
The purpose of this paper is not only to present cyclic
fatigue data, but also to describe damage evolution of
cyclic fatigue of Hi-NicalonTM fibers reinforced Sic
composite at room temperature.
In
situ observation of
cyclic
fatigue crack initiation and propagation has been
carried out by scanning electron microscopy.
MATERIALS AND EXPERIMENTAL
PROCEDURE
The composites used in the investigation were
processed by chemical vapor infiltration (CVI) of Sic
into plane woven 0"/90" Hi-NicalonTM fibers preformed
(Honeywell Advanced Composites Inc., Newark,
DE,
USA). Before the infiltration the Sic fibers were coated
with carbon by chemical vapor deposition (CVD) to
decrease interface bonding between fibers and the
matrix. This coating acts to raise strength and toughness
of the composites [14-151. The composite contained 40
vol% Sic fibers and about 10% porosity. The average
diameter of fibers is 14pm and each bundle consists of
500
bundles.
The testing specimen was machined from the panels
using diamond-cutting tools. The shape and dimensions
of the specimens for monotonic-tension and cyclic-
fatigue tests at room temperature (RT) are shown in
Fig.1. The specimen has two types of orientations in
fiber woven structure. One
is
called parallel orientation
(Fig.1 (a)), and the other is perpendicular orientation
(Fig.1 (b)).
A
difference between two kinds of
specimens
is
the different number of layers.
Perpendicular specimen has about
3
layers and parallel
specimen has about
8
layers.
233
The optical micrograph of transverse cross section of
Hi-NicalonTM /Sic composite that shows the fiber
distribution, porosity, boron-based particulate and
matrix are shown in Fig.2.
In
this micrograph the light
regions correspond to the CVI Sic matrix, the fibers of
from side
to
side are longitudinal bundle
(0"
direction),
the oval shapes are fibers in transverse bundle
(90"
direction), and large black regions are pores between
bundles. The pores are connected to each other with
these matrix-rich regions. Such pores form a continuous
network that facilitates matrix deposition during the
CVI
process [16].
The tensile and fatigue tests were performed using a
floor-type Shimadzu hydraulic-servo fatigue testing
system (Shimadzu Corp., Kyoto, Japan) combined with
a scanning electron microscopy (SEM). Fig.3 shows
configuration of the testing machine with a SEM.
In
site SEM observation of fatigue crack propagation
on
these specimens were carried out by this testing system
at RT. The monotonic tensile tests were conducted in
vacuum at RT under lading control. Strain was
measured by strain gages. The tension-tension fatigue
tests were carried out under load control with a
sinusoidal loading frequency of lOHz in vacuum at RT.
The stress ratio (R) was
0.1.
After fracture, the
specimens were examined by SEM.
Fig.1. Woven fiber structure of
2D
plain-woven Hi-
NicalonTM fiber/SiC composite for monotonic tension
and
cyclic
fatigue tests at
RT.
(a) parallel specimen, (b)
perpendicular specimen. (c) shape and dimensions of
perpendicular specimen.
Fig.2. Microstructure of Hi-NicalonTM fiber
/Sic composite.
Fig.3. Configuration of the fatigue testing machine with
a scanning electron microscopy.
RESULTS AND DISCUSSION
Monotonic tensile behavior
The stress versus strain curves of Hi-NicalonTM /Sic
composites at room temperature are shown in Fig.4.
The curves indicate a linear elastic behavior up to a
proportional limit of 70MPa, and this stress is -30%
of
the average ultimate tensile strength (UTS) in parallel
specimen and -35% of the average UTS
in
perpendicular specimen. Each exhibited nonlinearly at
high stress because of matrix crack. Matrix crack
propagates between layers because of delamination of
layers. The number of layers in perpendicular specimens
are more than those in parallel specimens,
so
that the
tensile stress of perpendicular specimen is lower than
that of parallel specimen. However the failure strain
varied rather widely, from 0.2-0.7%. This variation
is
probably associated with effect of manufacturing stage.
i.e.
the fiber damage may occur either during the
weaving
or
during another stage of the composites
manufacture
[
171,
hence the pores and
2D
plain-woven
architecture may certainly lead to a nonuniform stress
and strain field under an applied load.
The modulus calculated from the linear portion of the
curves are 260GPa in two kinds
of
specimens. The
average tensile stresses, fracture strain and Young's
modulus are shown in Table
I.
234
250
300
1
0.45
Hi-NicalonTM
(+0.25/
(Dm)
/Sic
-
0.7.5)
260
205
(+30/-33)
n
2
200
5
150
v)
Q)
v)
t
100
50
0
0.1
0.2
0.3
0.4
0.5
0.6
Strain
(%)
Fig.4. Monotonic tensile stress versus strain of Hi-
NicalonTM fiber/SiC composite
in
vacuum at RT.
Table
I.
Tensile result of
Hi-NicalonTM fiber/SiC comoosite
Tensile Young's
strength
I
Fz::y
I
modulus
I
Hi-NicalonTM
arallel
Loading-unloading behavior
The loading-unloading curves of Hi-NicalonTM/SiC
composites at room temperature are shown
in
Fig.5 (a:
parallel specimen, b: perpendicular specimen). From a
reduction in modulus, as manifested
in
the slopes of
hysteresis loops, the matrix cracking may initiate at
about 70MPa. The loops appear wide in relation to those
measured
on
other CMCs
[
181, implying that interfacial
debonding and sliding occur during loading. Indeed,
examination of fractured specimens reveals that the
crack deflection and debonding of the interfaces
between longitudinal fiber and matrix (Fig.6).
0
0.1
0.2
0.3
0.4
0.5
0
0.1
0.2
0.3
0.4
0.5
Strain
(%)
Strain
(%)
Fig.5. Stress versus strain curves and hysteresis loops of
Hi-NicalonTM fiber/SiC composite
in
vacuum at
RT.
(a) parallel specimen, (b) perpendicular specimen
matrix
20pm
-.
Fig.6. Crack deflection and debonding of the interfaces
between longitudinal fiber and matrix.
o.04t.
'.
.
I..
. .
I.
'
'.
8..
. .
I
'
'.
.
j
-
n
0.0351
71
1
8
-0-perpendicular
-
0.03
L
.-
2
0.025
cn
0)
E
0.015
1
Q
c
E
0.02
0.01
i
0.005.
0
50
100
150
200
250
Stress
(MPa)
Fig.7. Variation of the permanent strain versus stress of
Hi-NicalonTM fiber/SiC composite.
+parallel
-0-
perpendicular
0.8
0
0
50
100
150
200
250
Stress
(MPa)
Fig.8. Young's modulus reduction versus stress of Hi-
NicalonTM fiber/SiC composite.
Cyclic
fatigue behavior
The cyclic fatigue life
in
vacuum at RT is shown
Fig.9. The cycles to failure increase with decrease of
maximum stress and the fatigue life of perpendicular
specimen is lower than that of parallel specimen. The
fatigue life for parallel specimen at 10' cycles is
200MPa which is
80
%
of UTS, and for perpendicular
specimen at
lo6
cycles
is
140MPa which is 70
%
of
UTS.
The fatigue limit of the composite at RT is much higher
than the stress of the matrix cracking (about 70MPa).
This means that the composites can avoid the unsteady
propagation of the matrix cracks induced by the first
loading during cyclic fatigue at the stress of the fatigue
limit.
235
In
site observation of fatigue damage shows that most
of cracks initiate at the sharp corners of large pores at
the crossover points of the fiber weaving (Fig.10) and
matrix cracks do not appear at regular intervals likes
other composites [19]. Figs.11 and 12
show
in site
observation of cyclic propagation processes in
perpendicular specimens. Fig.11 (a) shows the original
state. Because
the
maximum stress (145MPa)
is
higher
than the proportional limit of stress, matrix cracks
initiate at the pores (Fig.11 (b), N=1@). Scattering pores
in rich matrix religions may make the S-N dates vary
widely. Crack propagation from pores makes interface
delaminate between layers
(Fig.11
(c),
N=105)
and
because
of
connection of cracks this composite fractures
(Fig.11 (d), N=1.4x105).
3007.
..
.
4
I
perpendicular
.-
Y
loo
F
7
parallel
f
501
I
+
tensile test
for
perpendicular
..
t
I
A
tensile
test
for
parallel
I
1
I
I
I
or.
I.
I
1
loo
10'
lo2
lo3
10'
lo5
lob
lo7
Cycles
to
failure
Fig.9. Maximum stress versus cycles
to
failure of Hi-
NicalonTM fiber/SiC composite in vacuum at
RT.
(a) N=O: Original state
(b) N=103: Cracks initiate at the pores
(c)
N=105: Delamination cracks between layers
(d) Np1.4~10~
Fig.11. Microscopic damage evaluation of perpendicular
specimen under fatigue. (am,,=145MPa,
R=O.
1, f=lOHz
at
RT
in vacuum)
Fig.10. Crack initiates at the pores at
the
crossover point
of the fiber weaving of parallel specimen.
236
(a)
N=O:
Original state (Crack initiation by applying
mean stress of 88MPa)
(b)
N=10':
Delaminaiton propagation
(c) N=104: Connection of transverse crack with
delamination crack
(d) N,=1.9x104
Fig.12. Microscopic damage evaluation
I
perpendicular specimen under fatigue. (am,,=160MPa,
R=0.1, f=lOHz at
RT
in
vacuum)
Fig.12 (a) shows the original state, because of mean
stress of 88MPa crack initiates. Delamination
propagation between longitudinal and transverse fiber
bundles under fatigue (Fig.12 (b), N=10"), Connection
of transverse crack with delamination crack (Fig.12 (c),
N=105)
and fractures (Fig.12 (d),
N=1.9x104).
Under
fatigue delaminations are observed (Fig.13) and they
have three types. i.e. they initiate between (1) each fiber,
(2)
longitudinal and transverse fiber bundles, (3)fabric
layers. From this observation, (3) type of delamination
is
main fracture mechanisms shown
in
Fig.14. This is
reason that loading-unloading under fatigue severs
fabric layers by the extend and contact of bending
longitudinal fiber.
This
may be an original reason
different from 3D-fiber/matrix composites.
So
perpendicular specimen has much fiber bundles and
fabric layers, that fatigue life of perpendicular specimen
is
lower than that of parallel specimen.
Delaminations in this composite easily occur,
consequently the crack propagation
is
retarded, there is
no
single crack propagation and the unsteady of crack
propagation are many observed at unbroken test piece at
10'.
So
the interface debonding strength
in
this
composite
is
not strong between the fiber and matrix,
that this fiber coating of Carbon
is
good property.
As
a result,
the main fatigue mechanisms of the
composites are interface debonding between layers and
crack initiation at pores.
Fig.13. Delaminatrin at the interfaces between
longitudinal
(0")
fiber and transverse
(90")
fiber bundles
of perpendicular specimen.
90"
bundle
\
0"
bundle
delanhated
+.--
layers loading-unloading
Fig.14. Delamination model between fabric layers
under fatigue.
237
CONCLUSION
In situ observation of cyclic fatigue propagation of
Hi-NicalonTM fiber/SiC composite has been carried out
by scanning electronic microscopy attached to servo
type fatigue testing machine. The following conclusions
were obtained.
1)
The tensile strength of Hi-NicalonTM/SiC composite
at RT is 205-250MPa and Young’s modulus
is
260GPa.
2)
The tensile strength and fatigue life in parallel
specimens are higher than those of perpendicular
specimens.
3)
The fatigue life
is
governed by delamination crack
initiation and propagation between fabric layers.
4)
The number of layers in perpendicular specimens is
more than
those
in parallel specimens.
This
is
the
reason for the lower tensile strength and fatigue
strength in perpendicular specimens.
REFERENCES
(
1)
R.
John, L.P. Zawada and J.L. Kroupa, Stress
Due
to
Temperature Gradients in Ceramic-
Matrix-Composite Aerospace Components, J.
Am. Ceram. SOC.,
82
[
11
(1999) 161-168.
(
2)
T.
Ishikawa, K. Bansaku, N. Watanebe, Y.
Nomura, M. Shibuya and
T.
Hirokawa,
Experimental Stress/Strain behavior of SiC-
Matrix Composites reinforced with Si-Ti-C-0
fibers and estimation of matrix elastic modulus,
Comp. Sci. and Tech.,
58 (1998) 51-63.
W.R. Fohey, J.M. Battison, T.A. Nielsen and
S.
Hastings, Ceramic Composites Turbine Engine
Component Evaluation, Ceram. Eng.
Sci.
Proc.,
D.J. Pysher, K.C. Goretta, R.S. Hodder Jr. and
R.E. Tressler, Strengths of Ceramic Fibers at
Elevated Temperatures, J. Am. Ceram.
SOC.,
72
M. Takeda, Y. Imai,
H.
Ichikawa,
T.
Ishikawa,
T.
Seguchi and K. Okumura, Thermo-
mechanical Analysis of the Low Oxygen
Silicon Carbide Fibers Derived from
Polycarbosilane, Ceram. Eng.
Sci.
Proc.,
14
M.
Takeda,
J.
Sakamoto,
A.
Saeki,
Y.
Imai and
H. Ichikawa, High performance Silicon carbide
Fiber Hi-Nicalon for ceramic Matrix
Composites, Ceram. Eng. Sci. Proc.,
16 [4]
C. Vahlas, Thermodynamic Approach to the
Oxidation of Hi-NicalonTM Fiber, J. Am. Ceram.
(
3
)
16 [4] (1995) 459-466.
(
4)
[2] (1989) 284-288.
(5)
[7-81 (1993) 540-547.
(
6)
(1995) 37-44.
(
7
)
SOC.,
82[9] (1999) 2514-2516.
(8)
L.P. Zawada and L.M. Butkus, Tensile and
Fatigue Behavior of Silicon Carbide Fiber-
Reinforced Ahminosilicate Glass, J. Am.
Ceram.,
74 [ll] (1991) 2851-2585.
S.J. Zhu, Y. Kagawa,
M.
Mizuno,
S.Q.
Guo,
Y.
Nagano and
H.
Kaya,
In
site observation of
cyclic fatigue crack propagation of SiC-
fiber/SiC composite at room temperature,
Materials Sci. and Eng., A
220
(1996) 100-108.
(1
0)
A.G. Evans, F.W. Zok and R.M. McMeeking,
Fatigue of ceramic matrix composites, Acta
metall.mater.
Vo1.43,
No.3
(1995)859-875.
(
1
1
)
S.F.
Shuler, J.W. Holmes and
X.
Wu, Influence
of Loading Frequency
on
the Room-
Temperature Fatigue of a Carbon-Fiber/SiC-
Matrix Composite, J. Am. Ceram.
SOC.,
76 [9]
(
1
2)
D. Rouby and P. Reynaud, Fatigue behavior
related to interface modification during load
cycling in ceramic-matrix fiber composites,
Composites Sci. and Tech.,
48 (1993) 109-118.
(
1
3)
M. Mizuno,
S.
Zhu, Y. Nagano, Y. Sakaida,
Y,Kagawa and M.Watanabe, Cyclic-Fatigue
Behavior of SiC/SiC Composites at Room and
High temperatures, J. Am. Ceram. SOC.,
79 [12]
(
1
4)
F.
Rebillat, J. Lamon and R. Naslain, Properties
of Multilayered Interphases
in
SiC/SiC
Chemical-Vapor-Infiltrated
Composites with
“Weak” and “Strong” Interfaces, J. Am. Ceram.
SOC.,
81 [9] (1998) 2315-2126.
(1
5)
S.
Bertrand, P. Forio,
R.
Pailler and J. Lamon,
Hi-Nicalon/SiC Minicomposites with
(Pyrocarbon/SiC), Nanoscale Multilayered
Interphases, J. Am. Ceram.
SOC.,
82 [9] (1999)
(
1
6)
M.
Takeda, Y. Kagawa,
S.
Mitsuno, Y. Imai and
H.
Ichikawa, Strength of a Hi-NicalonTM
/Silicon-Carbide-Matrix Composite Fabricated
by the Multiple Polymer Infiltration-Pyrolysis
Process, J. Am. Ceram. SOC.,
82 [6] (1999)
(
1
7)
A.J. Eckel and R.C. Bradt, Strength
Distribution of Reinforced Fibers
in
a Nicalon
Fiber/Chemically Vapor Infiltrated Silicon
Carbide Matrix Composite, J. Am. Ceram. SOC.,
(
1
8)
J.M.
Domergue, F.E. Heredia and
A.G.
Evans,
Hysteresis Loops and the Inelastic Deformation
of
0/90
Ceramic Matrix Composites,
J.
Am.
Ceram.
SOC.
79 [l] (1996) 161-170.
(
1
9)
W. Kuo and
T.
Chou, Multiple Cracking of
Unidirectional and Cross-Ply Ceramic Matrix
Composites, J. Am. Ceram. SOC.,
78
[3]
(1995)
745-755.
(9)
(1993) 2327-2336.
(1996) 3065-3077.
2465-2473.
1579- 158 1.
72 [3] (1989) 455-458.
23
8
MULTI-AXIAL STRENGTH DATA
FOR Al2O3-
AND MgO-Zr02-CERAMICS
S.
Kruger, T. Kentschke*,
H.-J.
Barth
Technical University Clausthal, Institute for Tribology and Machine Kinetics,
D-38678
Clausthal-Zellerfeld, Germany
ABSTRACT
Multi-axial strength tests were carried out using
tu-
bular specimen made from A1203 and MgO-Zr02. For
biaxial tensile stresses the effect
of
the multi-axial
stress state is negligible because of the large scattering
of material strength. Therefore the computation of
ceramic components can be simplified using the maxi-
mum principal stress hypothesis.
A
multi-axial stress
hypothesis is not required. Under tension-compression
the tensile strength decreases with increasing compres-
sion load, if the compression is higher than one third
of
the uniaxial compression strength. In this case a multi-
axial strength effect exists and has to be taken into
account when computing ceramic components.
INTRODUCTION
Multi-axial stress conditions are present almost in
each component. For a computation a multi-axial stress
hypothesis is required to compare the multi-axial stress
state with uniaxial strength data.
A
large number of
such hypotheses are known for ceramics in literature,
but the differences are enormous, especially when
compressive stresses have to be taken into account.
Furthermore the validity is mostly unclear, such that
the engineer has finally no possibility to choosing the
“right” hypotheses for a certain application.
It is the intention of the authors to supply multi-
axial strength data of different ceramic materials and to
make recommendations for practical computation of
ceramic components.
A
theoretical analysis of the data
leading to a new multi-axial strength hypothesis has
been done in
(1).
EXPERIMENTAL SET-UP
The highly limited ductility of ceramics makes the
determination
of
strength data difficult. Even small
errors in load transfer can cause uncontrolled stress
intensities and may lead to incorrect measurements. In
the frame
of
a European Research Project funded by
the European Commission a testing machine was de-
veloped, that allows multi-axial testing of tubular
specimen
(2).
The gripping system of this machine, see
figure
1,
uses a conical bronze ring
(3)
to assure a safe
friction contact with the specimen
(5).
The vertical
pistons
(1)
and
(9)
apply the axial loads (compression
and tension). The internal pressure is applied by the
inner piston
(1
0).
I
(10)
piston internal
(1)
piston axial
compression
(8)
protection shell
(2)
sphencal plate
(6)
strain gauges
figure
1:
gripping system
Three strain gauges
(6)
are located on the outer pe-
rimeter of the specimen with
an
angle of
120”
between
them. This allows measuring the bending stresses dur-
ing the gripping and testing procedure. The geometry
of the tubular specimen was designed with the help of
Finite Element Method and enables tests in all quad-
rants of the principal stress field.
Figure
2
provides an
overview
of
typical ceramic testing methods using the
principal stress diagram.
239
GI;
CJ2
:
principal
stresses
p.
:extemalpresarre
pi
:intemlpnsSure
figure
2:
ceramic testing procedures
lube
test
.I
61
344-pint
bending
tesl
tube
test
To
assure, that the specimen fails under all loading
TESTING PROCEDURES
case in a desired and reproducible way, numerous pa-
rameters had to be taken into account. Parameters were
the manufacturing process, the gripping loads, the
application of strain gauges, the local risk of rupture
and the total probability of failure. The result is a
tu-
Table
1
illustrates the number of tests per load case
and material (A1203, trade name F99.7, and
MgO-Zr02,
trade name FZM, manufacturer FRIATEC GmbH
(4)).
bular specimen with a- reduced wall thickness in the
middle (fracture zone), see figure
3.
table
1:
test plan
figure
3:
specimen geometry
The risk of rupture reaches its maximum at the fracture
zone at the inner surface for all load combinations.
Therefore, the specimen is suitable for multi-axial tests.
*
10
tests
per
manufaduring
spdmen
charge
The tests for instantaneous failure are performed
with a maximum load rate of
50
MPds. This led to
durations of
4
s
(A1203)
and
10
s
(ZrOz) for tensile tests
and
40
s
(Zr02) for axial compression tests. The deter-
mined strength data were interpreted as the inert
strength. For technical reasons, the axial compression
strength of A1203 (3700 MPa) could not be reached for
the specimen. The testing machine can only reach a
maximum stress of 2300MPa
in
the axial compression
test.
240
The step-load-procedure according to Nadler (3)
was used to investigate the sub-critical crack
growth.
For uni- and multi-axial tests the stress was set to 65
%
of the inert strength when starting the test. Accordingly
the average duration of an experiment is 5
h
(MgO-
Zr02) and 4
h
(A1203) for the tension tests and
7
h for
the compression tests (MgO-ZrOZ).
For multi-axial tests the combination of loads were
chosen in a way that each data point in the principal
stress field has almost the same distance from another
point. The used multi-axial stress combinations are
listed
in
table
2.
The ratio of the combined axial loads
and the internal pressure is an indicator for the most
injurious stress case leading to a faster fracture of the
specimen.
int. pressure 517<557<599
table
2:
stress combinations for the multi-axial sub-
critical crack growth tests
615480~755
lmaterial
lstress
I
caJcaxOt
I
ctarjctanm
I
ratio
I
in1
Pressure 542483427
EXPERIMENTAL DATA
885<980<1086
INSTANTANEOUS FAILURE
adjusted Weibull modulus [-I
The stress-strain-behavior of A1203 and MgO-ZrOz
(figure
4)
shows the different Young's moduli (A1203
410GPa
f
13GPa and ZrOz 215GPa
k
1GPa) and
fracture characteristics of the ceramics. A1203 has an
ideal-elastic behavior until rupture starts. MgO-ZrOz
shows an increasing stress-strain-proportion in both the
tension and the compression quadrant for high loads.
The degressive curve results from non-reversible de-
formation of the specimen, which has been proved
in
a
mechanical hysteresis test (1).
1000
-
Ah03
~
MgO-210~
strain
E
61
-1
max load reached
-2500
5.1 <8.5<13.8 11.5~20,0<27,9 tension
int
pressure
6.oci3.9<29,6 4.5ci5.7e6.0
no
data 12.544.0e73.0 shear
The maximum fiber stresses (tensile stress
tsz
for
axial tension, tangential stress
oti
at the internal surface
of the specimen for internal pressure) were used for the
Weibull-statistical evaluation according to
DIN
EN
843-5. A comparison of internal pressure tests and
tension tests is illustrated in the Weibull diagrams in
figure 5a,b,
wherein the probability of failure of the
i-
th specimen Pfi
=
(i-O,5)/N is a function of the charac-
teristic specimen strength
tscha.
The non-dashed lines
represent the best approach of the two-parametrical
Weibull distribution function to the experimental data
using the Maximum-Likelihood-Method. The wide
range of the 95
YO
confidence intervals (dashed lines)
results from the small number of experiments. Fur-
thermore the experimental data were analyzed assum-
ing the volume (crack starts from defects in material)
and the surface (crack starts from defects on surface)
model. For calculations according to the volume model
the effective stressed volume
Vefi
is needed. To com-
pare the internal pressure and tension test the average
was transformed into the characteristic reference ele-
ment strength
(V,
=
lmm3) with the equation
specimen strength
(oOV,specimen, Veff,,specimen
=
659mm3)
tension 214
IN
is the Weibull integral.
449
When surface defects are assumed, the total risk of
rupture
is
calculated as the sum of the local risks
of
rupture of the internal and external surface. The
strength data were referred to a surface element of
Oo
=
1mm2. The calculated Weibull parameters are shown
in
table
3.
table
3:
Weibull parameters
tension
95
X
confidence interval[MPa],
char
speclmen strength,
max.
lateral
fiber
stress
Ichar. reference elem strength. Itension
I
423<458<496
I
6M)c629<650
I
Ivolume
model Ishear
I
nodata
I
1021~1058~1099
I
]char reference elem. strength. Itension
I
458<495<538
I
630~652<673
I
lsurface
model
Ishear
I
nodata
I
1020~1056~1097
I
figure
4:
o-&-diagram of MgO-Zr02 and
Al2O3
24
1