NASA Technical Memorandum 85797

NASA_TM-8579719840025763

AREVIEW OF CHEVRON-NOTCHED FRACTURE SPECIMENS

J. C. NEWMAN JR. J

FOR ItEFEREl'1CEl
. . . . e-, I .-. . . " ' - - '

SEPTEMBER 1984

LIBRARY COpy

.lANGLEY RESEARCH CENTER

LIBRARY, NASA
HAMPTON, VIRGINIA

Nl\SI\
National Aeronautics and
Space Administration
Langley Research Center
Hampton, Virginia 23665
 A REVIEW OF CHEVRON-NOTCHED FRACTURE SPECIMENS

J. C. Newman, Jr.
NASA Langley Research Center
Hampton, Virginia

SUMMARY

This paper reviews the historical development of chevron-notched fracture

specimens; it also compares stress-intensity factors and load line displace-

ment solutions that have been proposed for some of these specimens. The

review covers the original bend-bar configurations up to the present day

"short" rod and bar specimens. In particular, the results of a recent

"analytical" round robin that was conducted by an ASTM Task Group on Chevron-

Notched Specimens are presented.

In the round robin, three institutions calculated stress-intensity

factors for either the chevron-notched round-rod or square-bar specimens.

These analytical solutions were compared among themselves, and then among the

various experimental solutions that have been proposed for these specimens.

The experimental and analytical stress-intensity factor solutions that were

obtained from the compliance method agreed within 3 percent for both speci-

mens. An assessment of the consensus stress-intensity factor (compliance)

solution for these specimens is made.

The stress-intensity factor solutions proposed for three- and four-point

bend chevron-notched specimens are also reviewed. On the basis of this

review, the bend-bar configurations need further experimental and analytical

calibrations.
 The chevron-notched rod, bar, and bend-bar specimens were developed to

determine fracture toughness of brittle materials, materials that exhibit

"flat" or "nearly flat" crack-growth resistance curves. The problems

associated with using such specimens for materials that have a rising ".
crack-growth resistance curve are reviewed.

NOMENCLATURE

A normalized stress-intensity factor defined by Barker

a crack length measured from either front face of bend bar or load
line

initial crack length (to tip of chevron notch)

a, crack length measured to where chevron notch intersects specimen

surface

b length of crack front

B thickness of bar specimen or diameter of rod specimen

C* normalized compliance, EBVL/P, for chevron-notched specimen

E Young's modulus of elasticity

F normalized stress-intensity factor for straight-through crack

specimen

F* normalized stress-intensity factor for chevron-notched specimen

F*c normalized stress-intensity factor determined from compliance for

chevron-notched specimen

F*m minimum normalized stress-intensity factor for chevron-notched

specimen

H half of bar specimen height or radius of rod specimen

K stress-intensity factor (Mode r)

minimum stress-intensity factor for chevron-notched specimen

plane-strain fracture toughness (ASTM E399)

plane-strain fracture toughness from chevron-notched specimen

crack-growth resistance

2
k shear-correction parameter in Bluhm's slice model

P applied load

maximum test (failure) load

load-point half-displacement

half-displacement measured at top of specimen along load line

w specimen width

x,y,z Cartesian coordinates

a crack-length-to-width (a/w) ratio

crack-length-to-width (ai/w) ratios defined in Fig. 2

Poisson's ratio

INTRODUCTION

Chevron-notched specimens (Fig. 1) are gaining widespread use for

fracture toughness testing of ceramics, rocks, high-strength metals, and other

brittle materials [1-7]. They are small (5- to 25-rnrn thick), simple, and

inexpensive specimens for determining the plane-strain fracture toughness,

denoted herein as KIcV • Because they require no fatigue precracking, they

are also well suited as quality control specimens. The unique features of a

chevron-notched specimen, over conventional fracture toughness specimens,

are: (1) the extremely high stress concentration at the tip of the chevron

notch, and (2) the stress-intensity factor passes through a minimum as the

crack grows. Because of the high-stress concentration factor at the tip of

the chevron notch, a crack initiates at a low applied load, so costly pre-

cracking of the specimen is not needed. From the minimum stress-intensity

factor, the fracture toughness can be evaluated from the maximum test load.

Therefore, a load-displacement record, as is currently required in the ASTM

E399 plane-strain fracture toughness (K IC ) test procedure, is not needed.

3
Because of these unique features, some of these specimens are being considered

for standardization by the American Society for Testing and Materials (ASTM).

This paper reviews the historical development of chevron-notched

fracture specimens. The paper also compares the stress-intensity factor and

load-line displacement solutions that have been proposed for some of these

specimens. The review is presented in four parts.

In the first part, the review covers the development of the original

chevron-notched bend bars, the present day "short" rod and bar specimens, and

the early analyses for these specimens.

In the second part, the results of a recent "analytical" round robin

conducted by the ASTM Task Group on Chevron-Notched Specimens are presented.

Three institutions participated in the calculations of stress-intensity

factors for either the chevron-notched round-rod or square-bar specimen. They

used either three-dimensional finite-element or boundary-integral equation

(boundary-element) methods. These analytical solutions were compared among

themselves and among the various experimental solutions that have been deter-

mined for the rod and bar specimens. An assessment of the consensus stress-

intensity factor (compliance) solution for these specimens is presented.

In the third part, some recent stress-intensity factor solutions, pro-

posed for three- and four-point bend chevron-notched specimens, are reviewed.

In the last part, the applicability of chevron-notched specimens to

materials that have a rising crack-growth resistance curve is discussed.

4
 HISTORY OF CHEVRON-NOTCHED SPECIMENS

In 1964, Nakayama [1,2] was the first to use a bend specimen with an

unsymmetrical chevron notch. His specimen configuration is shown in Figure 1.

He used it to measure fracture energy of brittle, polycrystalline, refractory

materials. All previous methods which had been developed for testing homo-

geneous materials were thought to be inadequate. This specimen is unique in

that a crack initiates at the tip of the chevron notch at a low load, then

propagates stably until catastrophic fracture. Because of the low load, the

elastic stored energy in the test specimen and testing apparatus was small so

that the fracture energy could be estimated from the area under a load-time

history record.

Tattersall and Tappin [3] in 1966 proposed using a bend bar with a

chevron notch symmetrical about the center line of the specimen, as shown in

Figure 1. They used this specimen to measure the work of fracture on

ceramics, metals and other materials. The work of fracture was determined

from the area under the load-displacement record divided by the area of the

fracture surfaces.

In 1972, Pook [4] suggested using a chevron-notched bend bar to determine

the plane-strain fracture toughness of metals. He stated that, "If the KI

against crack length characteristic is modified, by the introduction of

suitably profiled side grooves, so that there is a minimum at a/w ~ 0.5, and

the initial KI is at least twice this minimum, it should be possible to omit

the pre cracking stage, and obtain a reasonable estimate of KIc from the

maximum load in a rising load test. Pook's "suitably profiled side grooves"

is the present day chevron notch. However, he considered only the analytical

treatment needed to obtain stress-intensity factors as a function of crack

length for various types of chevron notches. He did not study the experimental

aspects of using a chevron-notched specimen to obtain KIC •

5
 The nomenclature currently used for a straight-sided chevron notch in a

rectangular cross section specimen is shown in Figure 2. The specimen width,

w, and crack length, a, are measured from the front face of the bend bar (or

from the load line in the knife-edge-loaded specimen). The dimensions aO

and a, are measured from the edge of the bend bar (or load line) to the

vertex of the chevron and to where the chevron intersects the specimen

surface, respectively. The specimen is of thickness B and the crack front

is of length b.

Pook [4] used the stress-intensity factor solution for a three-point bend

bar with a straight-through crack [B]and a side-groove correction proposed by

Freed and Kraft [9] to obtain approximate solutions for various shape chevron

notches. The stress-intensity factors for a chevron-notched specimen, KCN '

was given by

(B)' /2
KSTC b

where KSTC is the stress-intensity factor for a straight-through crack in a

bar having the same overall dimensions. Figure 3 illustrates the unique

stress-intensity factor solution for a chevron-notched specimen compared to a

straight-through crack specimen. The dashed curve shows the normalized

stress-intensity factors for the straight-through crack as a function of

a/we This curve is a monotonically increasing function with crack length.

The solid curve shows the solution for the chevron-notched specimen. For

a = aO' the stress-intensity factor is very large but it rapidly drops as the

crack length increases. A minimum value is reached when the crack length is

between aO and a,. For a > a" the stress-intensity factors for the

chevron-notched specimen and for the straight-through crack specimen are

identical because the configurations are identical.

6
 The analytical procedure used by Pook [4] to determine the stress-

intensity factor as a function of crack length was an engineering approxima-

tion. At that time, no rigorous analysis had been conducted to verify the

accuracy of equation (1).

In 1975, Bluhm [10] made the first serious attempt to analyze the

chevron-notched bend bars. The three-dimensional crack configuration was

analyzed in an approximate "two-dimensional" fashion. The specimen was

treated as a series of slices in the spanwise direction. Both beam bending

and beam shear effects on the compliance of each slice were considered but the

inter-slice shear stresses were neglected in the analysis. Then by a

synthesis of the slice behavior, the total specimen compliance was determined.

The slice model, however, introduced a "shear correction" parameter (k) which

had to be evaluated from experimental compliance measurements. Experimental

compliance measurements made on an "uncracked" chevron-notched bend bar

(aO = 0 and a1 = 1) were used to determine a value for the "shear

correction" parameter for three- and four-point bend specimens. Bluhm

estimated that the slice model was capable of predicting the compliance of the

cracked Tattersall-Tappin type specimen (see Fig. 1) to within 3 percent.

Bluhm did not, however, calculate stress-intensity factors from the compliance

equations. Later, Munz et ale [7] did use Bluhm's slice model to calculate

stress-intensity factors for various chevron-notch bar specimens.

In the following, the concept proposed by Pook [4] to determine the

KIc value for brittle materials using chevron-notched specimens will be

illustrated. Figure 4 shows stress-intensity factor, K, plotted against

crack length. The solid line beginning at a O and leveling off at KIC is

the "ideal" crack-growth resistance curve for a brittle material. The dashed

curves show the "crack-driving force" curves for various values of applied

load on a chevron-notched specimen. Because of the extremely large K value

7
at a = aO' a small value of load, like P" is enough to initiate a crack at

the vertex of the chevron. At load P" the crack grows until the crack-drive

value is equal to KIc ' that is, the intersection point between the dashed

curve and horizontal line at point A. Further increases in load are required

to extend the crack to point Band C. When the maximum load, Pmax ' is

reached the crack-drive curve is tangent to the KIc line at point D. Thus,

the K value at failure is equal to K!c. The tangent point also corresponds

to the minimum value of stress-intensity factor on the crack-drive curve

(denoted with a solid symbol). Therefore, KIcv is calculated by

P
K max F*
Icv m
BIW
where Pmax is the maximum failure load and F* is the minimum value of the
m

normalized stress-intensity factor. Because F* is a predetermined value for

m
the particular chevron-notched configuration, it is necessary only to measure

the maximum load to calculate KIcv •

This maximum load test procedure can only be applied to brittle materials

with flat or nearly flat crack-growth resistance curves. Many engineering

materials, however, have a rising crack-growth resistance curve. The problems

associated with using chevron-notched specimens for these materials will be

discussed later.

Chevron-Notched Rod and Bar Specimens

Although the bend bars were the first type of chevron-notched specimens

to be tested, the knife-edge loaded rod and bar specimens have received more

attention. In the next sections, the rod and bar specimens are reviewed.

This review also includes the analytical round robin; in which, the rod and

bar specimens were analyzed. In a later section, some recent results on the

chevron-notched bend bars are also reviewed.

8
 Barker [5,6] in the late 70's, proposed the "short" rod and bar specimens,

Figure 1, for determining plane-strain fracture toughness. These specimens are

loaded by a knife-edge loading fixture [5,7] resulting in an applied line

• load, P, at location L, as shown in Figure 5(a). Figure 5 shows the

coordinate system used to define dimensions of the most commonly used rod and

bar specimens. (Here the chevron notch intersects the specimen surface at

x =w or a1 = 1.)

Rod Specimens.- Since 1977, the chevron-notched rod specimen, with

w/B = 1 .45, has been studied extensively. Figure 6 shows a comparison of the

minimum normalized stress-intensity factor as a function of the year the

result was published. The open symbols denote the method by which the values

were obtained. Each method will be discussed. The solid symbols show the

results of corrections that have been made by the author.

In 1977, Barker [5] used the KIc value obtained from ASTM E399 compact

specimens made of 2014-T651 aluminum alloy to determine the minimum stress-

intensity factor for the rod configuration by a "matching" procedure. The

minimum stress-intensity factor was given by

P
max A
K
m

where A is Barker's normalized stress-intensity factor that accounts for the

configuration. By equating Km to KIC ' the value of A was 20.8.

Equation (3) can be rewritten into the form

P
max
Km = F*
m
(4)

where the value of F* is 26.3 (v = 0.3). (Equation (4) is the form commonly
m

used for compact and knife-edge loaded specimens. The same form will be used

herein.) Table 1 summarizes the minimum normalized stress-intensity factors

9
obtained by various investigators; also listed are particular dimensions of

the rod configuration used.

In 1979, Barker [11] replaced the term (1-V 2 ) in equation (3) with unity

without changing the value of A. Thus, the value of F* dropped by about

m
5 percent. The value of F* should have remained at 26.3 for v = 0.3.
m

Barker and Baratta [12] in 1980 extensively evaluated the fracture tough-

ness of several steel, aluminum, and titanium alloys using the rod specimen

and KIC values measured according to ASTM Standard Method of Test for Plane-

strain Fracture Toughness of Metallic Materials (E399-78). They found that

the critical stress-intensity factors, calculated from the rod specimen data

using F*
m
= 25.5 [12], were consistently low, averaging about 6 percent below

the values. They concluded that F* for the test configuration used in
m

their study should be increased by 4 percent to a value of 26.5.

Earlier, Barker and Guest [13] had conducted an experimental compliance

calibration on the rod specimen and had obtained a value of F* as 29.6.

m
Their specimen, however, had a w/B ratio of 1.474 [14]. Subsequently, the

value of F* was corrected to a value corresponding to a w/B ratio of 1 .45

m
by using a "constant moment" conversion described in reference 15. The

corrected value of F* (28.7) was about 3 percent lower than the compliance
m

value from reference 13, as indicated in Figure 6.

Beech and Ingraffea [16,17] were the first to rigorously numerically

analyze a chevron-notched specimen. They used a three-dimensional finite-

element method to determine stress-intensity factor distributions along the

crack front and stress-intensity factors from compliance for the chevron-

notched rod. The specimen they analyzed, however, differed from the proposed

standard (w/B = 1.45; aO = 0.332; and a1 = 1) specimen analyzed in the ASTM

round robin in three ways: (1) the load line was at the front face of the

10
 specimen rather than at 0.05B into the specimen mouth, (2) the slot height

(0.03B) was modelled (see Fig. 5(a)) as zero, and (3) the square- or V-shaped

cutout at the load line was not modelled. (The effects of these differences
• in specimen configuration on stress-intensity factors are discussed in

reference 15 and will not be repeated here.) The stress-intensity factors

reported in references 16 and 17 from their crack front evaluations were

considerably lower (6 to 17 percent) than their values determined from a

plane-strain compliance relation. They used their plane-strain compliance

results to obtain a minimum stress-intensity factor. The value of F* from

m

reference 17 was 4 percent higher than the value given in reference 16. The

difference in these results was due to the manner by which the compliance

derivative was evaluated. The values of F* given in Table 1 were their

m
plane-strain compliance values and, in parentheses, values obtained from a

plane-stress compliance relation. The reason for using plane stress, herein,

was that the displacements remote from the crack front are more nearly

controlled by plane-stress conditions and, consequently, the plane-stress

compliance relation would be more correct than using plane strain. (Also, all

other results reported in Table 1, which were determined from compliance, were

made with the plane-stress relation.) If the plane-strain compliance relation

(with v = 0.3) had been used, the F*

m
values would have been about 5 percent

higher than the plane-stress values (square and triangular symbols) shown in

Figure 6.

Bubsey et ale [18], Shannon et ale [19], and Barker [15] used the experi-

mental compliance (plane stress) relation to evaluate stress-intensity factors

for the "short" rod specimen. Bubsey et ale and Shannon et ale used aluminum

alloy specimens with w/B ratios of 1.5, 1.75 and 2 for a wide range in

aO. Their values in Table 1 and Figure 6 were interpolated for aO = 0.332

11
and extrapolated to w/B = 1.45 by using second degree polynominals in terms

of aO and w/B, respectively. Because the proposed standard dimensions are

quite close to those used in the experiments, the interpolation and

extrapolation procedure is expected to induce only a small error (probably

less than 2 percent). Barker [15], on the other hand, used fused quartz

(v = 0.17) on specimens with W/B = 1.45. He reported a value of A as

23.38, therefore, F*would be about 28.2.

m
Raju and Newman [20], using a three-dimensional finite-element method,

studied the effects of Poisson's ratio (v) on stress-intensity factors for the

rod specimen (w/B = 1.45). Their results indicated that a specimen with

v = 0.17 (fused quartz) would have a stress-intensity factor about 2 percent

lower than a specimen with v = 0.3 (aluminum alloy). Thus, if Barker [15]

had used an aluminum alloy specimen, his experimental compliance value (F*)
m
would have been about 28.8.

Raju and Newman [20] and Ingraffea et ale [21] determined the minimum

stress-intensity factors for the rod specimen (w/B = 1 .45) using compliance

calculations from three-dimensional finite-element analyses. Each used the

plane-stress compliance relation. Raju and Newman obtained a value of F*

m

as 28.4 (as plotted in Fig. 6) and Ingraffea et ale obtained a value of 28.3

(not plotted). The result from Raju and Newman, however, was estimated to be

about 1.5 percent below the "true" solution based on a convergence study.

Thus, the corrected value of F* would have been about 28.8.

m
Ingraffea et ale [21] also used a boundary-element (boundary-integral)

method to determine the minimum stress-intensity factor from compliance. They

obtained a value of F* as 28.3 (as plotted in Fig. 6), the same as from
m
their finite-element analysis. The results from Ingraffea et ale [21] and

Raju and Newman [20] were part of the analytical round robin, previously

mentioned, and these results will be discussed and compared later.

12
 A comparison of minimum stress-intensity factors for the rod specimen

(w/B 1 .45) shows several interesting features. First, the method of using

to determine F* gives results that are about 8 percent below experi-

m
• mental and analytical compliance methods. Although the specimen used by

Barker [5,11] and Barker and Baratta [12] was somewhat different than the
•
proposed standard specimen, these differences are not expected to be signifi-

cant (see ref. 15, page 309). The specimens used in references 11 and 12 had

chevron notches with curved sides instead of straight sides. Barker [14]

argues that the calibration should be the same in a straight-sided and a

curved-sided chevron-notch specimen, provided that the crack front length (b)

and the rate of change in b is the same in both specimens at the minimum

stress-intensity factor. He determined that the aO and a1 for an

"equivalent" straight-sided chevron-notch specimen should be 0.343 and 0.992,

respectively. These values are quite close to those for the specimen analyzed

in the ASTM round robin with straight-sided chevron notches. Therefore, at

present, the 8-percent discrepancy in the values of F* cannot be explained

m
from differences in specimen configuration.

One possible source of error in the Klc-matching procedure may be due to

the different loads used in each test procedure. In the KIc test, the

5-percent secant offset load, P , is used to calculate KIC • The PQ load

Q
is, always less than or equal to P max' the maximum test (failure) load.

Whereas, in the chevron-notch specimen test, the maximum load is always used

to calculate Klcv • For example, if Pmax was used to calculate KIc

Q, then
instead of P K would tend to be higher than the current value.
lc

Thus, the value of F* would also tend to be higher than the current value
m
(circular symbols in Fig. 6). This would make the value of F*, determined
m
from the KIC-matching procedure, in closer agreement with the experimental and

analytical compliance values shown in Figure 6.

13
 Second, the experimental [13, 15, 18, 19) and the recent analytical

[20, 21) compliance determination of the minimum stress-intensity factor agree

within about 3 percent of each other. Accounting for the fact that one of the

analyses [20) was about 1.5 percent low, based on convergence studies, and

that reference 15 used fused quartz, which has a low value of poisson's ratio

so that a slightly lower value of F* would be expected (about 2 percent),

m
the agreement generally is within about 1 percent. Thus, for the rod specimen

with W/B = 1.45, an = 0.332, and a1 (straight-sided chevron) the

value of F* is estimated to be 28.9 ± 0.3. The dashed lines in Figure 6

m
show the expected error bounds on F*.
m
Bar Specimens.- Two types of chevron-notched bar specimens have been

studied. In 1978, Barker [6, 15) proposed a rectangular cross-sectioned

bar specimen with an H/B ratio of 0.435 (see Fig. 1). This specimen was

designed in such a way that the same minimum stress-intensity factor was

obtained as for his rod specimen [5). However, because the early compliance

calibration for the rod specimen was about 8 percent low (see Fig. 6), it was

not clear whether the bar and rod specimens now have the same value. Raju and

Newman [20) analyzed both specimens and found that the compliance calibration

for the rectangular bar specimen was about 3.8 percent lower than the rod

specimen.

In 1980, Munz et ale [7) proposed a square cross-sectioned bar specimen

(H/B = 0.5). They conducted a very extensive experimental compliance calibra-

tion on bar specimens with w/B = 1.5 and 2 for aO ranging from 0.2 to 0.5

and a1 1. From these results, they obtained minimum values of stress-

intensity factors for each configuration considered. Using the assumption

that the change of compliance with crack length in a chevron-notch specimen

was the same as that for a straight-through crack specimen, they obtained an

14
equation that was identical to equation (1) as

F* F(:'_-a:O
= f2 = ~~r (5 )

for For specimens with an ao of about 0.2 and 0.35, the

difference between experimental and analytical (eqn. (5)) minimum normalized

stress-intensity factors was less than 1 percent. For an aO value of about

0.5, the difference was 3 to 3.5 percent. They concluded that equation (5)

should only be used to obtain minimum values because experimental and

analytical values differed greatly at small crack-length-to-width (a) ratios

near a O'

Shannon et ale [19] have developed minimum stress-intensity factor

expressions for chevron-notched bar (square) and rod specimens with a1 =

and aO < 0.5. These expressions were fitted to minimum stress-intensity

factors determined from experimental compliance measurements. For the square-

bar specimen, the w/B ratio was 1.5 or 2 and for the rod specimen, the

w/B ratio was 1.5, 1.75 or 2.

The use of chevron-notched specimens with materials that have a rising

crack-growth resistance curve may require stress-intensity factors as a

function of crack length instead of using only the minimum value. Recently,

Shannon et ale [22] have developed polynomial expressions that give the

stress-intensity factors and load-line displacements as a function of crack

length for square-bar and rod specimens (a1 = 1). These expressions were

obtained from experimental compliance measurements made for various w/B

ratios. The W/B ratio for the square-bar specimen was, again, 1.5 or 2, and

for the rod specimen was 1.5, 1.75, or 2. The expressions apply to aO

between 0.2 and 0.4, and a varying from no to 0.8. Some of these results

will be compared with the results from the ASTM analytical round robin in the

next section.

15
 Analytical Round Robin on Chevron-Notched Rod and Bar Specimens

In 1981, plans were formulated for a cooperative test and analysis

program on chevron-notched square-bar and round-rod specimens by an ASTM task

group on Chevron-Notched Specimen Testing. Four configurations were selected:

the square and round versions of a relatively short specimen (w/B = 1.45); and

the square and round versions of a longer specimen (w/B = 2). These configu-

rations were chosen so as to include as many features as possible of prior

work [5-7]. The coordinate system used to define the specimens is shown in

Figure 5. The specimens were loaded by a knife-edge loading fixture that

results in an applied load, P, at the load line L in Figure 5(a). Specimens

had either a square cutout [7] at the load line or a V-cutout [15] at the load

line (not shown). The chevron notch, Figure 5(b), had straight sides and

intersected the specimen sides at x = w (or a1 = 1). The following table

lists the dimensions of the specimens considered:

Specimen W/B H/B

Bar 1.45 0.332 0.5

Bar 2 0.2 0.5

Rod 1 .45 0.332 0.5

Rod 2 0.2 0.5

The analysts were asked to calculate results for crack-length-to-width

(a/w) ratios of 0.4, 0.5, 0.55, 0.6, and 0.7. The information required from

the analyses were:

(1) K-distribution as a function of z and a/w (see Fig. 5(b)).

(2) K-value from the plane-stress compliance relation as a function of

a/w:

16
 1 2
K = [EP dVL l / (6)
b da J
(3) Normalized displacements EVB/P at points Land T (see

Fig. 5(a» as a function of a/we

The participants in the round robin where:

INVESTIGATORS INSTITUTION

A. R. Ingraffea, R. perucchio, Cornell University

T. Y. Han, W. H. Gerstle,
and Y. P. Huang

A. Mendelson and L. J. Ghosn Case-Western

Reserve University

I. S. Raju and J. C. Newman, Jr. NASA Langley

Research Center

The following table lists the investigators, the three-dimensional methodes)

used in the analyses, and the particular configuration(s) analyzed:

ROD BAR
INVESTIGATORS METHOD W/B 1 .45 2 1 .45 2

Ingraffea et ale (21) Finite-element X

Boundary-element X

Mendelson and Boundary-element X

Ghosn [23]

Raju and Newman [20] Finite-element X X X X

All analyses were conducted on models of specimens with the square cutout at

the load line, as shown in Figure 5(a). The slot height (0.03B) shown in

Figure 5(a) was not modeled in any of the analyses (that is, the height was

taken as zero).

Rod Specimen - Ingraffea et ale [21] and Raju and Newman [20] determined

the distribution of normalized stress-intensity factors along the crack front

of a rod specimen (w/B = 1.45) with a = 0.55 using boundary-element and

finite-element methods, respectively. These results are compared in Figure 7.

17
The normalized stress-intensity factor (F*) is plotted against 2z/b. The

center of the specimen is at 2z/b = 0 and the crack intersects the chevron

boundary at 2z/b = 1, see insert. Ingraffea et al. used only one element, a

quarter-point singular element, to define one-half of the crack front length

(b/2)j they showed a nearly linear distribution. On the other hand, Raju and

Newman used five layers of singularity elements to define one-half of the

crack front and they showed nearly constant stress-intensity factors for

2z/b < 0.5. Their stress-intensity factors increased rapidly as 2z/b

approached unity. The results from Raju and Newman were 0 to 16 percent

higher than the results from Ingraffea et al. The difference is probably due

to Ingraffea et al. using only one element along the crack front.

A comparison of experimental and analytical load-point displacements

for the short chevron-notched rod (w/B = 1.45) is shown in Figure 8. The

normalized displacement, EBVL/P, is plotted against a/w. Load-point dis-

placements (V L ) were either measured or calculated at z = 0 (see Fig. 5(b»

as a function of crack length. Because the experiments and analyses were

conducted on materials with different Poisson ratios, thP. displacements have

been adjusted, using results from Raju and Newman [20] on the Poisson effect,

to displacements for a Poisson ratio of 0.3. Barker [15] measured load-point

displacements on fused quartz (v = 0.17) using a laser-interferometric

technique. His displacements have been reduced by 3 percent to compensate

for the differences in Poisson ratios; his data are shown as circular

symbols. In contrast, Shannon et al. [22] measured displacements (V T ) at the

top of aluminum alloy (v = 0.3) specimens (see Fig. 5(a». They measured

displacements for specimens with various values of ~O (0.2 ( ~O ( 0.4) and

with w/B equal to 1.5, 1.75, and 2. The results (square symbols) plotted in

Figure 8 were interpolated to ~ = 0.332 and extrapolated to w/B = 1.45,

18
respectively, using RAcond degree polynomials. These results agreed well with

Barker's results.

Load-point displacements from Raju and Newman's finite-element analysis

[20] and Ingraffea et al.'s [21] boundary-element analysis are also shown in

Figure 8. The displacements from Ingraffea et ala have been reduced by

1 percent to compensate for a slight difference in Poisson's ratio. Both

analytical results were from 4 to 6 percent below the experimental results.

Based on beam theory [24], however, about 2 percent of this difference is

caused by neglecting the notch (0.03B) made by a saw blade or chevron cutter

(see Fig. 5(a». These displacements were used by each investigator to

determine the stress-intensity factors from the plane-stress compliance

method. These results are described in the following section.

Experimental and analytical normalized stress-intensity factors (F*), as

functions of a/w, for the chevron-notched rod are compared in Figure 9. (Note

the use of a broken scale.) The experimental and analytical results were

obtained from the plane-stress compliance relation (eqn. (6» as

1 2
F*
1
= Bf; [EP dvL / =
(7)
c P b da]

where C* is the normalized compliance, EBVL/P. The load-point displacement

(V L ) was either measured or calculated at z = 0 as a function of crack

length. Barker [15] measured the load-point displacements on fused quartz

(v = 0.17) using a laser-interferometric technique. The displacements were

then fitted to an empirical equation in terms of crack length. This equation

was differentiated to obtain the compliance derivative. Barker's results are

shown as circular symbols. Shannon et ale [22] measured displacements (V ) at

T
the top of aluminum alloy (v = 0.3) specimens. They assumed that dVT/da was

19
equal to dVL/da to obtain stress-intensity factors. Again, these results

were interpolated and extrapolated to ~ = 0.332 and w/B = 1.45 using second

degree polynomials. Shannon's results (square symbols) are a few percent higher

than Barker's results. As previously mentioned, Raju and Newman [20] have shown

by a three-dimensional stress analysis that there is a slight difference (about

2 percent) between stress-intensity factors for v = 0.17 and 0.3; these results

agreed with the observed experimental differences.

The analytical results from Raju and Newman [20] and Ingraffea et ale [21]

are also shown in Figure 9. Based on a convergence study [20], the analytical

results are expected to lie about 1.5 percent below the "true" solution. The

analytical results agreed well (within 3 percent) with the experimental results

near the minimum value of F*.

c
Figure 10 compares how analyses and test results (F*) vary with a/w for
c
the chevron-notched rod with w/B = 2. The solid curve represents an equation

proposed by Bubsey et al. [18] for the rod specimens. The equation they used

was equation (5) where F was the normalized stress-intensity factor for a

straight-through crack in the same configuration [18].

Shannon et al.'s [22] results shown in Figure 10 were obtained from

equation {7) using measured load-line displacements (V T ) on the rod

specimen. Their results agreed well (within 1 percent) with the equation from

Bubsey et al., except at small values of ~. From previous work [7], i t was

recognized that equation (5) overestimates values of F*c for values of ~

approaching ~O. The finite-element results of Raju and Newman [20] were

about 2.5 percent below the results from Bubsey et ale and Shannon et ale

Based on all of these results, the value of the minimum normalized stress-

intensity factor (F*) is estimated to be in the range 36.2 ± 0.4.

m
Bar Specimen.- Mendelson and Ghosn [23], using the boundary-element

method, and Raju and Newman [20], using the finite-element method, determined

20
the distribution of boundary-correction factors along the crack front of a bar

specimen with W/B = 2 and a = 0.55. The results are compared in Figure 11.

Here F* is plotted against 2z/b. Mendelson and Ghosn, in contrast to

Ingraffea et al. [21], used five elements to define one-half of the crack front

length. Their elements were assumed to have either linear tractions or linear

displacements. They determined F* values by using either crack-surface dis-

placements or normal stresses near the crack front. For 2z/b < 0.9, their

results were 3 to 16 percent higher than the results from Raju and Newman,

whereas the previous results from Ingraffea et al., using the same (boundary-

element) method (Fig. 7), gave results on a rod specimen that were consistently

lower than the results from reference 20. The reason for the discrepancy

between references 20 and 23 on 'stress-intensity factor distributions is not

clear.

Experimental and analytical load-point displacements at z =0 for the

chevron-notched bar with w/B =2 are compared in Figure 12. Normalized

displacement is plotted against a/w. Shannon et al. [22] measured displace-

ments at the top of aluminum alloy specimens (circular symbols). The solid

curve represents a polynomial equation from reference 22 that was fitted to

the experimental data. The finite-element results from Raju and Newman [20],

v = 0.3, ranged from 3.5 to 6 percent lower than the experimental data • . And

the boundary-element results from Mendelson and Ghosn [23] were 8 to 11 per-

cent lower than the experimental data. (Results from reference 23, v = 1/3,

were increased by 1 percent to compensate for the small difference in

Poisson's ratio from v = 0.3.) Again, these displacements were used by each

investigator to determine the stress-intensity factors from the plane-stress

compliance method (eqn. (7».

The normalized stress:intensity factors (F~), as functions of a/w, for

the bar specimen with w/B = 2 are shown in Figure 13. The experimental

21
results and polynomial equation of Shannon et ale [22] are shown as circular

symbols and solid curve, respectively. The dashed curve shows an equation

proposed by Munz et ale [7] for bar specimens. For the chevron-notched

specimen, Munz et ale used equation (5) where F was the normalized stress-

intensity factor for a straight-through crack in the same configuration [7].

Again, equation (5) overestimates F~ for a approaching aD. But for

larger values of a, the equation underestimates F~ based, at least, on the

present experimental results [22].

The analytical results of Mendelson and Ghosn [23] and Raju and Newman [20]

are also shown in Figure 13. Near the minimum F*c value, the results from

Mendelson and Ghosn were about 1.5 percent lower than the experimental results

but overestimated F* on either side of the minimum. The results from Raju
c
and Newman were about 2.5 percent lower than the experimental results. From

all of the experimental and analytical results, the minimum F* is estimated

m
to be 29.8 ± 0.3.

In Figure 14, experimental and analytical normalized stress-intensity

factors, as functions of a/w, are compared for the bar specimen with

w/B = 1.45. The experimental results from Shannon et ale [22] were, again,

obtained by interpolation and extrapolation to aa = 0.332 and W/B = 1.45

from results obtained from specimens with various aD and w/B ratios.

The solid curve shows the equation proposed by Munz et ale [7]. Near the

minimum F*, the equation agreed well with the experimental results (within
c
1 percent) but, again, overestimated results for a/w ratios less than about

0.55. The analytical results from Raju and Newman [20] were 0 to 1.5 percent

lower than the experimental results. The minimum value from Raju and Newman

was 24.43, from Shannon et ale was 24.85, and from Munz et ale was 24.66.

From these results, the minimum value of F* is estimated to be 24.8 ± 0.3.

m

22
 Chevron-Notched Bend Bars

As previously mentioned, Nakayama [1,2], and Tattersall and Tappin [3]

were the first to introduce and to determine fracture energies from chevron-

notched bend bars. Pook [4] and Bluhm [10] were the first to provide

approximate stress-intensity factors and compliance expressions, respectively,

for these specimens. This section reviews the more recent experimental and

analytical stress-intensity factor solutions that have been proposed for

chevron-notched bend bars.

Munz et ale [25] compared stress-intensity factors for various four-point

bend specimens with 0.12 ( a (0.24, 0.9 C;; a ( 1, and W/B = 1 or 1.25.
O 1
Two analytical methods were studied. The first was by the use of equation (7)

wherein dC*/da, the compliance derivative of the chevron-notched specimen,

was assumed to be equivalent to dC/da, the compliance derivative of a

straight-through crack. Under this assumption, equation (7) reduces to

~quation (5) or Pook's equation [4]. The second method was by using Bluhm's

slice model [10]. Bluhm's slice model is probably more accurate than Pook's

equation, but neither method has been substantiated by experimental compliance

measurements or by more rigorous analytical (three-dimensional elasticity)

methods. The slice model was, however, calibrated to experimental compliance

measurements made on uncracked chevron-notch bend bars. A comparison of the

two analytical results showed that the differences ranged from -5 to 10 per-

cent for the particular configurations considered.

In 1981, Shih [26] proposed a "standard" chevron-notched bend bar

configuration for three-point loading with a major-span-to-width ratio (s/w)

of 4. The w/B ratio was 1.82 with a = 0.3 and a1 = 0.6. Shih [26] used

the KIC value from 7079-T6 aluminum alloy and the failure (maximum) load on

the chevron-notched bend bars to estimate the minimum stress-intensity factor;

this value is shown in Figure 15 as the horizontal dashed line. The equation

23
proposed by Pook [4] (upper solid curve) gave a minimum value very close to

the value determined by Shih. Later, however, Shih [27] re-evaluated the

minimum by testing 7079-T6 aluminum alloy compact specimens and chevron-

notched specimens made from the same plate. The new KIc value dropped by

19 percent from the old value and, consequently, the minimum value (F*)
m
dropped to 10.17, as shown by the dash-dot line in Figure 15.

Wu [28] used equation (5) to determine the stress-intensity factors for

three-point bend chevron-notched specimens. His equation gave essentially the

same results (within 1 percent) as that shown for Pook in Figure 15. Wu [29]

also used Bluhm's slice model to determine specimen compliance and then used

equation (7) to determine F* as a function of ex (or a/w). His equation

c
was used herein to calculate F* in Figure 15. Here the minimum value from
c
Wu's equation was about 4 percent higher than the new minimum values proposed

by Shih [27]. From these results, it is obvious that Pook's equation and

Bluhm's slice model give drastically different values of stress-intensity

factors, and that the determination of minimum values by matching K and

IC

Km must be approached with caution.

EFFECTS OF MATERIAL FRACTURE TOUGHNESS BEHAVIOR

For a brittle material, a material which exhibits a "flat" crack-growth

resistance curve as shown in Figure 4, the use of a chevron-notched specimen

to obtain KIcv is well justified. But what if the material has a "rising"

crack-growth resistance curve as shown in Figure 16? Because most engineering

materials, under non-plane-strain conditions, have rising crack-growth

resistance curves or KR-curves, the answer to this question is of 'utmost

importance. The objective of this paper, however, is not to answer this

question, but to review some of the problems associated with using these

specimens for such materials.

24
 Figure 16 illustrates the application of the ~-curve concept [30] to a

material with a rising KR-curve. The stress-intensity factor is plotted

against crack length. The hypothetical KR-curve (solid curve) begins at the

initial crack length, a O' The dashed curves show the "crack-driving force"

curves for various values of applied load on a chevron-notched specimen

(w = constant). As the load is increased, the crack grows stably into the

material (point A, to B, to C, to D) unti~ the load reaches Pmax • At this

load and crack length, crack growth becomes unstable (point D). As can be

seen, the instability point (tangent point between crack-drive curve and

KR-curve) does not correspond to the minimum K value (solid symbol).

Consequently, the maximum load and minimum K value cannot be used to compute

the stress-intensity factor at failure, although the difference might be

small. But if the specimen width is smaller than that used in Figure 16, then

the instability point would occur at a lower point on the KR-curve.

Conversely, the instability point for a larger width specimen would occur at a

higher point on the KR-curve. Thus, a specimen size (or width) effect exists

and it has been the subject of several papers on chevron-notched specimens

[12, 31-35].

25
 DISCUSSION

Chevron-Notched Test Specimens

Many investigators have shown the advantages of using chevron-notched

specimens for determination of plane-strain fracture toughness of brittle

materials. The following table summarizes some of the advantages and

disadvantages of these specimens:

ADVANTAGES DISADVANTAGES

• SMALL SPECIMENS • RESTRICTED TO "BRITTLE" MATERIALS

o NO FATIGUE PRECRACKING • MATERIAL THICKNESS LIMITATIONS

o SIMPLE TEST PROCEDURE • NOTCH MACHINING DIFFICULTY

• MAXIMUM LOAD TEST

• SCREENING TEST

• NOTCH GUIDES CRACK PATH

• HIGH CONSTRAINT AT CRACK FRONT

The chevron-notched specimens can be small because their width and height are

of nearly the same size as their thickness (5 to 25 mm), so only a small

amount of material is needed. Consequently, they are very useful as quality

control specimens. They may be useful in alloy development programs where

small amounts of material are produced. They can also be used to determine

toughness profiles through the thicknesses of large plates. Because they

require no fatigue precracking, they cost less than current fracture toughness

specimens. For brittle materials, the test procedure is very simple; once the

minimum stress-intensity factor has been obtained, it is only necessary to

record the maximum failure load to calculate fracture toughness. Even for

ductile materials, the specimens may be used in screening tests to rank

materials.

The chevron notch tends to guide the crack path and, therefore, these

specimens can be used to test particular regions of a material, such as

26
heat-affected zones. The notch also constrains the crack front, which helps

.set up an approximate plane-strain condition around the crack front.

The major disadvantage in using chevron-notched specimens with the

maximum load test procedure - for plane-strain fracture toughness testing - is

that they are restricted to brittle materials, such as ceramics, rocks, high-

strength metals, and other low toughness materials. Further studies are

needed on more ductile materials to see if these specimens can be used for

fracture toughness evaluation. They are also limited in the thickness that

can be tested. Thin materials, less than about 5 mm, cannot be easily tested.

Stress-Intensity Factors

Several methods have been used to determine stress-intensity factors and

minimum stress-intensity factors for these specimens. In the first method,

the minimum value was obtained by matching ~ to KIc from ASTM E399

standard specimens. For the "short" rod specimen, the minimum value obtained

from KIc-matching [5,11,12] was about 8 percent below several experimental

compliance calibrations and two recent three-dimensional elasticity solutions.

In more recent applications of the KIc-matching procedure [26,27], the minimum

values for a three-point bend specimen differed by about 20 percent. Thus,

the KIc-matching procedure should be used with caution.

The second method is derived from the assumption that the change in

compliance with crack length of the chevron-notch specimen is equal to the

change in compliance of a straight-through crack specimen. The stress-intensity

factors derived from this method match those from Pook's equation [4]. For the

rod and bar specimens, researchers have shown that this method gives accurate

values of minimum stress-intensity factors, but is unreliable on either side of

the minimum. In contrast, this method gave very large differences on a three-

point bend specimen. Again, this method must be used with caution.

27
 The third, a more refined approximate method for chevron-notched

specimens, is the slice model proposed by Bluhm [10]. This model has been

used extensively on three- and four-point (chevron-notched) bend specimens.

Munz et ale [7] has used this model on chevron-notched bar specimens. The

problem associated with this method is the "shear-correction" parameter (k)

that must be determined from experimental compliance measurements. If the

shear-correction parameter, k, is determined experimentally from uncracked

chevron-notched specimens close to the desired configuration, then this method

will probably give reliable results. But a systematic study to evaluate the

accuracy of stress-intensity factors computed from the slice model has not

been undertaken.

The fourth method is three-dimensional elasticity solutions, such as

finite-element and boundary-integral equation methods. These methods can give

accurate stress-intensity factors if care is taken especially in conducting

convergence studies. These methods, however, tend to be expensive if a large

number of solutions are desired.

The last method is experimental compliance calibration. This method can

also give accurate stress-intensity factors if the tests are done carefully.

But the method is limited to the particular specimen configurations studied.

Coupled with Bluhm's slice model, this method may provide a reliable and

inexpensive way of obtaining stress-intensity factors for a wide range of

configuration parameters.

A summary of the consensus minimum normalized stress-intensity factor,

F~, for the four configurations considered in the analytical round robin and

for the rectangular bar specimen [6, 15, 20] are shown in the following table.

28
 Specimen wlB aO a, HIB F*m

Bar 1.45 0.332 0.435 27.8 ± 0.3

Bar , .45 0.332 0.5 24.8 ± 0.3
Bar 2 0.2 0.5 29.8 ± 0.3
Rod , .45 0.332 0.5 28.9 ± 0.3
Rod 2 0.2 0.5 36.2 ± 0.4

The stress-intensity factor solutions for three- and four-point bend

chevron-notched specimens have only been obtained from the KIc-matching

procedure, Pook's equation, and Bluhm's slice model. Of these, the slice

model is probably the most reliable. However, it is recommended that a

detailed finite-element or boundary-element analysis, or careful experimental

compliance calibrations, be performed on various chevron-notched bend bar

configurations.

CONCLUSIONS

The historical development of chevron-notched fracture specimens and the

stress-intensity solutions that have been proposed for these specimens was

reviewed. The review covered the three- and four-point bend bars as well as

the "short" rod and bar specimens. The stress-intensity factor solutions and

minimum stress-intensity value for these specimens had been obtained by using

several different methods, either experimental or analytical. Results of a

recent ASTM analytical round robin on the rod and bar specimens were

summarized. Some problems associated with using these specimens for materials

with rising crack-growth resistance curves were discussed. Based on this

review, the following conclusions were drawn:

29
1. For the chevron-notched round-rod and bar specimens, the experimental

compliance calibrations and the analytical (finite-element,

boundary-element, and some approximate methods) calculations agreed

within 3 percent. When the lower bound convergence of the finite-

element and boundary-element techniques were accounted for, the

agreement was generally within about 1 percent.

2. Chevron-notched bend bars need further experimental and analytical

stress-intensity factor calibrations. Although some recent stress-

intensity factor solutions agreed within 5 percent, they were

obtained from methods which have not been adequately substantiated.

3. Further studies are needed on using chevron-notched specimens with

materials that exhibit a "rising" crack-growth resistance curve

behavior.

30
 REFERENCES

[1] Nakayama, J.: A Bending Method for Direct Measurement of Fracture

Energy of Brittle Material, Japan Journal of Applied Physics, Vol. 3,

1964, pp. 422-423.

[2] Nakayama, J.: Direct Measurement of Fracture Energies of Brittle

Heterogeneous Materials, Journal of the American Ceramic Society,

pp. 103-108.

[5] Barker, L. M.: A Simplified Method for Measuring plane Strain Fracture

Toughness, Engineering Fracture Mechanics, Vol. 9, 1977, pp. 361-369.

[6] Barker, L. M.: Short Bar Specimens for KIC Measurements, Fracture

Mechanics Applied to Brittle Materials, ASTM STP 678, American Society

for Testing and Materials, philadelphia, PA, 1979, pp. 73-82.

(Proceedings Eleventh National Symposium Fracture Mechanics,

Blacksburg, VA, June 1978.)

[7] Munz, D.; Bubsey, R. T.; and Srawley, J. E.: Compliance and Stress-

Intensity Coefficients for Short Bar Specimens with Chevron Notches,

International Journal of Fracture, Vol. 16, No.4, 1980, pp. 359-374.

[8] Brown, W. F. and srawley, J. E.: Plane strain Crack Toughness Testing

of High Strength Metallic Materials, ASTM STP 410, American Society

for Testing and Materials, philadelphia, 1966.

31
 [9] Freed, C. N. and Kraft, J. M.: Effect of Side Grooving on Measurements

of Plane-Strain Fracture Toughness, Journal of Materials, Vol. 1,

No.4, 1966, pp. 770-790.

[10J Bluhm, J. I.: Slice Synthesis of a Three-Dimensional "Work of Fracture"

Specimen, Engineering Fracture Mechanics, Vol. 7, 1975, pp. 593-604.

[11] Barker, L. M.: Theory for Determining KIC from Small, Non-LEFM

Specimens, Supported by Experiments on Aluminum, International Journal

of Fracture, Vol. 15, NO.6, 1979, pp. 515-536.

[12] Barker, L. M. and Baratta, F. I.: Comparisons of Fracture Toughness

Measurements by the Short Rod and ASTM standard Method of Test for

Plane-Strain Fracture Toughness of Metallic Materials (E399-78),

Journal of Testing and Evaluation, Vol. 8, No.3, 1980, pp. 97-102.

[13] Barker, L. M. and Guest, R. V.: Compliance Calibration of the Short Rod

Fracture Toughness Specimen, Terra Tek Report TR 78-20, April 1978.

[14] Barker, L. M.: Discussion of "Compliance Calibration of the Short Rod

Chevron-Notch Specimen for Fracture Toughness Testing of Brittle

. Materials", by Bubsey, R. T.; Munz, D.; Pierce, W. S.; and Shannon,

J. L., Jr., International Journal of Fracture, Vol. 19, 1982,

pp. R3-R5.

[15] Barker, L. M.: Compliance Calibration of a Family of Short Rod and

Short Bar Fracture Toughness Specimens, Engineering Fracture

Mechanics, Vol. 17, No.4, 1983, pp. 289-312.

[16] Beech, J. F. and Ingraffea, A. R.: Three-Dimensional Finite Element

Calibration of the Short-Rod Specimen, Geotechnical Engineering Report

80-3, Cornell University, 1980.

[17] Beech, J. F. and Ingraffea, A. R.: Three-Dimensional Finite Element

Calibration of the short-Rod Specimen, International Journal of

Fracture, Vol. 18, No.3, 1982, pp. 217-229.

32
[18] Bubsey, R. T.i Munz, D.i Pierce, W. S. and Shannon, J. L., Jr.:

Compliance of the Short Rod Chevron-Notch Specimen for Fracture

Toughness Testing of Brittle Materials, International Journal of

Fracture, Vol. 18, No.2, 1982, pp. 125-133.

[19] Shannon, J. L., Jr.i Bubsey, R. T.i pierce, W. S.i and Munz, D.:

Extended Range stress Intensity Factor Expressions for Chevron-Notched

Short Bar and Short Rod Fracture Toughness Specimens, International

Journal of Fracture, Vol. 19, 1982, pp. R55-R58.

[20] Raju, I. S. and Newman, J. C., Jr.: Three-Dimensional Finite-Element

Analysis of the Chevron-Notched Fracture Specimens, symposium on

Chevron-Notched Specimen: Testing and Stress Analysis, April 21,

1983, Louisville, KY.

[21] Ingraffea, A. R.i perucchio, R.i Han, T. Y.i Gerstle, W. H. and Huang,

Y. P.: Three-Dimensional Finite and Boundary Element Calibration of

the Short-Rod Specimen, Symposium on Chevron-Notched Specimens:

Testing and Stress Analysis, April 21, 1983, Louisville, KY.

[22] Shannon, J. L., Jr.i Bubsey, R. T. and Pierce, W. S.: Closed-Form

Expressions for crack-Mouth Displacements and stress-Intensity Factors

for Chevron-Notched Short Bar and Rod Specimens Based on Experimental

Compliance Measurements, NASA Lewis Research Center, NASA TM 83796,

1984.

[23] Mendelson, A. and Ghosen, L. J.: Three-Dimensional Analysis of Short

Bar Chevron-Notched Specimens by Boundary Integral Method, Symposium

on Chevron-Notched Specimens: Testing and Stress Analysis, April 21,

1983, Louisville, KY.

[24] Timoshenko, S. and Goodier, J. N.: Theory of Elasticity, second

edition, McGraw-Hill Book Co., New York, 1951.

33
[25] Munz, D. G.; Shannon, J. L., Jr. and Bubsey, R. T.: Fracture Toughness

Calculation from Maximum Load in Four Point Bend Tests of Chevron-

Notch Specimens, International Journal of Fracture, Vol. 16, 1980,

R137-R14l.

[26] Shih, T. T.: Chevron V-Notched Bend Specimen for KIC Measurement of

Brittle Materials, Journal of Testing and Evaluation, Vol. 9, No.1,

1981, pp. 50-55.

[27] Shih, T. T.: An Evaluation of the Chevron V-Notched Bend Bar Fracture

Toughness Specimen, Engineering Fracture Mechanics, Vol. 14, No.4,

1981, pp. 821-832.

[28] Wu, Shang-Xian: An Investigation of Chevron-Notch Three-Point Bend

Fracture Toughness Specimen, International Journal of Fracture,

Vol. 19, 1982, pp. R27-R30.

[29] Wu, Shang-Xian: Compliance and Stress-Intensity Factor of Chevron-

Notched Three-Point Bend Specimen, symposium on Chevron-Notched

Specimens: Testing and Stress Analysis, April 21, 1983,

Louisville, KY.

[30] Fracture Toughness Evaluation by R-Curve Method, ASTM STP 527, ed. by

D. E. MCCabe, American Society for Testing and Materials,

Philadelphia, PA, 1973.

[31] Munz, D.; Bubsey, R. T. and Shannon, J. L., Jr.: Performance of

Chevron-Notch Short Bar Specimen in Determining the Fracture Toughness

of Silicon Nitride and Aluminum oxide, Journal of Testing and

Evaluation, Vol. 8, No.3, 1980, pp. 103-107.

[32] Munz, D.: Determination of Fracture Toughness of High strength Aluminum

Alloys with Chevron-Notched Short Rod and Short Bar Specimens,

Engineering Fracture Mechanics, Vol. 15, No. 1-2, 1981, pp. 231-236.

34
[33] Munz, D.; Himsolt, G.; and Eschweiler, J.: Effect of Stable Crack

Growth on Fracture Toughness Determination for Hot-Pressed Silicon

Nitride at Elevated Temperatures, Fracture Mechanics Methods for

Ceramics, Rocks, and Concrete, ASTM STP 745, S. W. Freiman and

E. R. Fuller, Eds., American Society for Testing and Materials, 1981,

pp. 69-84.

[34] Barker, L. M.: Specimen Size Effects in Short Rod Fracture Toughness

Measurements, Symposium on Chevron-Notched Specimen: Testing and

Stress Analysis, April 21, 1983, Louisville, KY.

[35] Shannon, J. L., Jr. and Munz, D. G.: Specimen size and Geometry

Effects on Fracture Toughness of A1 2 0 3 Measured with Short Rod

and Short Bar Chevron-Notched Specimens, symposium on Chevron-

Notched Specimens: Testing and Stress Analysis, April 21, 1983,

Louisville, KY.

35
 Table 1.- Chronological development of m1n1mum normalized stress-intensity
factors (F*) for short chevron-notched rod
m

Investigator( s) Year Ref. w/B aO/w a 1 /w F*

m

Barker a 1977 5 1.45 0.31 b 0.96 b 26.3

Barker and GuestC 1978 13,14 1 .474 0.343 d 0.992 d 29.6

Barker 1979 11 1.45 0.31 b 0.96 b 25.1

Barker and Barattaa 1980 12 1 .45 0.343 d 0.992 d 26.5

Beech and Ingraffea e 1980 16 1.5 0.35 1.0 31.4(30.0)f

Beech and Ingraffeae 1982 17 1 .5 0.35 1 .0 32.7(31.2)f

Bubsey et al. c 1982 18 1.45 0.332 1.0 29.0

Barker c 1983 15 1.45 0.332 1 .0 28.2

w
en
Shannon et aLc 1983 22 1.45 0.332 1.0 29.1

Raju and Newmane 1983 20 1 .45 0.332 1 .0 28.4

Ingraffea et al.e,g 1983 21 1.45 0.332 1.0 28.3

aF~ determined from matching KIc from ASTM E399 specimens.

bDimensions estimated from photograph.
cF~ determined from experimental compliance.
dCurved-sided chevron notch used in test, equivalent dimensions for straight-sided
chevron notch used for table.
eF~ determined from finite-element analysis and compliance.
fValues are from plane-strain (plane-stress) assumption.
gF~ determined from boundary-element analysis and compliance.
Nakayama (1964) Tattersall and Tappin
(1966)

BEND BAR BEND BAR

Barker (1978)

BAR

Figure 1.- Various chevron-notched fracture specimen configurations.

 FACE PARAMETERS:
OF a = o/w
BEND BAR"
OR aO= °O/w
LOAD a 1 = °l /w
W
DO
LINE w/B Ratio

==_0==:-1
I:: w _I

Figure 2.- Chevron-notched fracture specimen nomenclature ..

 K B~ ~all /
/
P
II l;;ti /
W
\.0 l'L-------: /~ T
B
~a--; ~

-- ------ ------ ---/

t=0--.i vi
,1
,1

00 01
w W

Crock-length-to-width ratioJ o/w

Figure 3.- Comparison of normalized stress-intensity factors for chevron-notched

and straight-through crack specimens.
 Pmax
KIev = BVW F*
m

o
Crack length" a

Figure 4.- Fracture of "brittle" material using a chevron-notched specimen.

 y

iT

T p
2H P
L O.03B
!
~===:::::E============================::;=:~ -_. x

ll~:- a ~_t----+l
(a) z = 0 plane.

TB - -•• x
1- O.IB
L l-.l-.---_---=:::::..J

z
(b) y = 0 plane.

Figure 5.- Coordinate system used to define dimensions of

knife-edge loaded chevron-notched rod and bar
specimens.

41
 32
Rod
w/B = 1.45
Beech and Ingraffea
30 Barker and Guest [13J .
=l~~~~= =~ )l_Sha~non eti~'p:~:~nt
/ Raj u and Newman [20J
28
Bubsey et al, [18J Ingraffea et aI, [21J
F*m Barker [15J
)D ~
26 Barker [5J Barker and Baratta [12]
J
Barker [11J o Match K1c
24 Q Experimental compliance
8 Finite-element compliance
v Boundary-element compliance

75 77 79 81 83 85
Year

Figure 6.- Comparison of minimum normalized stress-intensity factor for

chevron-notched rod.
 40
Rod
w/B = 1.45
aO = 0.332 Raju and Newman [20]
a1 =1 (Finite-element method)
30 a = 0.55
--;...-----
--\-
--- -- --Ingraffea et al, [21]
F* 20 (Boundary-element method)

~
w

p -x
K= - F*
10 BfW O,lB

I
z

o0 .2 .4 .6 .8 1
2z
-
b

Figure 7.- Comparison of distribution of normalized stress-intensity factors

along crack front for short chevron-notched rod.
 200
Rod
w/B = 1.45 II
a = 0.332 <:>
O Barker [15] ~
a =1
1 (Experimental)
150 v = 0.3
<:>

J..-... Raj u and Newman [20]

Shannon et 01. [22] -fI)
(Finite-element)
(Experimental) •
100 <:>
( .

~ Ingraffea et al. [21J

~ (Boundary~element)
50
....- .~

o L- -.J. ..L- ---iL...- --'- ----l

o .2 .4 .6 .8 1
a
w

Figure 8.- Comparison of experimental and analytical load-point displacements for

short chevron-notched rod.

•
 •

40 Rod •
w/B = 1.45
etO = 0.332
et 1 = 1
35 • • Ingraffea et al, [21J
~ ~ (Boundary-element)
Shannon et a1. [22J ---:. ~
F*c 30 (Experimental) I mi" Raj u and Newman [20J
~~ (Finite-element)

Barker [15J
(Experimental)
25

a
w

Figure 9.- Comparison of experimental and analytical normalized stres-s-intensity

factors for short chevron~notched rod.
 45
Rod
w/B = 2
aO = 0.2
a1 = 1
40
Bubsey et al, [18J
(Equation)
Shannon et al,
(Experimental)
F*c 35
! B B
Raju and Newman [20]
(Finite-element)

30

.2 .4 .6 .8 1
a
w

Figure 10.- Comparison of experimental and analytical normalized stress-intensity

factors for long chevron-notched rod.
 •

50

40
Mendelson and Ghosn [23J
(Boundary-element) .
./'
Displacement __ ~: ~.::::~-""".--.

_.-
- _. _0--l--
. . --- --- -~_.-. -=~ .
30
Stress Nodal-force
F*
Raju and Newman [20J
20 (Finite-element)
Bar
w/B = 2
a
O= 0.2
10 a =1
1
a = 0,55

Oa...-..-----'----.-..------.Ao----------
a .2 .4 .6 ,8 1
2z
b

Figure 11.- Comparison of distribution of normalized stress-intensity factors

along crack front for long chevron-notched bar.
 200
Bar
w/B = 2
Ct
a = 0,2
Ct
1 =1
150 v = 0.3 Shannon et 01. [22]
(Equation)

EBV L 100 Shannon et al. [22]

p (Experimental) Raju and Newman [20]
(Finite-element)

50
Mendelson and Ghosn [23J
(Boundary-element)

.2 .4 .6 ,8 1
a
w

Figure 12.- Comparison of experimental and analytical load-point displacements

for long chevron-notched bar.

•
•
 • of •

40
Bar
w/B = 2
oro = 0,2 \
Otl = 1 \
35 \
\ LMunzetal, [7]
Shannon
\ / (Equation)
\A /
et a1. [22J
---"!~---:
)zf
(Eauation)
F*c 30
Shannon et aI,
tl Mendelson and Ghosn [23J
[22]
(Boundary-element)
(Experimental)
.p.
\D
Raju and Newman [20J
(Finite-element)

,2 ,4 ,6 o
,0 1
a
w
Figure 13.- Comparison of experimental and analytical normalized stress-intensity
factors for long chevron-notched bar.
 40
Bar
w/B = 1.45
CiO = 0.332
Cil = 1
35 Munz et 01. [7J
(Equation)

F*c 30
Shannon et aI.
(}l (Experimental)
o
Raju and Newman [20J
(Finfte-element)

.2 .4 .6 ,8 1
a
w

Figure 14.- Comparison of experimental and analytical normalized stress-intensity

factors for short chevron-notched bar •

•
 • •

30
Bend Bar (s/w = 4)
w/B = 1.82
CiO = 0.3
Ci1= 0.6

Wu [28]
20 (Equation) Pook [4]
(Equation)

WU [29]
(Equation)
Shih [26]
-._._._._._._.- Shih [27]
10
(Match K1c )

.2 .4 .6 .8 1
a
w

Figure 15.- Comparison of normalized stress-intensity factors for chevron-notched

three-point bend bar.
 \
\
\
w = constant \\
\\
\\\
\ \ \ Minimum K
\ \ ',~ P3
K \" D /
\ " _//
U'I
N \ C / P2
........ /'
--~

o
Crack lengthJ a

Figure 16.- Fracture of "rising KR-curve" material using a chevron-notched specimen .

•
 1. Report No.
NASA TM-85797
I 2. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle 5. Report Dilte

A REVIEW OF CHEVRON-NOTCHED FRACTURE SPECIMENS September 1984
6. Performing O~nization Code
505-33-23-02
7. Author(s) 8. Performing Organization Report No.

J. C. Newman, Jr.
~-----------------------------l
10. Work Unit No.
9. Performing Organization Name and Address

NASA Langley Research Center 11. Contract or Grant No.

Hampton, VA 23665
13. Type of Report and Period Covered
12. Sponsoring Agency Name and Address Technical Memorandum
National Aeronautics and Space Administration 14. Sponsoring Agency Code
Washington, DC 20546
15. Supplementary Notes

16. Abstract
This paper reviews the historical development of chevron-notched fracture speci-
mens; it also compares stress-intensity factors and load line displacement solutions
that have been proposed for some of these specimens. The review covers the original
bend-bar confi gurati ons up to the present day "short" rod and bar specimens. In
particular, the results of a recent "analytical" round robin that was conducted by an
ASTM Task Group on Chevron-Notched Specimens are presented.
In the round robin, three institutions calculated stress-intensity factors for
either the chevron-notched round-rod or square-bar specimens. These analytical solu-
tions were compared among themselves, and then among the various experimental solu-
tions that have been proposed for these specimens. The experimental and analytical
stress-intensity factor solutions that were obtained from the compliance method
agreed within 3 percent for both specimens. An assessment of the consensus stress-
intensity factor (compliance) solution for these specimens is made.
The stress-intensity factor solutions proposed for three- and four-point bend
chevron-notched specimens are also reviewed. On the basis of this review, the bend-
bar configurations need further experimental and analytical calibrations.

17. Key Words (Suggested by Author(s)) 18. Oistribution Statement •

Fracture mechanics, stress-intensity factor,
cracks, finite-element method, boundary- Unclassifed - Unlimited
element method, crack-opening displacement, Subject Category 39 of

chevron-notch specimen

19. Se<:urity Classif. (01 this report) 20. Security Ciani I. (of this pagel 21, No. of PagM 22. Price
Unclassified Unclassified 53 A04

N-JC5 For sale by the National Technical Information Service, Springfield, Virginia 22161