This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles

for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.

Designation: E1921 − 19

Standard Test Method for

Determination of Reference Temperature, To, for Ferritic
Steels in the Transition Range1
This standard is issued under the fixed designation E1921; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.

1. Scope recommended that the specimen type be reported along with

1.1 This test method covers the determination of a reference
temperature, To, which characterizes the fracture toughness of
ferritic steels that experience onset of cleavage cracking at
elastic, or elastic-plastic KJc instabilities, or both. The specific
types of ferritic steels (3.2.1) covered are those with yield
strengths ranging from 275 to 825 MPa (40 to 120 ksi) and
weld metals, after stress-relief annealing, that have 10 % or
less strength mismatch relative to that of the base metal.
1.2 The specimens covered are fatigue precracked single-
edge notched bend bars, SE(B), and standard or disk-shaped
compact tension specimens, C(T) or DC(T). A range of
specimen sizes with proportional dimensions is recommended.
The dimension on which the proportionality is based is
specimen thickness.
1.3 Median KJc values tend to vary with the specimen type
at a given test temperature, presumably due to constraint
differences among the allowable test specimens in 1.2. The
degree of KJc variability among specimen types is analytically
predicted to be a function of the material flow properties (1)2
and decreases with increasing strain hardening capacity for a
given yield strength material. This KJc dependency ultimately
leads to discrepancies in calculated To values as a function of
specimen type for the same material. To values obtained from
C(T) specimens are expected to be higher than To values
obtained from SE(B) specimens. Best estimate comparisons of
several materials indicate that the average difference between
C(T) and SE(B)-derived To values is approximately 10°C (2).
C(T) and SE(B) To differences up to 15°C have also been
recorded (3). However, comparisons of individual, small data-
sets may not necessarily reveal this average trend. Datasets
which contain both C(T) and SE(B) specimens may generate
To results which fall between the To values calculated using
solely C(T) or SE(B) specimens. It is therefore strongly
1
This test method is under the jurisdiction of ASTM Committee E08 on Fatigue
and Fracture and is the direct responsibility of E08.07 on Fracture Mechanics.
Current edition approved Feb. 1, 2019. Published February 2019. Originally
approved in 1997. Last previous edition approved in 2018 as E1921 – 18a. DOI:
10.1520/E1921-19.
2
The boldface numbers in parentheses refer to the list of references at the end of
this standard.
the derived To value in all reporting, analysis, and discussion of
results. This recommended reporting is in addition to the
requirements in 11.1.1.
1.4 Requirements are set on specimen size and the number
of replicate tests that are needed to establish acceptable
characterization of KJc data populations.
1.5 To is dependent on loading rate. To is evaluated for a
quasi-static loading rate range with 0.1< dK/dt < 2 MPa√m/s.
Slowly loaded specimens (dK/dt < 0.1 MPa√m) can be
analyzed if environmental effects are known to be negligible.
Provision is also made for higher loading rates (dK/dt > 2
MPa√m/s) in Annex A1. Note that this threshold loading rate
for application of Annex A1 is a much lower threshold than is
required in other fracture toughness test methods such as E399
and E1820.
1.6 The statistical effects of specimen size on KJc in the
transition range are treated using the weakest-link theory (4)
applied to a three-parameter Weibull distribution of fracture
toughness values. A limit on KJc values, relative to the
specimen size, is specified to ensure high constraint conditions
along the crack front at fracture. For some materials, particu-
larly those with low strain hardening, this limit may not be
sufficient to ensure that a single-parameter (KJc) adequately
describes the crack-front deformation state (5).
1.7 Statistical methods are employed to predict the transi-
tion toughness curve and specified tolerance bounds for 1T
specimens of the material tested. The standard deviation of the
data distribution is a function of Weibull slope and median KJc.
The procedure for applying this information to the establish-
ment of transition temperature shift determinations and the
establishment of tolerance limits is prescribed.
1.8 The procedures described in this test method assume
that the data set represents a macroscopically homogeneous
material, such that the test material has uniform tensile and
toughness properties. Application of this test method to an
inhomogeneous material will result in an inaccurate estimate of
the transition reference value To and nonconservative confi-
dence bounds. For example, multi-pass weldments can create
heat-affected and brittle zones with localized properties that are

Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
1
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
quite different from either the bulk or weld materials. Thick-
section steels also often exhibit some variation in properties
near the surfaces. Metallography and initial screening may be
necessary to verify the applicability of these and similarly
graded materials. Section 10.6 provides a screening criterion to
assess whether the data set may not be representative of a
macroscopically homogeneous material, and therefore, may
not be amenable to the statistical analysis procedures employed
in this test method. If the data set fails the screening criterion
in 10.6, the homogeneity of the material and its fracture
toughness can be more accurately assessed using the analysis
methods described in Appendix X5.
1.9 This standard does not purport to address all of the
safety concerns, if any, associated with its use. It is the
responsibility of the user of this standard to establish appro-
priate safety, health, and environmental practices and deter-
mine the applicability of regulatory limitations prior to use.
1.10 This international standard was developed in accor-
dance with internationally recognized principles on standard-
ization established in the Decision on Principles for the
Development of International Standards, Guides and Recom-
mendations issued by the World Trade Organization Technical
Barriers to Trade (TBT) Committee.
2. Referenced Documents
2.1 ASTM Standards:3
E4 Practices for Force Verification of Testing Machines
E8/E8M Test Methods for Tension Testing of Metallic Ma-
terials
E23 Test Methods for Notched Bar Impact Testing of Me-
tallic Materials
E74 Practices for Calibration and Verification for Force-
Measuring Instruments
E111 Test Method for Young’s Modulus, Tangent Modulus,
and Chord Modulus
E177 Practice for Use of the Terms Precision and Bias in
ASTM Test Methods
E208 Test Method for Conducting Drop-Weight Test to
Determine Nil-Ductility Transition Temperature of Fer-
ritic Steels
E399 Test Method for Linear-Elastic Plan-Strain Fracture
Toughness KIc of Metallic Materials
E436 Test Method for Drop-Weight Tear Tests of Ferritic
Steels
E561 Test Method for KR Curve Determination
E691 Practice for Conducting an Interlaboratory Study to
Determine the Precision of a Test Method
E1820 Test Method for Measurement of Fracture Toughness
E1823 Terminology Relating to Fatigue and Fracture Testing
2.2 ASME Standards:4
ASME Boiler and Pressure Vessel Code, Section II, Part D
3
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM
Standards volume information, refer to the standard’s Document Summary page on
the ASTM website.
4
Available from American Society of Mechanical Engineers (ASME), ASME
International Headquarters, Two Park Ave., New York, NY 10016-5990, http://
www.asme.org.
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
2
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
3. Terminology
3.1 Terminology given in Terminology E1823 is applicable
to this test method.
3.2 Definitions:
3.2.1 ferritic steels—typically carbon, low-alloy, and higher
alloy grades. Typical microstructures are bainite, tempered
bainite, tempered martensite, and ferrite and pearlite. All
ferritic steels have body centered cubic crystal structures that
display ductile-to-cleavage transition temperature fracture
toughness characteristics. See also Test Methods E23, E208
and E436.
3.2.1.1 Discussion—This definition is not intended to imply
that all of the many possible types of ferritic steels have been
verified as being amenable to analysis by this test method.
3.2.2 stress-intensity factor, K [FL– 3/2]—the magnitude of
the mathematically ideal crack-tip stress field coefficient (stress
field singularity) for a particular mode of crack-tip region
deformation in a homogeneous body.
3.2.2.1 Discussion—In this test method, Mode I is assumed.
See Terminology E1823 for further discussion.
3.2.3 J-integral, J [FL–1]—a mathematical expression; a
line or surface integral that encloses the crack front from one
crack surface to the other; used to characterize the local
stress-strain field around the crack front (6). See Terminology
E1823 for further discussion.
3.3 Definitions of Terms Specific to This Standard:
3.3.1 control force, Pm [F]—a calculated value of maximum
force, used in 7.8.1 to stipulate allowable precracking limits.
3.3.2 crack initiation—describes the onset of crack propa-
gation from a preexisting macroscopic crack created in the
specimen by a stipulated procedure.
3.3.3 effective modulus, Eeff [FL–2]—an elastic modulus that
allows a theoretical (modulus normalized) compliance to
match an experimentally measured compliance for an actual
initial crack size, ao.
3.3.4 effective yield strength, σY [FL-2]— an assumed value
of uniaxial yield strength that represents the influence of plastic
yielding upon fracture test parameters.
3.3.4.1 Discussion—It is calculated as the average of the
0.2 % offset yield strength σYS, and the ultimate tensile
strength, σTS as follows:
σ YS1σ TS
σY 5
2
3.3.5 elastic modulus, E' [FL–2]—a linear-elastic factor
relating stress to strain, the value of which is dependent on the
degree of constraint. For plane stress, E' = E is used, and for
plane strain, E/(1 – v2) is used, with E being Young’s modulus
and v being Poisson’s ratio.
3.3.6 elastic plastic Jc [FL–1]—J-integral at the onset of
cleavage fracture.
3.3.7 elastic-plastic KJ [FL–3/2 ]—An elastic-plastic equiva-
lent stress intensity factor derived from the J-integral.
3.3.7.1 Discussion—In this test method, KJ also implies a
 E1921 − 19
stress intensity factor determined at the test termination point
under conditions that require censoring the data by 8.9.2.
3.3.8 elastic-plastic KJc [FL–3/2]—an elastic-plastic equiva-
lent stress intensity factor derived from the J-integral at the
point of onset of cleavage fracture, Jc.
3.3.9 equivalent value of median toughness, K Jceq~ med!
[FL-3/2]—an equivalent value of the median toughness for a
multi-temperature data set.
3.3.10 Eta (η)—a dimensionless parameter that relates plas-
tic work done on a specimen to crack growth resistance defined
in terms of deformation theory J-integral (7).
3.3.11 failure probability, pf—the probability that a single
selected specimen chosen at random from a population of
specimens will fail at or before reaching the KJc value of
interest.
3.3.12 initial ligament length, bo [L]— the distance from the
initial crack tip, ao, to the back face of a specimen.
˙ LL[LT-1]—rate of in-
3.3.13 load-line displacement rate,∆
crease of specimen load-line displacement.
3.3.14 pop-in—a discontinuity in a force versus displace-
ment test record (8).
3.3.14.1 Discussion—A pop-in event is usually audible, and
is a sudden cleavage crack initiation event followed by crack
arrest. The test record will show increased displacement and
drop in applied force if the test frame is stiff. Subsequently, the
test record may continue on to higher forces and increased
displacements.
3.3.15 precracked Charpy, PCC, specimen—SE(B) speci-
men with W = B = 10 mm (0.394 in.).
3.3.16 provisional reference temperature, (ToQ) [°C]—
Interim To value calculated using the standard test method
described herein. ToQ is validated as To in 10.5.
3.3.17 reference temperature, To [°C]—The test temperature
at which the median of the KJc distribution from 1T size
specimens will equal 100 MPa√m (91.0 ksi√in.).
3.3.18 SE(B) specimen span, S [L]—the distance between
specimen supports (See Test Method E1820 Fig. 3).
3.3.19 specimen thickness, B [L]—the distance between the
parallel sides of a test specimen as depicted in Fig. 1–3.
3.3.19.1 Discussion—In the case of side-grooved
specimens, the net thickness, BN, is the distance between the
roots of the side-groove notches.
3.3.20 specimen size, nT—a code used to define specimen
dimensions, where n is expressed in multiples of 1 in.
3.3.20.1 Discussion—In this method, specimen proportion-
ality is required. For compact specimens and bend bars,
specimen thickness B = n inches.
3.3.21 temperature, TQ [°C]—For KJc values that are devel-
oped using specimens or test practices, or both, that do not
conform to the requirements of this test method, a temperature
at which KJc (med) = 100 MPa√m is defined as TQ. TQ is not a
provisional value of To.
3.3.22 time to control force, tm [T],—time to Pm.
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
3
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
3.3.23 Weibull fitting parameter, K0—a scale parameter
located at the 63.2 % cumulative failure probability level (9).
KJc = K0 when pf = 0.632.
3.3.24 Weibull slope, b—with pf and KJc data pairs plotted in
linearized Weibull coordinates obtainable by rearranging Eq
18, b is the slope of a line that defines the characteristics of the
typical scatter of KJc data.
3.3.24.1 Discussion—A Weibull slope of 4 is used exclu-
sively in this method.
3.3.25 yield strength, σYS [FL−2]—the stress at which a
material exhibits a specific limiting deviation from the propor-
tionality of stress to strain at the test temperature. This
deviation is expressed in terms of strain.
3.3.25.1 Discussion—It is customary to determine yield
strength by either (1) Offset Method (usually a strain of 0.2 %
is specified) or (2) Total-Extension-Under-Force Method (usu-
ally a strain of 0.5 % is specified although other values of strain
may be used).
3.3.25.2 Discussion—Whenever yield strength is specified,
the method of test must be stated along with the percent offset
or total strain under force. The values obtained by the two
methods may differ.
4. Summary of Test Method
4.1 This test method involves the testing of notched and
fatigue precracked bend or compact specimens in a tempera-
ture range where either cleavage cracking or crack pop-in
develop during the loading of specimens. Crack aspect ratio,
a/W, is nominally 0.5. Specimen width in compact specimens
is two times the thickness. In bend bars, specimen width can be
either one or two times the thickness.
4.2 Force versus displacement across the notch at a speci-
fied location is recorded by autographic recorder or computer
data acquisition, or both. Fracture toughness is calculated at a
defined condition of crack instability. The J-integral value at
instability, Jc, is calculated and converted into its equivalent in
units of stress intensity factor, KJc. Censoring limits are based
on KJc to determine the suitability of data for statistical
analyses.
4.3 A minimum of six tests are required to estimate the
median KJc of the Weibull distribution for the data population
(10). Extensive data scatter among replicate tests is expected.
Statistical methods are used to characterize these data popula-
tions and to predict changes in data distributions with changed
specimen size.
4.4 The statistical relationship between specimen size and
KJc fracture toughness is assessed using weakest-link theory,
thereby providing a relationship between the specimen size and
KJc (4). Limits are placed on the fracture toughness range over
which this model can be used.
4.5 For the definition of the toughness transition curve, a
master curve concept is used (11, 12). The position of the curve
on the temperature coordinate is established from the experi-
mental determination of the temperature, designated To, at
which the median KJc for 1T size specimens is 100 MPa√m
(91.0 ksi√in.). Selection of a test temperature close to that at
which the median KJc value will be 100 MPa√m is encouraged
 E1921 − 19
and a means of estimating this temperature is suggested. Small
specimens such as precracked Charpy’s may have to be tested
at temperatures below To where KJc(med) is well below 100
MPa√m. In such cases, additional specimens may be required
as stipulated in 8.5.
4.6 Tolerance bounds can be determined that define the
range of scatter in fracture toughness throughout the transition
range. The standard deviation of the fitted distribution is a
function of Weibull slope and median KJc value, KJc(med).
5. Significance and Use
5.1 Fracture toughness is expressed in terms of an elastic-
plastic stress intensity factor, KJc, that is derived from the
J-integral calculated at fracture.
5.2 Ferritic steels are microscopically inhomogeneous with
respect to the orientation of individual grains. Also, grain
boundaries have properties distinct from those of the grains.
Both contain carbides or nonmetallic inclusions that can act as
nucleation sites for cleavage microcracks. The random location
of such nucleation sites with respect to the position of the crack
front manifests itself as variability of the associated fracture
toughness (13). This results in a distribution of fracture
toughness values that is amenable to characterization using the
statistical methods in this test method.
5.3 The statistical methods in this test method assume that
the data set represents a macroscopically homogeneous
material, such that the test material has both the uniform tensile
and toughness properties. The fracture toughness evaluation of
nonuniform materials is not amenable to the statistical analysis
procedures employed in this test method. For example, multi-
pass weldments can create heat-affected and brittle zones with
localized properties that are quite different from either the bulk
or weld materials. Thick-section steels also often exhibit some
variation in properties near the surfaces. Metallographic analy-
sis can be used to identify possible nonuniform regions in a
material. These regions can then be evaluated through me-
chanical testing such as hardness, microhardness, and tensile
testing for comparison with the bulk material. It is also
advisable to measure the toughness properties of these nonuni-
form regions distinctly from the bulk material. Section 10.6
provides a screening criterion to assess whether the data set
may not be representative of a macroscopically homogeneous
material, and therefore, may not be amenable to the statistical
analysis procedures employed in this test method. If the data
set fails the screening criterion in 10.6, the homogeneity of the
material and its fracture toughness can be more accurately
assessed using the analysis methods described in Appendix X5.
5.4 Distributions of KJc data from replicate tests can be used
to predict distributions of KJc for different specimen sizes.
Theoretical reasoning (9), confirmed by experimental data,
suggests that a fixed Weibull slope of 4 applies to all data
distributions and, as a consequence, standard deviation on data
scatter can be calculated. Data distribution and specimen size
effects are characterized using a Weibull function that is
coupled with weakest-link statistics (14). An upper limit on
constraint loss and a lower limit on test temperature are defined
between which weakest-link statistics can be used.
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
4
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
5.5 The experimental results can be used to define a master
curve that describes the shape and location of median KJc
transition temperature fracture toughness for 1T specimens
(15). The curve is positioned on the abscissa (temperature
coordinate) by an experimentally determined reference
temperature, To. Shifts in reference temperature are a measure
of transition temperature change caused, for example, by
metallurgical damage mechanisms.
5.6 Tolerance bounds on KJc can be calculated based on
theory and generic data. For added conservatism, an offset can
be added to tolerance bounds to cover the uncertainty associ-
ated with estimating the reference temperature, To, from a
relatively small data set. From this it is possible to apply a
margin adjustment to To in the form of a reference temperature
shift.
5.7 For some materials, particularly those with low strain
hardening, the value of To may be influenced by specimen size
due to a partial loss of crack-tip constraint (5). When this
occurs, the value of To may be lower than the value that would
be obtained from a data set of KJc values derived using larger
specimens.
5.8 As discussed in 1.3, there is an expected bias among To
values as a function of the standard specimen type. The
magnitude of the bias may increase inversely to the strain
hardening ability of the test material at a given yield strength,
as the average crack-tip constraint of the data set decreases
(16). On average, To values obtained from C(T) specimens are
higher than To values obtained from SE(B) specimens. Best
estimate comparison indicates that the average difference
between C(T) and SE(B)-derived To values is approximately
10 °C (2). However, individual C(T) and SE(B) datasets may
show much larger To differences (3, 17, 18), or the SE(B) To
values may be higher than the C(T) values (2). On the other
hand, comparisons of individual, small datasets may not
necessarily reveal this average trend. Datasets which contain
both C(T) and SE(B) specimens may generate To results which
fall between the To values calculated using solely C(T) or
SE(B) specimens.
6. Apparatus
6.1 Precision of Instrumentation—Measurements of applied
forces and load-line displacements are needed to obtain work
done on the specimen. Force versus load-line displacement
shall be recorded digitally on computers or autographically on
x-y plotters. For computers, digital signal resolution shall be at
least 1/32,000 of the displacement transducer signal range and
shall be at least 1/4,000 of the force transducer signal range.
6.2 Grips for C(T) Specimens—A clevis with flat-bottom
holes is recommended. See Test Method E399, Fig. A6.2, for a
recommended design. Clevises and pins should be fabricated
from steels of sufficient strength to elastically resist indentation
loads (greater than 40 Rockwell hardness C scale (HRC)).
6.3 Bend Test Fixture—A suitable bend test fixture scheme
is shown in Fig. A3.2 of Test Method E399. It allows for roller
pin rotation and minimizes friction effects during the test.
Fixturing and rolls should be made of high-hardness steel
(HRC greater than 40).
 E1921 − 19
6.4 Displacement Gage for Compact Specimens:
6.4.1 Displacement measurements are made so that J values
are determined from area under force versus displacement test
records (a measure of work done). If the test temperature
selection recommendations of this practice are followed, crack
growth measurement will probably prove to be unimportant.
Results that fall within the limits of uncertainty of the
recommended test temperature estimation scheme will prob-
ably not have significant slow-stable crack growth prior to
instability. Nevertheless, crack growth measurements are rec-
ommended to provide supplementary information, and these
results may be reported.
6.4.2 Unloading compliance is the primary recommendation
for measuring slow-stable crack growth. See Test Method
E1820. When multiple tests are performed sequentially at low
test temperatures, there will be condensation and ice buildup
on the grips between the loading pins and flats of the clevis
holes. Ice will interfere with the accuracy of the unloading
compliance method. Alternatively, crack growth can be mea-
sured by other methods such as electric potential, but care must
be taken to avoid specimen heating when low test temperatures
are used.
6.4.3 In compact C(T) specimens, displacement measure-
ments on the load-line are recommended for J determinations.
However, the front face position at 0.25W in front of the
load-line can be used with interpolation to load-line
displacement, as suggested in 7.1.
6.4.4 Clip gage (or other similar displacement gage) accu-
racy shall be verified according to 6.2.2 of Test Method E1820.
6.5 Displacement Gages for Bend Bars, SE(B):
6.5.1 The SE(B) specimen has two displacement gage
locations. A load-line displacement transducer is primarily
intended for J computation, but may also be used for calcula-
tions of crack size based on elastic compliance, if provision is
made to subtract the extra displacement due to the elastic
compliance of the fixturing. The load-line gage shall display
accuracy of 1 % over the working range of the gage. The gages
used shall not be temperature sensitive.
6.5.2 Alternatively, a crack-mouth opening displacement
(CMOD) gage can also be used to determine the plastic part of
J. However, it is necessary to employ a plastic eta (η) value
developed specifically for the CMOD location (19) or infer
load-point displacement from CMOD using an expression that
relates the two displacements (20). In either case, the procedure
described in 9.1.4 is used to calculate the plastic part of J.
However, it is recommended that the plastic part of J be
estimated from the direct CMOD or load-line displacement
measurement rather than inferring load-line displacement from
CMOD. Additionally, CMOD measurement is more accurate
than load-line displacement for estimating crack length from
compliance.
6.5.3 Crack growth can be measured by alternative methods
such as electric potential, but care must be taken to minimize
specimen heating effects in low-temperature tests (see also
6.4.2) (21).
6.6 Force Measurement:
6.6.1 Testing shall be performed in a machine conforming to
Practices of E4 and Test Methods E8/E8M. Applied force may
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
5
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
be measured by any transducer with a noise-to-signal ratio less
than 1/2,000 of the transducer signal range.
6.6.2 Calibrate force measurement instruments by way of
Practice E74, 10.2. Annual calibration using calibration equip-
ment traceable to the National Institute of Standards and
Technology is a mandatory requirement.
6.7 Temperature Control—Specimen temperature shall be
measured with thermocouple wires and potentiometers. It is
recommended that the two thermocouple wires be attached to
the specimen surface separately, either by welding, spot
welding, or by being affixed mechanically. Mechanical attach-
ment schemes must be verified to provide equivalent tempera-
ture measurement accuracy. The purpose is to use the test
material as a part of the thermocouple circuit (see also 8.6.1).
Accuracy of temperature measurement shall be within 3°C of
true temperature and repeatability among specimens shall be
within 2°C. Precision of measurement shall be 61°C or better.
The temperature measuring apparatus shall be checked every
six months using instruments traceable to the National Institute
of Standards and Technology in order to ensure the required
accuracy.
7. Specimen Configuration, Dimensions, and Preparation
7.1 Compact Specimens—Three recommended C(T) speci-
men designs are shown in Fig. 1. One C(T) specimen configu-
ration is taken from Test Method E399; the two with cutout
sections are taken from E1820. The latter two designs are
modified to permit load-line displacement measurement. Room
is provided for attachment of razor blade tips on the load-line.
Care should be taken to maintain parallel alignment of the
blade edges. When front face (at 0.25W in front of the
load-line) displacement measurements are made with the Test
Method E399 design, the load-line displacement can be in-
ferred by multiplying the measured values by the constant 0.73
(22). The ratio of specimen height to width, 2H/W is 1.2, and
this ratio is to be the same for all types and sizes of C(T)
specimens. The initial crack size, ao, shall be 0.5W 6 0.05W.
Specimen width, W, shall be 2B.
7.2 Disk-shaped Compact Specimens—A recommended
DC(T) specimen design is shown in Fig. 2. Initial crack size,
ao, shall be 0.5W6 0.05W. Specimen width shall be 2B.
7.3 Single-edge Notched Bend—The recommended SE(B)
specimen designs, shown in Fig. 3, are made for use with a
span-to-width ratio, S/W = 4. The width, W, can be either 1B or
2B. The initial crack size, ao, shall be 0.5W 6 0.05W.
7.4 Machined Notch Design—Three designs of fatigue
crack starter notches are shown in Fig. 4. These notches can be
straight through the specimen thickness or incorporate the
chevron form (Fig. 4). To facilitate fatigue cracking at low
stress intensity levels, it is recommended that the root radius
for either a straight-through slot terminating in a V-notch, or a
narrow notch, be 0.08 mm (0.003 in.) or less. If a chevron form
of notch is used, the root radius may be 0.25 mm (0.010 in.) or
less. In the case of a notch ending in a drilled hole, a sharp
stress raiser at the end of the hole will facilitate fatigue
precracking and help ensure that the precrack centering re-
quirement in 7.8.2 is met.
 E1921 − 19

FIG. 1 Three Compact Specimen Designs That Have Been Used Successfully for Fracture Toughness Testing

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
6
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

NOTE 1—A surfaces shall be perpendicular and parallel as applicable to within 0.002W TIR.
NOTE 2—The intersection of the crack starter notch tips with the two specimen surfaces shall be equally distant from the top and bottom extremes of
the disk within 0.005W TIR.
NOTE 3—Integral or attached knife edges for clip gage attachment may be used. See also Fig. 6, Test Method E399.
FIG. 2 Disk-shaped Compact Specimen DC(T) Standard Proportions

NOTE 1—All surfaces shall be perpendicular and parallel within 0.001W TIR; surface finish 64v.
NOTE 2—Crack starter notch shall be perpendicular to specimen surfaces to within6 2°.
FIG. 3 Recommended Bend Bar Specimen Design

Fig. 5 summarizes the maximum starter notch dimensions.
The notch cutout for measurement gages shall be no greater
than 0.2W wide by 0.1W deep. The allowable starter notch
height shall be no greater than 0.063W. The centerline of the
crack-starter notch shall not deviate from the specimen center-
line by more than 0.005W. Fig. 5 also defines the notch tip
length for both a V-notch type and a narrow-notch type. The
narrow notch is often machined using a wire or plunge
electrical discharge machining technique with no additional
machining to further sharpen the notch root radius. Finally, Fig.
5 also summarizes the fatigue precrack extension requirements.
The average length of the fatigue precrack extension from the
machined notch, ∆apc, (determined using the measured initial
crack length defined in 8.8.1) shall equal or exceed the larger
of 0.5N, (∆ash + ∆af), or 0.25 mm. Further, the sum of ∆apc and
the notch tip length shall exceed 2.0N. Additional fatigue
precracking requirements are contained in 7.8.
7.5 Specimen Dimension Requirements—The crack front
straightness criterion defined in 8.9.1 must be satisfied. The
specimen remaining ligament, bo, must have sufficient size to
maintain a condition of high crack-front constraint at fracture.
The maximum KJc capacity of a specimen is given by:

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
7
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

FIG. 4 Envelope Crack Starter Notches

FIG. 5 Envelope of Fatigue Crack and Crack Starter Notches

K Jclimit 5 Œ Ebo σ YS
30~ 1 2 v 2 !
(1)
Measurement of σys at the test temperature (T) using Test
Methods E8/E8M is preferred for use in Eq 1. When σys has not
been measured at T, any of the following three methods are
where: acceptable for estimating σys at T for use in Eq 1:
bo = W-ao

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
8
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.

(1) Using a value of σys measured at a higher temperature
than T.
(2) Interpolating between measurements of σys at tempera-
tures above and below T, as long as the σys measurement
temperatures are within 75 °C of each other. Extrapolation of
the fit to infer yield strength outside of the measurement points
is not allowed with this method. This method is applicable over
a test temperature range of - 200 °C and 300 °C
(3) Determining σys from the following equation which can
be used for temperatures between -200°C and 300°C. (23) See
Note 1.
σ ys 5 σ ysRT 1105 ⁄ ~ 491 1 1.8 T ! 2 189~ MPa!
where:
T = test temperature (°C), and
σysRT = the material yield strength at room temperature
(MPa)
NOTE 1—Eq 2 should not be used to determine σysRT from σys values
obtained at other temperatures.
KJc data that exceed this requirement (that is, Eq 1) are used
in a data censoring procedure. Details of this procedure are
described in 10.2.1.
7.6 Small Specimens—At high values of fracture toughness
relative to specimen size and material flow properties, the
values of KJc that meet the requirements of Eq 1 may not
always provide a unique description of the crack-front stress-
strain fields due to some loss of constraint caused by excessive
plastic flow (5). This condition may develop in materials with
low strain hardening. When this occurs, the highest KJc values
of the data set could possibly cause the value of To to be lower
than the value that would be obtained from testing specimens
with higher constraint.
7.7 Side Grooves— Side grooves are optional. Precracking
prior to side-grooving is recommended, despite the fact that
crack growth on the surfaces might be slightly behind. Speci-
mens may be side-grooved after precracking to decrease the
curvature of the initial crack front. In fact, side-grooving may
be indispensable as a means for controlling crack front
straightness in bend bars of square cross section. The total
side-grooved depth shall not exceed 0.25B. Side grooves with
an included angle of 45° and a root radius of 0.5 6 0.2 mm
(0.02 6 0.01 in.) usually produce the desired results.
7.8 Precracking:
7.8.1 Fatigue Loading Requirements—Allowable fatigue
force values are limited to keep the maximum stress intensity
factor applied during precracking, Kmax, well below the mate-
rial fracture toughness measured during the subsequent test.
The fatigue precracking shall be conducted with the specimen
fully heat-treated to the condition in which it is to be tested. No
intermediate heat treatments between precracking and testing
are allowed. There are several ways of promoting early crack
initiation: (1) by providing a very sharp notch tip, (2) by using
a chevron notch (Fig. 4), (3) by statically preloading the
specimen in such a way that the notch tip is compressed in a
direction normal to the intended crack plane (to a force not to
exceed Pm), and (4) by using a negative fatigue force ratio; for
a given maximum fatigue force, the more negative the force
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
E1921 − 19
ratio, the earlier crack initiation is likely to occur. The peak
compressive force shall not exceed Pm as defined in the
equations below:
0.5Bb 2o σ Y
For SE~B! specimens, P m 5
S
(3)
0.4Bb 2o σ Y
For C~T! and DC~T! specimens, P m 5 (4)
2W1a o
7.8.2 Fatigue Precracking Procedure—Fatigue precracking
can be conducted under either force control, displacement
control, or K control. If the force cycle is maintained constant,
(2) the maximum K and ∆K will increase with crack size; if the
displacement cycle is maintained constant, the reverse will
happen. If K is maintained constant, force has to be reduced as
a function of increasing crack size. Fatigue cycling is con-
ducted using a sinusoidal waveform and a frequency close to
the highest practical value. There is no known marked fre-
quency effect on fatigue precrack formation up to at least 100
Hz in the absence of adverse environments. The specimen shall
be accurately located in the loading fixture to achieve uniform,
symmetric loading. The specimen should be carefully moni-
tored until crack initiation is observed on one side. If crack
initiation is not observed on the other side before appreciable
growth is observed on the first side, then fatigue cycling should
be stopped to try to determine the cause and find a remedy for
the unsymmetrical behavior. Sometimes, simply turning the
specimen around in relation to the fixture will solve the
problem.
Precracking can be performed either by some method of
smoothly and continually decreasing the maximum stress
intensity factor (Kmax) or by using discrete steps. It is suggested
that the reduction in Kmax between any discrete step be no
greater than 20 % because reducing Kmax too rapidly can result
in precrack growth rate retardation. It is also suggested that
measurable crack extension occur before proceeding to the
next step. Precracking is generally most effectively conducted
using R = Pmin/Pmax = 0.1. Maximum force values shall be
accurate to within 6 5 % of their target values.
Fig. 6 shows the allowable envelope for Kmax during
precracking. The precracking Kmax and crack extension re-
quirements are summarized in Table 1, and Table 2. Precrack-
ing can be conducted in any manner such that Kmax remains
within the envelope and the maximum fatigue force is less than
Pm. The Kmax applied to the specimen shall not exceed 25
MPa√m (22.8 ksi√in) at any crack length, and may be limited
by Pm for small specimens or low yield strength materials, or
both. As the testing temperature decreases compared to the
precracking temperature, the warm prestressing effect
increases, which can elevate the measured fracture toughness.
To minimize the warm prestressing effect, the maximum K that
may be applied to the specimen during ∆af (Kf in Fig. 6) shall
not exceed 15 MPa√m (13.7 ksi√in). Alternatively, when the
testing temperature is equal to or above the precracking
temperature, Kf shall not exceed 20 MPa√m (18.3 ksi√in). The
minimum length of ∆af (Fig. 6) is 0.2 mm (0.008 in.). ∆ash is
greater than or equal to the change in plastic zone size in going
from a maximum K of 25 MPa√m (22.8 ksi√in) to Kf. The
9
 E1921 − 19

FIG. 6 Envelope of Allowable Kmax During Precracking

TABLE 1 Kmax Requirements

Intial: Kmax cannot exceed 25 MPa=m (22.8 ksi=in.) and the maximum fatigue force cannot exceed Pm.

Final: Kf depends on the test temperature:

Test Temperature Kf throughout ∆af

< precracking temperature < 15MPa=m (13.7 ksi=in.)
$ precrackeing temperature < 20MPa=m (18.3 ksi=in.)

TABLE 2 Crack Extension Requirements

∆apc $ MAX {0.5N, (∆ash + ∆af),

0.25 mm}
∆ash $ rp1 – rp2

Where:

rpl =
r p1 5 S D
1 K max
with
3π σ ys
2

Kmax ≤ 25 MPa√m (22.8

ksi√in.)

rp2 = r p2 5 S D
1 Kf
3π σ ys
2

∆af $ 0.2 mm (0.008 in.)

minimum value for ∆ash defines the condition where the Also, as stipulated previously in 7.4 and summarized in Fig.
leading edge of the plastic zone remains stationary as Kmax is 5, the length of the fatigue precrack extension from the
decreased. machined notch, ∆apc (determined using the measured initial
∆a sh $ r p1 2 r p2 (5) crack length defined in 8.8.1), shall equal or exceed the larger
of 0.5N, (∆ash + ∆af), or 0.25 mm at each of the nine
where: measurement locations defined in 8.8.1. Additionally, the sum
of ∆apc and the notch tip length shall exceed 2.0N at each of
r p1 5
1
3π S D
K max
σ ys
2
with K max the nine measurement locations defined in 8.8.1. The precrack
must also meet the curvature requirement in 8.9.1. Ensuring
5 25 MPa=m ~ 22.8 ksi=in.! that the average ∆apc is long enough such that the minimum
fatigue precrack extension occurs at each measurement point in
1 Kf 2
r p2 5
3π σ ys S D 8.8.1 for a crack having the maximum curvature allowed in
NOTE 2—If the yield strength (σys) is not known, a low estimate should 8.9.1 will provide some confidence that these requirements are
be used to obtain a conservatively high estimate of ∆ash. met before testing the specimen.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
10
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.

8. Procedure
8.1 Testing Procedure—The objective of the procedure de-
scribed here is to determine the J-integral at the point of crack
instability, Jc. Crack growth can be measured by partial
unloading compliance, or by any other method that has
precision and accuracy, as defined below. However, the
J-integral is not corrected for slow-stable crack growth in this
test method.
8.2 Test Preparation—Prior to each test, certain specimen
dimensions should be measured, and the average starting crack
size estimated. The average starting crack size can be estimated
using a variety of techniques including precrack compliance,
back-face strain, and using the average of the optical side face
measurements.
NOTE 3—When side-grooving is to be used, first precrack without side
grooves and then visually estimate the precrack size.
If estimates are available from multiple techniques, the user
shall select the value that is believed to be most representative
of the average crack size.
8.2.1 The dimensions B, BN, and W shall be measured to
within 0.05 mm (0.002 in.) accuracy or 0.5 %, whichever is
larger.
8.2.2 Follow Test Method E1820, 8.5 for crack size
measurement, 8.3.2 for testing compact specimens and 8.3.1
for testing bend specimens.
8.3 The required minimum number of KJc results that are
uncensored is specified according to the value of KJc(med). See
also 8.5.
8.4 Test Temperature Selection—It is recommended that the
selected temperatures be close to that at which the KJc(med)
value will be about 100 MPa√m for the specimen size selected.
8.4.1 Quasi-static loading rates—If loading rate complies
with the limits stated in 8.7.1, Charpy V-notch data can be used
as an aid for predicting a viable test temperature. If a Charpy
transition temperature, TCVN, is known corresponding to a 28 J
Charpy V-notch energy or a 41 J Charpy V-notch energy, the
constant C can be chosen from Table 3 corresponding to the
test specimen size (defined in 3.3.20), and used to estimate5 the
test temperature from (12, 24).
T 5 T CVN1C
5
Standard deviation on this estimate has been determined to be 15°C.
TABLE 3 Constants for Test Temperature Selection Based on
Charpy Results
Specimen Size, Constant C (°C)
(nT)
28 J
0.4A −32
0.5 −28
1 −18
2 −8
3 −1
4 2
A
For precracked Charpy specimens, use C = −50 or −56°C.
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
E1921 − 19
8.4.2 The procedure outlined in 8.4.1 is only appropriate for
determining an initial test temperature. The iterative scheme
described in 10.3.1 may be necessary to refine this test
temperature in order to increase To accuracy. Testing below the
temperature specified in Eq 6 may be appropriate for low
upper-shelf toughness materials to avoid ductile crack growth
before cleavage onset, and for low yield strength materials to
avoid obtaining data that must be censored because it exceeds
KJclimit in accordance with Eq 1.6
8.5 Testing Below Temperature, To—When the equivalent
value of KJc(med) for 1T specimens is greater than 83 MPa√m,
the required number of uncensored KJc values to perform the
analyses covered in Section 10 is six. However, small speci-
mens such as precracked Charpy specimens can develop
excessive numbers of KJc values that exceed the KJclimit (Eq 1)
when testing close to the To temperature. In such cases it is
advisable to test at temperatures below To, where most, if not
all, KJc data developed can be uncensored. The disadvantage
here is that the uncertainty in To determination increases as the
lower-shelf toughness is approached. This increase in uncer-
tainty can be countered by testing more specimens thereby
increasing the KJc(med) accuracy. The number of specimens
required for obtaining a valid To measurement as a function of
the test temperature is provided in 10.3 (10.4.1 for the special
case of testing at a single temperature).
8.6 Specimen Test Temperature Control and Measurement—
For tests at temperatures other than ambient, any suitable
means (liquid, gas vapor, or radiant heat) may be used to cool
or heat the specimens, provided the region near the crack tip
can be maintained at the desired temperature as defined in 6.7
during the conduct of the test.
8.6.1 The most dependable method of monitoring test tem-
perature is to weld or spot weld each thermocouple wire
separately to the specimen, spaced across the crack plane. The
specimen provides the electrical continuity between the two
thermocouple wires, and spacing should be enough not to raise
any question of possible interference with crack tip deforma-
tion processes. Alternative attachment methods can be me-
chanical types such as drilled hole, or by a firm mechanical
holding device so long as the attachment method is verified for
accuracy and these practices do not disturb the crack tip stress
(6)
field of the specimen during loading.
8.6.2 To verify that the specimen is properly seated into the
loading device and that the clip gage is properly seated,
estimate the specimen crack size while working-in the test
setup at test temperature. Working-in is accomplished via
repeated preloading and unloading in the linear elastic range
between force values of 0.2 Pmax and Pmax (where Pmax is the
largest allowable precracking force of the finishing cycles as
prescribed in 7.8.2) at least three times. For each unloading/
reloading sequence, estimate the precrack size using Test
41 J
Method E1820, Eq A2.12 for C(T) specimens and Eq A1.12 for
−38
−34
SE(B) specimens. The elastic modulus, E, used in these
−24 calculations shall be the nominal value for the material at the
−14 test temperature. The nominal value of E shall come from
−7
−4
6
Data censoring is covered in 8.9.2 and Section 10.
11
 E1921 − 19
either handbook values or dedicated modulus testing per Test TABLE 4 SE(B) Specimen Rate Estimation-C(T) Specimen Rate
Method E111 or equivalent. Tensile test results do not provide Estimation
accurate elastic modulus values. Alternatively, the following SE(B) Specimen Rate Estimation C(T) Specimen Rate Estimation
equation can be used to determine the nominal value of E.
a/W t M K̇ E∆˙ LL a/W t M K̇ E∆˙ LL
E 5 204 2 T⁄16 GPa (7) σ Y œW dK σ Y œW dK
W W
dt œ dt œ
where: 0.45 0.346 5.064 0.45 0.412 3.475
T = test temperature in °C. 0.50 0.333 5.263 0.50 0.386 3.829
0.55 0.318 5.522 0.55 0.361 4.212
This equation was derived from fitting the tabular values for 0.60 0.302 5.851 0.60 0.336 4.635
ferritic steels contained in ASME Section II, Part D. The fit is 0.65 0.283 6.267 0.65 0.312 5.118
0.70 0.263 6.798 0.70 0.287 5.696
valid for –200°C ≤ T ≤ 300°C.
8.6.3 Check the estimated crack size slope against the
average precrack size defined in 8.2. The test setup is consid-
ered acceptable when the last three consecutive estimated crack
sizes are all within 10 % of the final precrack size and no geometry. Pm is nominally 40 % of limit force; see 7.8.1. The
individual estimated crack size differs from the mean by more actual crosshead rate used must be adjusted to account for test
than 60.002W. To minimize the difference between the pre- machine compliance if the load-line displacement rate of Table
crack size and the working-in estimated crack size, the nominal 4 is used. The crosshead speed during periodic partial
E value may be adjusted up to 10 %. Modulus adjustments unloadings, if applied, may be as slow as needed to accurately
should only be made when the force transducer and clip gage estimate crack growth, but shall not be faster than the rate
calibrations are known to be within acceptable limits of specified for loading.
accuracy, see Section 6. The value of E in use for the final three 8.7.2 Slow Loading Rates—For loading rates less than
acceptable working-in unloading/reloading sequences shall be 0.1 MPa√m ⁄s, the procedures of this test method can be used if
used for all crack size estimates throughout the remainder of the failure mode remains cleavage. The corresponding refer-
the given test. If the repeatability or the accuracy, or both of the ence temperature is then reported as To,X using the convention
estimated crack sizes are outside the prescribed limits, the test described in A1.2.1.1.
setup is questionable and should be thoroughly rechecked. It is 8.8 Test Termination— After completion of the test, opti-
essential that the specimen temperature and clip gage are stable cally measure initial crack size and the extent of slow-stable
and that the clip gage knife edges are sharp in order to meet crack growth or crack extension due to crack pop-in, or both,
these requirements. Be aware that ice buildup at the loading when applicable.
clevis hole between tests can affect accuracy. Therefore, the 8.8.1 When the failure event is full cleavage fracture,
loading pins and devices should be dried before each test. determine the initial fatigue precrack size, ao, as follows:
8.7 Testing for KJc—All tests shall be conducted under measure the crack size at nine equally spaced points centered
displacement control. Force versus load-point displacement about the specimen centerline and extending to 0.01B from the
measurements shall be recorded. Periodic partial unloading can free surfaces of plane sided specimens or near the side groove
be used to determine the extent of slow-stable crack growth if roots on side grooved specimens. Average the two near-surface
it occurs. Alternative methods of measuring crack extension, measurements and combine the average of these two readings
for example the potential drop method, can be used (21). If with the remaining seven crack measurements. Determine the
displacement measurements are made at a location other than average of those eight values. Measure the extent of slow-
at the load point, the ability to infer load point displacement stable crack growth if it develops applying the same procedure.
within 2 % of the absolute values shall be demonstrated. In the The measuring instruments shall have an accuracy of 0.025
case of the front face for compact specimens (7.1), this mm (0.001 in.).
requirement has been sufficiently proven so that no demonstra- 8.8.2 Post-Test Check—Compare the estimated fatigue pre-
tion is required. For bend bars, see 6.5.2. Crack size prediction crack length determined in 8.6.3 with the optical average value
from partial unloading slopes at a different location will require determined in 8.8.1. The pre-test estimate shall not differ from
different compliance calibration equations than those recom- the post-test optical value by more than 5 %. If the error
mended in 8.6.2. Table 2 in Practice E561 contains equations exceeds 5 %, then a unique value of E (Ei) can be found for
that define compliance for other locations on the compact each test to match the optical average value as closely as
specimen. possible using Test Method E1820, Eq A2.12 for C(T) speci-
8.7.1 Quasi-static Loading—Load specimens at a rate such mens and Eq A1.12 for SE(B) specimens. Ei values shall be
that K̇ during the initial elastic portion is between 0.1 and 2 within 10 % of the nominal value of E identified in 8.6.2.
MPa√m/s. Variation of the loading rate within these limits 8.9 Qualification of Data:
allows obtaining a To which is insensitive to the loading rate 8.9.1 The KJc datum shall be considered invalid if any of the
within 10°C (25). The testing machine loading rate associated nine physical measurements of the starting crack size differ by
with this allowable range can be determined in terms of the more than 0.1(boBN)1/2 from the average defined in 8.8.1. The
time to reach Pm (tm) or the specimen load-line displacement datum is also invalid if the calculated crack length for the test
rate ∆˙ LL. Table 4 is provided to determine the time to tm or ∆˙ LL determined in 8.8.2 differs from the optical average value
as a function of K̇, W, E, and σYS for each allowable specimen determined in 8.8.1 by more than 5 %.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
12
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
8.9.2 A KJc datum requires censoring if the specimen where:
exceeds the KJclimit requirement of 7.5, or if a test has been Ke = [P/(BBNW)1/2] f (ao/W),
discontinued at a value of KJ without cleavage fracture after
~ 21a o /W !
surpassing KJclimit. Another limit, KJc∆a, is violated in tests that f ~ a o /W ! 5 0.88614.64 ~ a o /W ! 2 13.32~ a o /W ! 2
~ 1 2 a o /W ! 3/2 @
terminate in cleavage after slow stable crack growth that
114.72~ a o /W ! 3 2 5.6~ a o /W ! 4 # , (10)
exceeds the smaller of either 0.05(W-ao) or 1 mm (0.040 in.) at
the longest crack size dimension measured by 8.8.1. A KJc and ao = initial crack size.
datum exceeding KJc∆a also requires censoring. Censored KJ or 9.1.2 For disk-shaped compact specimens, DC(T), the elas-
KJc values contain statistically useable information and are tic component of J is calculated as follows:
used in the evaluation as described in 10.2.1. If KJclimit is ~ 1 2 v 2 ! K 2e
violated, the KJc datum shall be replaced with KJclimit in the Je 5 (11)
E
analysis. See, for example, X1.3.1. If KJc∆a is violated, the KJc
where:
datum shall be replaced with KJc∆a (as determined in 10.2.1) in
the analysis. See, for example, X1.3.6. If both KJclimit and KJc∆a Ke = [P/(BBNW)1/2] f(ao/W),
are violated, the lower value of the two shall be used to replace ~ 21a o /W !
f ~ a o /W ! 5 0.7614.8~ a o /W ! 2 11.58~ a o /W ! 2
the KJc datum for data censoring purposes in the analysis. ~ 1 2 a o /W ! 3/2 @
8.9.3 For any test terminated without cleavage fracture, and 111.43~ a o /W ! 3 2 4.08~ a o /W ! 4 # , (12)
for which the final KJ value does not exceed either censoring and ao = initial crack size.
limit cited in 8.9.2, the test is judged to be a nontest, the result 9.1.3 For SE(B) specimens of both B × B and B × 2B cross
of which shall be discarded. sections and span-to-width ratios of 4, the elastic component of
8.9.4 Data sets that contain all uncensored KJc values are J is calculated as follows:
used without modification in Section 10. Data sets that contain
~ 1 2 v 2 ! K 2e
some censored data, but that meet the minimum data require- Je 5
E
(13)
ments of 8.5, are used in the evaluation as described in 10.2.1.
Remedies for excessive censored data points include (1) testing where:
at a lower test temperature, (2) testing with larger specimens, K e = {PS/[(BBN)1/2 W3/2]}f (ao/W),
or (3) testing more specimens to satisfy the minimum data 3 ~ a o /W ! 1/2
requirements. f ~ a o /W ! 5 (14)
2 @ 112 ~ a o /W ! #
8.9.5 A discontinuity in a force-displacement record prior to
attaining maximum force may be a pop-in event. Test equip- 1.99 2 ~ a o /W ! ~ 1 2 a o /W ! @ 2.15 2 3.93~ a o /W ! 12.7~ a o /W ! 2 #
ment can, at times, introduce a discontinuity in the force- ~ 1 2 a o /W ! 3/2 ,
displacement record. However, a pop-in is uniquely character- and ao = the initial crack size.
ized by a decrease in force and a simultaneous increase in crack 9.1.4 The plastic component of J is calculated as follows:
opening displacement. After a pop-in event, both force and
ηA p
crack opening displacement continue to increase. Discontinui- Jp 5 (15)
B Nb o
ties in the force displacement record after attaining maximum
force are not pop-in events and are not addressed in this where:
section. If a single pop-in or multiple pop-ins are observed A p = A – 1/2CoP2 ,
during testing, assess the estimated crack mouth opening A = Ae + Ap (see Fig. 7),
compliance change by 9.2 to determine if it changes by more Co = reciprocal of the initial elastic slope, ∆V/∆P (Fig. 7),
than 2 % as a result of the pop-ins. If the change is greater than and
2 %, the pop-in shall be treated as significant and the KJc value bo = initial remaining ligament.
associated with the pop-in shall be used as the result of the test. 9.1.4.1 For standard and disk-shaped compact specimens,
Note that an estimated compliance change of 2 % in accor- Ap is based on load-line displacement (LLD) and η = 2 + 0.522
dance with 9.2 predicts an increase in crack size of no more bo/W. For bend bar specimens of both B × B and B × 2B cross
than 1 % depending on the specimen type. sections and span-to-width ratios of 4, Ap may be based on
either LLD or crack-mouth opening displacement (CMOD).
9. Calculation of Jc and KJc Values from the Input Data Using LLD, η = 1.9. Using CMOD, η = 3.667–2.199(a/W)+
0.4376(a/W)2. Determination of η for bend bars based on
9.1 Determine the J-integral at onset of cleavage fracture as CMOD is also discussed in 6.5.2.
the sum of elastic and plastic components: 9.1.5 KJc is determined for each datum from J at onset of
J c 5 J e 1J p (8) cleavage fracture, Jc using:
9.1.1 For compact specimens, C(T), the elastic component
of J is calculated as follows:
K Jc 5 ΠJc
E
1 2 v2
(16)

~ 1 2 v 2 ! K 2e The nominal value of E identified in 8.6.2 for each test

Je 5 (9)
E temperature shall be used in the conversion of Jc to KJc.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
13
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

FIG. 7 Definition of the Plastic Area for Jp Calculations

9.2 Pop-in Evaluation—Test records used for KJc analyses
are those that exhibit complete specimen separation due to
cleavage fracture and those that exhibit significant pop-ins. As
discussed in 8.9.5, if a force-displacement record exhibits a
small but perceptible discontinuity that is characterized by a
drop in force with a simultaneous increase in crack opening
displacement, a pop-in may have occurred. As depicted in Fig.
8, the accumulated change in crack mouth opening compliance
before and after single or multiple pop-in events are estimated7
as:
7
Machine compliance can result in differences between this estimated crack
mouth opening compliance and the actual specimen compliance. However, the
estimated compliance change is always equal to or greater than the specimen
compliance.

NOTE—C0 is the initial compliance

NOTE—The pop-ins have been exaggerated for clarity.
FIG. 8 Schematic of Pop-in Magnitude Evaluation

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
14
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
1 2 C o· S Pn 2 yn
v n 1x n D (17)
where:
n = sequential number (see Fig. 8) of the last of the
particular series of pop-ins being assessed (When only
one pop-in occurs, n = 1.),
v1 = elastic displacement at pop-in No. 1 (see Fig. 8),
Pn = force at the nth pop-in, and
νn = elastic displacement at the nth pop-in. (νn may be
determined graphically or analytically, see Fig. 8).
yn = force drop at the nth pop-in, and
xn = displacement increase at the nth pop-in. 8
For a single pop-in, if S
12C o · D
P 1 2y 1
v 1 1x 1
$0.02
the pop-in shall be considered significant. When multiple
S P 2y
D
pop-in events occur, 12C o · v i1x i shall be assessed sequen-
i i
tially for each pop-in. The KJc value corresponding to the
S D
P 2y
pop-in that causes 12C o · v i1x i to exceed 0.02 shall be used
S D
i i
P 2y
as the test result. If 12C o · v n1x n ,0.02, then the KJc value at
n n
the final specimen fracture shall be used as the test result.
9.3 Outlier—Occasionally, an individual KJc datum will
appear to deviate greatly from the remainder of the data set. It
may be possible to reduce the influence of the outlier datum on
KJc(med) by testing additional specimens. However, no valid KJc
data shall be discarded from the data utilized to calculate
KJc(med) unless justification is provided by the tester that this
data is not representative of the intended test material.
10. Data Analysis and Evaluation of the Reference
Temperature, To
10.1 Evaluation of Data Sets Based on Weibull Model:
10.1.1 Test Replication—A data set consists of at least six
uncensored KJc test results.
10.1.2 Relationship between the Scale Parameter, Ko and
KJc Test Results—The three-parameter Weibull model is used
to define the relationship between individual KJc results and the
cumulative probability for failure, pf. The term pf is the
probability for failure at or before KJc for an arbitrarily chosen
specimen taken from a large population of specimens. Data
samples of six or more specimens are used to estimate the true
value of scale parameter, Ko, for the following Weibull model:
p f 5 1 2 exp$ 2 @ ~ K Jc 2 20! / ~ K o 2 20! # 4 % (18)
10.1.3 Ferritic steels with yield strengths ranging from 275
to 825 MPa (40 to 120 ksi) will have fracture toughness
cumulative probability distributions of nearly the same shape,
independent of specimen size and test temperature. Scale
parameter, Ko, is the data fitting parameter determined when
8
Although an individual pop-in may be ignored on the basis of these criteria, this
does not necessarily mean that the lower bound of fracture toughness has been
measured. For instance, in an inhomogeneous material such as a weld, a small
pop-in may be recorded because of fortuitous positioning of the fatigue precrack tip.
Thus, a slightly different fatigue precrack position may give a larger pop-in, which
could not be ignored. In such circumstances the specimens should be sectioned after
testing, and examined metallographically to ensure that the crack tips have sampled
the weld or base metal region of interest (26).
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
15
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
using the maximum likelihood statistical method of data fitting
(27). When KJc and Ko in Eq 18 are equal, pf = 0.632.
10.1.4 Size Effect Predictions—The statistical weakest-link
theory is used to model specimen size effects in the transition
range between lower shelf and upper shelf fracture toughness.
The following relationship shall be used to size-adjust indi-
vidual KJc values, KJc(med), or Ko. KJc serves as the example
case:
K Jc~ x ! 5 201 @ K Jc~ o ! 2 20# S D
Bo
Bx
1/4
(19)
where:
KJc(x) = KJc for a specimen size Bx,
KJc(o) = KJc for a specimen size Bo,
Bo = gross thickness of test specimens (side grooves
Bx
ignored), and
= gross thickness of prediction (side grooves ignored).
10.2 Computing To from KJc Test Results:
10.2.1 Data Censoring—Replace all censored KJc values
with appropriate KJc limit values (8.9.2). If censoring is
required due to violation of KJclimit, Eq 1, the experimental KJc
value shall be replaced by KJclimit for the specimen sizes used.
Determine KJclimit at each test temperature using the material
yield strength corresponding to that temperature. If censoring
is required due to violation of KJc∆a, the KJc test value shall be
replaced with the highest uncensored KJc in the data set
obtained at any specimen size and test temperature because
KJc∆a should be size independent and also largely insensitive to
test temperature. The KJIc value defined in E1820 can also be
used for KJc∆a, if JIc is known for the test material. As specified
in 8.9.2, if both KJclimit and KJc∆a are violated, replace the KJc
test value with the lower of the two limits.
10.2.2 Size Correction of KJc Data—If the data are gener-
ated from specimens of other than 1T size, all data, including
uncensored and censored values, shall be converted to 1T size
equivalence using Eq 19 (see 3.3.20). For determining KJc(1T)
using Eq 19, Bo and KJc(o) are the measured specimen thickness
and KJc test result (either censored or uncensored),
respectively, and Bx = 25.4 mm.
10.2.3 Calculation of Provisional To Value, ToQ—After
censoring the KJc input values and converting the uncensored
or censored KJc results to 1T equivalence using Eq 19, the
following equality shall be used to determine the provisional
ToQ using an iterative procedure (27, 28).
N
exp@ 0.019 ~ T i 2 T oQ! #
(δ (20)
i51
i
11.0176.7exp@ 0.019 ~ T i 2 T oQ! #
N
~ K Jc~ i ! 2 20! 4 exp@ 0.019 ~ T i 2 T oQ! #
2 ( $ 11.0176.7exp@ 0.019 ~ T 2 T
i51 i oQ !#%5
50
where:
N = number of specimens tested,
Ti = test temperature corresponding to KJc(i),
KJc(i) = either an uncensored KJc datum or a datum replaced
with a censoring value (8.9.2),
δi = 1.0 if the datum is uncensored or zero if the datum is
a censored value,
 E1921 − 19

11.0 = approximation of 10/(ln2)1/4 MPa√m to 3 significant where:

digits, and r = number of uncensored data as determined in 8.9.2, and
76.7 = approximation of 70/(ln2)1/4 MPa√m to 3 significant N = total number of uncensored and censored data.
digits.
K Jc~ med! 5 201 ~ K o 2 20! @ ln~ 2 ! # 1/4 , MPa =m (23)
Solve Eq 20 for ToQ temperature by iteration.
10.3 Requirement for Size of Data Set—Data generated at and
test temperatures in the range of To - 50°C to To - 14°C are
considered to make reduced accuracy contribution to To T oQ 5 T 2 S 1
0.019
lnD S~ K Jc~ med! 2 30!
70 D (24)
determinations. As a consequence, more data development
within the aforementioned temperature range is required. The 10.4.1 Single Temperature Test Sample Size Requirements—
following weighting system specifies the required number of For the special case where all tests are conducted at a single
data: temperature, Eq 21 and Table 5 in 10.3, which provide the
3
number of uncensored KJc test results required to evaluate To,
can be simplified according to Table 6. If KJc(med) of a data set
(r
i51
i ni $ 1 (21)

where ri is the number of uncensored data within the i-th TABLE 6 Number of Uncensored KJc Test Results Required to
temperature range, (T−To), and ni is the specimen weighting Evaluate To
factor for the same temperature range as shown in Table 5. Number of
Possible number
A uncensoredKJc
(T − To) range KJc(med) range of censored KJc
results
results by Eq 1B
required
TABLE 5 Weighting Factors for Multi-Temperature Analysis (°C) (MPa=m)

(T − To) rangeA 1T KJc(med) rangeA Weighting factor 50 to –14 212 to 84 6 3

–15 to –35 83 to 66 7 1
(°C) (MPa=m) ni –36 to –50 65 to 58 8 0
50 to −14 212 to 84 1/6 A
Convert KJc(med) to 1T equivalence using Eq 19. Round off to nearest whole digit.
−15 to −35 83 to 66 1/7 B
Established specifically for precracked Charpy specimens. Use this column for
−36 to −50 65 to 58 1/8
total specimen needs.
A
Rounded off to the closest integer.

is lower than 58 MPa√m, then the To determination using that

data set shall not be allowed.
10.3.1 Since the valid test temperature range is only known See Appendix X1 and Appendix X2 for example calcula-
after ToQ has been determined, the following iterative scheme tions.
may be helpful for identifying proper test temperature. Choose
10.5 Validation of ToQ as To: To = ToQ if all of the following
an initial test temperature as described in 8.4.1 using the value
requirements are met:
of “C” appropriate for the test specimen size. Conduct tests at
10.5.1 The apparatus requirements of Section 6 are met or
this temperature to obtain 3–4 uncensored results. Evaluate an
exceeded,
estimated ToQ value (ToQ(est)) value using Eq 20. Base all
10.5.2 The specimen configuration and dimensions meet the
subsequent test temperatures on ToQ(est). See Appendix X3 for
requirements of Section 7,
an example calculation.
10.5.3 The specimen precracking was completed within the
10.3.2 Certain data sets may result in an oscillating iteration
requirements of 7.8,
between two (or more) distinct ToQ values upon satisfying the
10.5.4 The specimens were tested according to the require-
ToQ 6 50°C limit required by 10.5.5. In these instances, the
ments of Section 8, including qualification of the data accord-
ToQ value reported shall be the average of the calculated
ing to 8.9, and
values. One example is for hypothetical data with toughness
10.5.5 The number of specimens tested within the allowable
values such that the initial ToQ estimation requires that data at
temperature range ToQ 6 50°C, meets the requirements of 10.3
one temperature be excluded. The second iteration then results
or 10.4.1 as applicable.
in the inclusion of this same data. Subsequent ToQ iterations
will then oscillate between the original first and second 10.6 Homogeneity Screening Procedure:
estimations. This phenomenon is more likely for sparse data 10.6.1 The following screening procedure, based on the
sets when test results exist near the ToQ 6 50°C limit. More SINTAP method (29), shall be followed to determine if the data
testing near the average ToQ will likely resolve this problem. set may be representative of a macroscopically inhomogeneous
material. If the material fails the screening criterion, the
10.4 Single Temperature Analysis—In the special case that
homogeneity of the material and its fracture toughness can be
all tests are conducted at a single temperature, T, the iterative
more accurately assessed using the analysis methods described
solution of Eq 20 can be replaced with a direct evaluation of
in Appendix X5.
Ko, KJc(med), and ToQ using the following relationships:
10.6.2 Screening Procedure:
Ko 5 F( ~
N

i51
K Jc~ i ! 2 20! 4
r G 1/4

120, MPa =m (22)

10.6.2.1 Determine KCENSi at each test temperature (Ti) for
the data set using:

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
16
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
K CENSi 5 30170·exp@ 0.019 ~ T i 2 T o ~ step1 ! ! # MPa=m (25) design curves (27, 30). For this method, the shape of the
median KJc toughness, KJc(med), for 1T specimens (3.3.20) is
where:
described by:
To(step1) = ToQ or To, as appropriate (that is, the reference
temperature as determined in 10.5 for the data set) K Jc~ med! 5 30170 exp@ 0.019~ T 2 T o ! # , MPa=m (27)

10.6.2.2 Compare each 1T-adjusted KJc value in the data set where:
(KJci) with the KCENSi value at the corresponding Ti. If KJci ≥ T = test temperature (°C), and
KCENSi, then KJci = KCENSi and δi = 0 for that data point. If KJci To = reference temperature (°C).
< KCENSi, then δi = 1 and KJci retains its original value. If δi =
The Weibull scale parameter, Ko, is given by:
0, the data point is censored, while if δi = 1, the data point is
uncensored in the subsequent analysis. @ K Jc~ med! 2 20#
Ko 5 120, MPa =m (28)
10.6.2.3 Using the new data set determined in 10.6.2.2, @ ln ~ 2 ! # 1⁄4
calculate a revised reference temperature (To(step2)) in accor- 10.8 Standard Deviation—The standard deviation of the
dance with 10.2. fitted Weibull distribution is a mathematical function of
10.6.2.4 If To(step2) ≥ To(step1) + 0.5 °C, recalculate the KJc(med) and is given by:
KCENSi at each test temperature using Eq 25, with To(step2) in
σ 5 0.28K Jc~ med! @ 1 2 20 ⁄ K Jc~ med! # (29)
place of To(step1). Repeat 10.6.2.2 and 10.6.2.3 to determine the
next revised reference temperature (To(step3)). 10.9 Tolerance Bounds—Upper and lower tolerance bounds
10.6.2.5 Continue the iteration procedure described in for the master curve can be calculated using the following
10.6.2.1 through 10.6.2.4, determining individual values of the equation:
revised reference temperature (To(stepi)), until To(stepn) -
To(stepn-1) < 0.5°C, where To(stepn) is the nth revised reference
temperature. Less than 5 iterations are typically required to
K Jc~ 0.xx! 5 201 ln FS 1
1 2 0.xx DG 1/4

$ 11177 exp@ 0.019~ T 2 T o ! # %

(30)
satisfy the above inequality.
10.6.2.6 Set Toscrn equal to the maximum value of the where 0.xx represents the selected cumulative probability
individual To(stepi) values. level; for example, for the 2 % tolerance bound, 0.xx = 0.02.
10.6.3 Screening Criterion:
10.10 Margin Adjustment—The margin adjustment is an
10.6.3.1 The data set can be considered to be representative upward temperature shift of the tolerance bound curve, Eq 30.
of a material that is macroscopically homogeneous, if: Margin is added to cover the uncertainty in To that is associated

T oscrn 2 T o ~ step1 ! # 1.44 Œ β2

r
(26)
with the use of only a few specimens to establish To. Additional
uncertainty can result from other sources including material
inhomogeneity and experimental errors. The standard devia-
where: tion on the estimate on To which incorporates sample size and
β = sample size uncertainty factor defined in 10.10 (that is, experimental uncertainties is given by:
corresponding to To(step1)), and
r = total number of uncensored data defined in 10.3 (that is
, corresponding to To(step1)).
σ5 Œ β2
r
1σ 2exp (31)

10.6.3.2 If the inequality in Eq 26 is not fulfilled, the data where:

set may be representative of a macroscopically inhomogeneous β = sample size uncertainty factor,
material, and the data set can be analyzed using the procedures r = total number of uncensored data used to establish the
of Appendix X5 to more accurately characterize the material. value of To, and
10.6.4 The confidence that the screening criterion has cor- σexp = contribution of experimental uncertainties. If standard
rectly identified the material as macroscopically inhomoge- calibration practices are followed, σexp = 4°C may be
neous increases as the number of data points in the data set used.
increases, as the differences in the fracture toughness between 10.10.1 An equivalent value of the median toughness for a
lower and higher-toughness material constituents contained data set, K Jceq~ med! , is defined as (31):
within the data set increases, and as the relative amounts of
these lower and higher-toughness constituents become more
equivalent. However, it is not possible to have reasonable
eq
K Jc ~ med! 5 F 1
r
r

( 30170 exp~ 0.019

i51
~ T i 2 T o !! G (32)

confidence in this screening criterion unless the data set When K Jceq~ med! is equal to or greater than 83 MPa√m, β = 18°C
contains at least 20 KJc values. An example demonstrating (32). If the 1T equivalent K Jceq~ med! is below 83 MPa√m, values of
application of the screening procedure is provided in Appendix β must be increased according to the following table:
X5.
eq
K Jcs medd
β
10.7 Establishment of a Transition Temperature Curve 1T equivalentA (°C)
(Master Curve)—Transition temperature KJc data tend to con- (MPa√m)
form to a common toughness versus temperature curve shape 83 to 66 18.8
65 to 58 20.1
in the same manner as the ASME Klc and KIR lower-bound

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
17
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
A eq
Round off K Jcs medd to the nearest whole number.
10.10.2 To estimate the uncertainty in To, a standard two-
tail normal deviate, Z, shall be taken from statistical handbook
tabulations. The selection of the confidence limit for To
adjustment is a matter for engineering judgment.
10.11 Uses for Master Curve—The master curve can be
used to define a transition temperature shift related to metal-
lurgical damage mechanisms. Fixed values of Weibull slope
and median KJc define the standard deviation; hence the
representation of data scatter. This information can be used to
calculate tolerance bounds on toughness for the specimen
reference size chosen. The data scatter characteristics modeled
here can also be of use in probabilistic fracture mechanics
analysis, bearing in mind that the master curve pertains to a 1T
size specimen. The master curve determined by this procedure
pertains to cleavage fracture behavior of ferritic steels. Exten-
sive ductile tearing beyond the censoring limits set in 8.9.2
may precede cleavage as the upper-shelf range of temperature
is approached. Such data can be characterized by separate
methods (see Test Method E1820).
11. Report
11.1 Report the following information for each specimen:
11.1.1 Specimen type, specimen thickness, B, net thickness,
BN, specimen width, W,
11.1.2 Crack plane orientation according to Terminology
E1823,
11.1.3 Number of uncensored data, r, and number of speci-
mens tested at each temperature,
11.1.4 Crack pop-in and compliance ratio, Ci/Co, if
applicable,
11.1.5 Material yield strength and tensile strength, at each
test temperature,
11.1.6 The location of displacement measurement used to
obtain the plastic component of J (load-line, front-face, or
crack-mouth),
11.1.7 A list of individual data for each specimen including:
11.1.7.1 KJc(i), KJclimit, KJc∆a, ao(i), T(i), and the extent of
visually measured slow-stable crack growth prior to the onset
of cleavage, if present,
11.1.7.2 Difference between maximum and minimum initial
crack size expressed as a percentage of the initial specimen
thickness, B,
11.1.8 Weibull scale parameter, Ko, the master curve refer-
ence temperature, To (°C) or To,X (°C), Toscrn, and whether the
material is classified as homogeneous or inhomogeneous as per
10.6.3 (that is, the material can be considered homogeneous if
the inequality in Eq 26 is satisfied).
11.1.9 Weibull scale parameter, Ko and the master curve
reference temperature, To (°C) or To,X (°C), and
11.1.10 Fatigue precracking Kmax for the final precracking
step (see 7.8.2).
11.1.11 Values for the nominal elastic modulus, E, used to
convert Jc to KJc at each test temperature.
11.1.12 Values for Ei and the associated calculated crack
size for each test from 8.8.2.
11.2 The report may contain the following supplementary
information:
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
18
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
11.2.1 Specimen identification codes,
11.2.2 Measured pop-in crack extensions for applicable test
results,
11.2.3 Provisional value ToQ (°C) and reason for invalidity,
if applicable, and
11.2.4 Force-displacement records.
12. Precision and Bias
12.1 Precision—The precision of To measurements has been
examined through an interlaboratory round-robin study with
nine participating laboratories. A nuclear grade pressure vessel
weld metal was tested using precracked Charpy-type (PCC)
specimens (33). All the values of To used for the calculations
satisfy the validity requirements of this test method. Values of
To were originally calculated and reported by participants using
the single-temperature approach (that is, one value of To per
test temperature). However, test results obtained by each
laboratory at different temperatures have also been analyzed
using the multi-temperature approach (that is, one value of To
per laboratory). While the data for this interlaboratory study do
not strictly conform to the requirements of Practice E691 (that
is, the reported reference temperatures and their standard errors
do not represent typical means and standard deviations based
on repeated measurements), alternative definitions that satisfy
the intent of Practice E691 have been used. The results are
summarized in 12.1.1.2 and 12.1.2.2. For more details, see
ASTM Research Report RR:E08-1008.9
12.1.1 Analysis of Single-Temperature Estimates of To:
12.1.1.1 Justification and Theoretical Background—For the
analysis of single-temperature estimates of To, laboratories
reported values of To and their associated standard errors based
on Eq 31 at different test temperatures. The number of test
temperatures varied for each laboratory. The arithmetic average
of the single To values is the laboratory’s estimated reference
temperature. In addition, the standard errors associated with
each single To value were combined to obtain an estimate of
the standard deviation for each laboratory. The use of Practice
E691 for computing repeatability and reproducibility standard
deviations for this case is not applicable due to the differing
number of test temperatures for each laboratory. Instead, the
repeatability standard deviation and the reproducibility stan-
dard deviation are defined as follows. The repeatability stan-
dard deviation is the square root of the average of the estimated
laboratory variances obtained from Eq 31. The reproducibility
standard deviation is the standard deviation of the reference
temperatures reported for each laboratory. For the analysis of
single-temperature estimates of To, the reported values of
repeatability and reproducibility standard deviations provide
some general insights, but do not strictly correspond to the
definitions provided in Practice E691.
12.1.1.2 Results—The statistics reported in Table 7 were
calculated for the single-temperature (that is, single-T) ap-
proach from test results obtained at various temperatures.
Individual To values are based on the number of uncensored
data points reported by each laboratory at each test tempera-
ture. The number of uncensored data points ranged from 6 to
9
Supporting data have been filed at ASTM International Headquarters and may
be obtained by requesting Research Report RR:E08-1008.
 E1921 − 19
TABLE 7 Precision using PCC Specimens (Single-T Analysis)
Parameter Type Number Average Repeatability Reproducibility Repeatability Reproducibility
of of Standard Standard Limit, Limit,
Analysis Determinations Deviation, Deviation, r R
Sr SR

To, °C Single-T 9 -75 7.8A 7.8B 21.8 21.8

A
The Repeatability Standard Deviation used in this analysis does not strictly correspond to the repeatability standard deviation as defined in Practice E691. Standard errors
for each laboratory are given in ASTM Research Report RR:E08-1008.
B
The provisional reproducibility standard deviation is defined as the standard deviation of the laboratory averages. Since the provisional reproducibility standard deviation
of 6.3 °C is less than the repeatability standard deviation, the “Reproducibility Standard Deviation” is set to be equivalent to the repeatability standard deviation. The
reproducibility standard deviation used in this analysis does not strictly correspond to the reproducibility standard deviation as defined in Practice E691.

11. Note that the statistics reflecting the precision of the test
method from this interlaboratory study (Table 7) also incorpo-
rate any variability resulting from estimating To at different test
temperatures.
NOTE 4—Repeatability and reproducibility limits have been calculated
by multiplying the respective standard deviations by 2.8, as recommended
by Practice E177. Repeatability and reproducibility limits are considered
general guides, and the associated 95th percentiles provide an estimate of
the differences that may be expected when comparing test results from
laboratories similar to those in the study.
12.1.2 Analysis of Multi-Temperature Estimates of To:
12.1.2.1 Justification and Theoretical Background—For the
analysis of multi-temperature estimates of To, each laboratory
provided a single value of the reference temperature based on
differing numbers of uncensored test results, and Eq 31 from
this test method was used to estimate the standard error of the
reported reference temperature. Although the data do not
strictly comply with Practice E691, the definitions of repeat-
ability standard deviation and reproducibility standard devia-
tion from Appendix X1.1 of Practice E691 remain applicable.
It is assumed in this analysis that the standard error of the
reference temperature, Eq 31, is equal to the standard deviation
of the measurement. Because the standard errors of the
reference temperatures were similar, they could be combined to
obtain the repeatability standard deviation.
12.1.2.2 Results—The statistics reported in Table 8 were
calculated for the multi-temperature (multi-T) approach. Eq 31
of this Test Method was used to determine the repeatability
standard deviation n, and reproducibility standard deviation
was computed using Practice E691 Eq 8, where n=1.
NOTE 5—Repeatability and reproducibility limits have been calculated
by multiplying the respective standard deviations by 2.8, as recommended
by E177. Repeatability and reproducibility limits are considered general
guides, and the associated 95th percentiles provide an estimate of the
differences that may be expected when comparing test results from
laboratories similar to those in the study.
12.1.3 The terms repeatability and reproducibility are used
as specified in Practice E177.
12.2 Bias—Since there is no accepted reference material,
method, or laboratory suitable for determining the bias in T0
using the procedure in this test method, no statement of bias is
being made.

TABLE 8 Precision using PCC Specimens (Multi-T Analysis)

Parameter Type Number Average Repeatability Reproducibility Repeatability Reproducibility
of of Standard Standard Limit, r Limit, R
Analysis Determinations Deviation, Sr Deviation, SR

To, °C Multi-T 9 -74 5.7A 7.6 16.0 21.3

A
These values do not strictly correspond to the repeatability standard deviation as defined in Practice E691. Individual within-laboratory standard deviations for each
laboratory are given in ASTM Research Report RR:E08-1008.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
19
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
ANNEX
(Mandatory Information)
A1. SPECIAL REQUIREMENTS FOR DETERMINING THE REFERENCE TEMPERATURE, To,X, AT ELEVATED LOADING
RATES
A1.1. Scope
A1.1.1 This annex covers the determination of the rate-
dependent reference temperature, To,X, under conditions where
the loading rate exceeds the limit allowed for conventional
quasi-static loading in 1.5 and when the minimum test time is
greater than the limit defined in Test Method E1820, Annex
A14 “Special Requirements for Rapid-Load J-integral Fracture
Toughness Testing”.
A1.1.2 This international standard was developed in accor-
dance with internationally recognized principles on standard-
ization established in the Decision on Principles for the
Development of International Standards, Guides and Recom-
mendations issued by the World Trade Organization Technical
Barriers to Trade (TBT) Committee.
A1.2. Terminology
A1.2.1 Definitions of Terms Specific to This Standard:
A1.2.1.1 reference temperature,To,X—Reference tempera-
ture analogous to To determined from testing performed at a
rate higher than the quasi-static range specified in 1.5. The
index X corresponds to the logarithm of the average loading
rate, log(K̇I), (where K̇I is expressed in MPa√m/s) for all tests
performed, rounded to the nearest integer.
A1.2.1.2 test loading rate K̇I [FL-3/2T-1]—rate of increase of
applied stress intensity factor.
A1.2.1.2.1 Discussion—It is generally evaluated as the ratio
between KJc and the corresponding time to cleavage. For tests
where partial unloading/reloading sequences are used to mea-
sure compliance, an equivalent time to cleavage tc shall be used
to calculate the loading rate. The value of tc is calculated as the
ratio between the value of load-line displacement at cleavage
and the load-line displacement rate applied during the mono-
tonic loading portions of the test (that is, the periods between
partial unloading/reloading sequences used for compliance
measurement).
A1.2.1.3 Elevated loading rate test—A test where the speci-
men loading rate in terms of K̇I exceeds 2 MPa√m/s.
A1.2.1.4 Estimated reference temperature, To,Xest [°C]—
estimated value of the reference temperature corresponding to
an elevated loading rate, X, to be used only for test temperature
selection in accordance with A1.4.2.
A1.3 General Considerations
A1.3.1 This annex describes how to obtain data measured at
a nominally constant loading rate, K̇I, from tests conducted at
either multiple or single temperature and calculates a rate
dependent measure of To, called To,X, where X corresponds to
the logarithm of the average loading rate, log(K̇I ), (where K̇I is
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
20
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
expressed in MPa√m/s) for all tests performed, rounded to the
nearest integer. K̇I values are considered to be nominally
constant if they deviate by no more than a factor of 3 from the
average K̇I value.
A1.3.2 The summary of test method (Section 4) and signifi-
cance and use (Section 5) are applicable to elevated loading
rate data. The apparatus (Section 6), and specimen
configuration, dimensions, and preparation (Section 7) are also
generally applicable. However, if the time taken to reach Pm is
less than 0.1 minutes as stipulated in Test Method E1820,
Annex A14, additional considerations pertaining to the test
apparatus in Test Method E1820, Annex A14 are also appli-
cable and modifications to the test specimens may be necessary
to accommodate the test apparatus as discussed in Test Method
E1820, Annex A14. For testing precracked Charpy-type speci-
mens using an instrumented impact pendulum machine, the
requirements of Test Method E1820, Annex A17, “Fracture
Toughness Tests at Impact Loading Rates using Precracked
Charpy-Type Specimens”, apply.
A1.4. Procedure
A1.4.1 General—The procedures described in Section 8 for
quasi-static testing are applicable, unless superseded by the
procedures in this section. To evaluate To,X, the loading rate for
all tests shall be of the same order of magnitude (that is,
correspond to the same X) and shall be higher than the
quasi-static range specified in 1.5.
A1.4.2 Initial Test Temperature Selection—If the value of To
under quasi-static loading rate conditions is known, the fol-
lowing relationship can be used to derive an estimated value of
To,X (To,Xest) to facilitate test temperature selection (34).
est
~ T o 1273.15! ·Γ
T o,X 5 • 2 273.15 (A1.1)
~ !
Γ 2 ln K I
where:
K̇I = in MPa√m/s and To is in °C.
The function Γ is given by:
Γ 5 9.9·exp FS T o 1273.15
190 D S D G
1.66
1
σ ys
722
1.09
(A1.2)
σys = yield strength measured (as per Test Methods E8/E8M)
or estimated at room temperature for quasi-static rates
(~ 10-6 to 10-4 s-1).
Eq A1.1 and Eq A1.2 shall not be used for calculating and
reporting values of reference temperatures corresponding to
elevated loading rates.
A1.4.2.1 While testing is allowed at any temperature within
To,X 6 50°C, the most accurate estimates of To,X are obtained
 E1921 − 19
by testing as close to To,X as possible. One recommended
strategy is to conduct initial testing at To,Xest, determine an
initial value of To,X, and then conduct subsequent testing as
close as possible to To,X to increase the overall accuracy of the
To,X estimation.
A1.4.3 Testing Procedure—If the time taken to reach Pm is
greater than 0.1 min., specimen testing shall be conducted as
per Test Method E1820. Testing at higher loading rates, shall
be conducted as per Test Method E1820, Annex A14 or as per
E1820, Annex 17 for precracked Charpy-type specimens tested
at impact loading rates with an instrumented striker. The
corresponding requirements of Test Method E1820, Annex
A14, or Annex17, as applicable, shall be fulfilled in order for
a test result to be considered valid.
A1.4.3.1 Testing shall proceed beyond the point of speci-
men instability due to cleavage to be considered a valid test. If
a single or multiple pop-ins occur prior to the final specimen
instability point, the significance of the pop-ins should be
evaluated using 8.9.5 and 9.2 to determine the point in the
load-displacement record used to calculate JQc(t) for this
method.
A1.4.3.2 After the test, measure the initial crack size and the
extent of any slow stable crack growth prior to either final
specimen instability or a significant pop-in event (9.2), as
applicable, as per 8.8.1.
A1.4.3.3 Determine the specimen compliance from the test
record using the method and requirements from Test Method
E1820, Annex 14; or E1820, Annex 17, as applicable. Then,
estimate the initial crack size based on this compliance using
Test Method E1820, equation A1.12 for SE(B) specimens, Test
Method E1820, equation A2.12 for C(T) or Test Method
E1820, equation A3.11 for DC(T) specimens. Compare, and
adjust if needed, the estimated initial crack size with the
measured initial crack size as described in 8.8.2.
A1.4.3.4 Calculate JQc(t) for each specimen in the data set
at the point of instability as per either Test Method E1820,
Section 9 or Test Method E1820, Annex 14; or E1820, Annex
17, as applicable for the test loading rate and test method
chosen. The J-estimation formulas for the basic method pro-
vided in Test Method E1820, Annex A1 – A3 for each
applicable specimen type shall be used.
A1.4.4 Qualification of JQc(t) Data—For JQc(t) to be a valid
Jc(t), the following requirements shall be met:
A1.4.4.1 Requirements on testing equipment in Test
Method E1820, Section 6 unless superseded by requirements in
Test Method E1820, Annex 14; or Test Method E1820, Annex
17, at which point the requirements in the applicable annex
shall apply.
A1.4.4.2 Requirements pertaining to specimen
configuration, dimensions, and machining tolerances in Test
Method E1820, Section 7.
A1.4.4.3 Specimen precracking requirements in 7.8.
A1.4.4.4 Requirements on fixture alignment in Test Method
E1820, Section 8.
A1.4.4.5 Requirements on specimen test temperature con-
trol and measurement in 8.6.
A1.4.4.6 Requirements on the accuracy of the post-test
check of the estimated crack size in 8.8.2.
Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
21
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
A1.4.4.7 Requirements on initial crack size measurement in
8.9.1.
A1.4.4.8 For testing conducted as per Test Method E1820,
the requirements on test rate in Test Method E1820, Section 8.
A1.4.4.9 For testing conducted as per Test Method E1820,
Annex A14, the requirements in Test Method E1820, Annex
A14 on qualification of data.
A1.4.4.10 For testing conducted as per Test Method E1820,
Annex A17, the requirements in Test Method E1820, Annex
A17 on specimen size, configuration, and preparation; on
apparatus; and on test procedures and measurements that are
relevant for KJcd values.
A1.4.5 Determination of KJc(t) data for To,X calculation.
A1.4.5.1 Calculate KJc(t) for each Jc(t) datum using Eq 16
in 9.1.5.
A1.4.5.2 KJc(t) data that exceed the KJclimit or KJc∆a require-
ments specified in 8.9.2 are considered to be censored data that
are still analyzed as described in 10.2.1.
A1.4.5.2.1 The use of the material’s quasi-static yield
strength in the KJclimit equation (Eq 1 in 7.5) is conservative
since the quasi-static yield strength will always be lower than
the dynamic yield strength. Alternatively, the dynamic yield
strength can be measured at the strain rate corresponding to K̇I
for the fracture test and used in the KJclimit equation. The
corresponding dynamic tensile testing strain rate, ε̇ , is calcu-
lated by (35, 36):
•
• 2σ YS K I
ε5
¯ ·E
(A1.3)
K
Jc
σYS = average quasi-static yield strength at the test tempera-
ture
E = Young’s modulus at the test temperature,
¯ = average cleavage toughness value of the elevated rate
K Jc
tests
K̇I = average loading rate of the elevated rate tests.
The tensile test shall then be conducted at the strain rate
calculated by Eq A1.3 to obtain the dynamic yield strength.
A1.4.6 Calculation of TQo,X—Calculate a provisional To,X
(TQo,X) according to Section 10 , where X corresponds to the
logarithm of the average loading rate, log(K̇I), (where K̇I is
expressed in MPa√m/s) for all tests performed, rounded to the
nearest integer. For example, if the average calculated loading
rate is 3×104 MPa√m/s, the corresponding reference tempera-
ture shall be designated To,4. As indicated in A1.3.1, the
loading rate for all specimens used in the calculation must
deviate by no more than a factor of 3 from the average K̇I value
for the data set.
A1.4.7 Qualification of TQo,X—The provisional value of
TQo,X is equal to To,X if the following requirements are met.
A1.4.7.1 The number of test specimens meets the minimum
data set size requirements in Table 5.
A1.4.7.2 The test temperatures, T, for all test specimens
satisfy -50°C ≤ T –To,X ≤ 50 °C.
A1.4.7.3 Once To,X has been determined, the provisions in
10.7, 10.8, 10.9, and 10.10 can be followed to determine a
transition temperature curve, standard deviation, tolerance
 E1921 − 19
bounds, and a margin adjustment for To,X. The uses of the
transition temperature curve in 10.11 are also applicable.
A1.5. Report
A1.5.1 The report shall include the information required in
Test Method E1820, 10.2.1 – 10.2.7, and 10.2.9.
A1.5.2 If testing was conducted as per Test Method E1820,
Annex A14 or Test Method E1820, Annex A17, the report shall
include the reporting information required in the applicable test
method.
A1.5.3 The report shall include the information required in
Section 11.1 and may include the supplementary information in
11.2.
A1.5.4 The report shall also include the following:
A1.5.4.1 Load-line displacement rate ∆˙ LL, and
A1.5.4.2 Test loading rate, K̇I
A1.5.5 The report may contain the following supplementary
information:
A1.5.5.1 Force-time and force-displacement records.
A1.6. Precision and Bias
A1.6.1 An interlaboratory study (33) for the determination
of the master curve reference temperature, To, from precracked
Charpy-type specimens tested at impact loading rates, was
conducted in ten laboratories using a reactor pressure vessel
steel denominated JRQ (ASTM A533B Cl.1).10
A1.6.1.1 Justification and Theoretical Background—The
data analyzed for this interlaboratory study do not strictly
conform to the requirements of Practice E691 because the
reported reference temperatures and their standard deviations
10
Supporting data have been filed at ASTM International Headquarters and may
be obtained by requesting Research Report RR:E08-1011. Contact ASTM Customer
Service at service@astm.org.
do not represent typical means and standard deviations based
on repeated measurements. Each laboratory provides a single
value of the reference temperature based on differing numbers
of valid measurements, and Eq 31 from this test method is used
to estimate the standard deviation of the reported reference
temperature. Although the data do not comply with Practice
E691, the definitions of repeatability and reproducibility from
Practice E691, Appendix X1.1 remain applicable. Each refer-
ence temperature is considered to be based on a sample of size
1 for analysis purposes. The standard deviations of the refer-
ence temperatures were very similar, so these estimates were
“pooled” to obtain the repeatability standard deviation.
A1.6.1.2 Statistical Analysis of the Interlaboratory Study
Results—The statistics reported in Table A1.1 were calculated
using the standard deviation of the laboratory reference tem-
peratures (when single values are reported for each laboratory,
this is the provisional reproducibility standard deviation) and
Eq 31 of this test method (repeatability standard deviation).
Note that the statistics in Table A1.1 also reflect any contribu-
tions from test temperature effects on the precision of the test
method. Individual To values are based on a number of valid
data points ranging from 7 to 10. One of the data sets was
excluded after being statistically classified as an outlier.
A1.6.1.3 The terms repeatability and reproducibility are
used as specified in Practice E177.
NOTE A1.1—Repeatability and reproducibility limits have been calcu-
lated by multiplying the respective standard deviations by 2.8, as
recommended by Practice E177. Repeatability and reproducibility limits
are considered general guides, and the associated 95 % probability is a
rough indicator of what can be expected when comparing two test results
from laboratories similar to those participating in the study.
A1.6.2 Bias—Since there is no accepted reference material,
method, or laboratory suitable for determining the bias in To
using the procedure described in this annex, no statement of
bias is being made.

TABLE A1.1 Precision using impact-tested PCC specimens

Parameter Number Average Repeatability Reproducibility Repeatability Reproducibility
of Standard Standard Limit, r Limit, R
Determinations Deviation, Sr Deviation, SR

To, °C 9 -0.1 7.40A 7.40B 20.7 20.7

A
The repeatability standard deviation is the square root of the pooled laboratory variances that were computed using Eq 31 of this test method. These values do not strictly
correspond to the repeatability standard deviation as defined in Practice E691. Individual within-laboratory standard deviations for each laboratory are given in ASTM
Research Report RR:E08-1011.
B
The provisional reproducibility standard deviation (the standard deviation of the reference temperatures reported by each laboratory) was estimated to be 5.63. Since the
reproducibility standard deviation is less than the repeatability standard deviation, the reproducibility standard deviation is set equal to the repeatability standard deviation
as specified in Practice E691.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
22
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

APPENDIXES

(Nonmandatory Information)

X1. WEIBULL FITTING OF DATA11

X1.1 Description of the Weibull Model: X1.3.2 The example data is artificially generated with the
X1.1.1 The three-parameter Weibull model is used to fit the following assumptions:
relationship between KJc and the cumulative probability for Material yield strength at test temperature = σYS = 482 MPa
E = 202 GPa, ν = 0.3
failure, pf. The term pf is the probability for failure at or before Test temperature T = 38°C
KJc for an arbitrarily chosen specimen from the population of Six 1/2T and six 1T specimens, all with a/W = 0.5
specimens. This can be calculated from the following: X1.3.3 KJclimit values in MPa√m from Eq 1 for 0.5T and 1T
p f 5 1 2 exp$ 2 @ ~ K Jc 2 K min! / ~ K o 2 K min! # b % (X1.1) specimens.
X1.1.2 Ferritic steels of yield strengths ranging from 275 to 0.5T 1T
825 MPa (40 to 120 ksi) will have fracture toughness distri- KJclimit 212.6 300.7
butions of nearly the same shape when Kmin is set at 20 MPa√m KJclimit 1T equivalent 182.0 300.7

(18.2 ksi√in.). This shape is defined by the Weibull exponent, X1.3.4 Simulated Data Set:
b, which is constant at 4. The scale parameter, Ko, is a Raw Data Size Adjusted
data-fitting parameter. The procedure is described in X1.2. (KJc, MPa=m) (KJc(1T), MPa=m)
1/2T 1T 1/2TA 1T
X1.2 Determination of Scale Parameter, Ko, and KJc(med)—
138.8 119.9 119.9 119.9
The following example illustrates the analysis of all uncen- 171.8 147.6 147.6 147.6
sored data tested at a single temperature. The data came from 195.2 167.3 167.3 167.3
(216.2) 185.0 (182.0) 185.0
tests that used 4T compact specimens of A533 grade B steel all (238.5) 203.7 (182.0) 203.7
tested at -75°C. None of the data require censoring so r = N = (268.3) 228.8 (182.0) 228.8
6, and the next step is to convert all data to the 1T equivalent A
KJc(1T) = (KJc(0.5T) − 20) (1/2 / 1)1/4 + 20 MPa=m
specimen size for analysis:
X1.3.5 Calculations using the censored, size-adjusted data
Rank KJc(4T) KJc(1T)
(i) (MPa=m) Equivalent
and the relationships of 10.4 with N = 12 and r = 9 yield: Ko
(MPa=m) = 190.0 MPa√m, KJc(med) = 175.1 MPa√m and To = ToQ =
1 59.1 75.3 -0.4 °C.
2 68.3 88.3
3 77.9 101.9 X1.3.6 Censoring When Ductile Crack Extension Limit is
4 97.9 130.2
5 100.9 134.4
Violated—The following example also illustrates the analysis
6 112.4 150.7 of censored data where all tests have been conducted at a single
test temperature of 38 °C. The test material has properties as
Eq 22 gives the Weibull scale parameter as: defined in X1.3.2 and toughness data as defined in X1.3.4.
Ko 5 F( ~
N

i51
K Jc~ i ! 2 20! 4
N G 1/4

120, MPa=m (X1.2)

However, for this example assume that the steel has a low
upper shelf. The maximum allowable crack growth (8.9.2) is
Since N = r = 6, Eq 22 gives Ko = 123.4 MPa√m. 0.64 mm for the 0.5T specimen and 1 mm for 1T specimen.
The median KJc is obtained from Eq 23 as: The raw data below provides the crack growth associated with
each KJc value.
K Jc~ med! 5 201 ~ K o 2 20! @ ~ 1n 2 ! 1⁄4 # 5 114.4 MPa=m (X1.3)
Raw Data Censored Data 1T Size
and the reference temperature (°C) is obtained from Eq 24: Adjusted Censored

T o 5 T oQ 5 T 2 S D F
1
0.019
ln
K Jc~ med! 2 30
70
G
5 284.8 °C 0.5T 1T 0.5T 1T 0.5TA
Data
1T
∆ap, KJc, ∆ap, KJc, KJc, KJc, KJc,
(X1.4) mm MPa=m mm MPa=m MPa=m MPa=m MPa=m
0.00 138.8 0.00 119.9 138.8 119.9 119.9 119.9
X1.3 Data Censoring Examples: 0.25 171.8 0.15 147.6 171.8 147.6 147.6 147.6
X1.3.1 Censoring When KJclimit is Violated—The following 0.50 195.2 0.20 167.3 195.2 167.3 167.3 167.3
0.67 (216.2) 0.55 185.0 (195.2) 185.0 (167.3) 185
example illustrates the analysis of censored data when all tests 0.70 (238.5) 1.10 (203.7) (195.2) (195.2) (167.3) (195.2)
have been conducted at a single test temperature. 0.71 (268.3) 1.15 (228.8) (195.2) (195.2) (167.3) (195.2)
A
KJc(1T) = KJc(0.5T) − 20) (0.5 / 1)1/4 + 20 MPa=m
11
The results in the example problems in Appendix X1 through Appendix X4 are As per 10.2.1, the highest KJc value at any specimen size that
provided to the nearest tenth place. While such precision is not supported by the
doesn’t exceed the maximum allowable crack growth is the
standard, it is provided so that analysis procedures can be verified. These results are
obtained by rounding values only when the results are reported and not during 0.5T specimen test with KJc = 195.2 MPa√m. Therefore KJca
intermediate steps in the calculation. = 195.2 MPa√m as per 10.2.1. Since KJca is less than KJc(limit)

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
23
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
for both the 0.5T and 1T specimens, KJca is the controlling
censoring value for both specimen sizes in this example.
X1.3.7 Calculations using the censored, size-adjusted data
and the relationships of 10.4 with N = 12 and r = 7 yield: Ko
= 189.0 MPa√m, KJc(med) = 174.2 MPa√m and To = ToQ = 0.0
°C.

X2. MASTER CURVE FIT TO DATA11

X2.1 The data set for this example is defined by: X2.3 Determine Ko, KJc(med) and ToQ using 10.4 to obtain:
X2.1.1 Six 0.5T compact specimens, Ko = 115.8 MPa√m,
KJc(med) = 20 + (Ko – 20)[ln(2)]1/4 = 107.4 MPa√m, and
X2.1.2 A 533 grade B base metal, and To = ToQ = -75 - ln[(107.4 - 30)/70]/0.019 = -80.3 °C.
X2.1.3 Test temperature, T = –75°C.
X2.4 Master Curve—The master curve of Eq 27 with To =
X2.2 In this data set, there are no censored data. -80.3 °C for this case is presented in Fig. X2.1 along with the
Rank KJc(1/2T) KJc(1T) 1T adjusted data.
Equivalent
(I) (MPa=m) (MPa=m)
1 91.4 80.0
2 103.1 89.9
3 120.3 104.3
4 133.5 115.4
5 144.4 124.6
6 164.0 141.1

FIG. X2.1 Master Curve for 1T Specimens Based on 0.5T Data Tabulated in X2.2

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
24
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

X3. EXAMPLE To DETERMINATION WITH DATA OBTAINED AT MULTIPLE TEST TEMPERATURES11

X3.1 Material and Specimen Geometries: T o ~ est! 5 T 28J 1C 5 25°218° 5 223°C

T o ~ est! 5 T 41J 1C 5 16°224° 5 28°C
A533 Grade B plate
Quenched and tempered Conduct four 1T SE(B) tests at −20°C.
900°C WQ (water quench); and 440°C (5 h) temper X3.6 To Estimation Procedure #2 from Results of First Four
Specimen types: Tests:
0.5T C(T) with ao/W = 0.5
1T SE(B) with ao/W = 0.5 and W = 2B First four test results at T = −20°C:
KJc, MPa=m
X3.2 Mechanical Properties: 135.1
108.9
Yield strength (room temperature): 641 MPa 177.1
Tensile strength (room temperature): 870 MPa 141.7
Charpy V-Notch Toughness: Calculate preliminary To(est)#2 from data to determine allow-
28-J temperature T28J = −5°C able test temperature range using 10.4:
41-J temperature T41J = 16°C
NDT: 41°C K Jc~ med! 5 136.6 MPa=m;
T 0 ~ est! #2 5 242.1°C
X3.3 Yield Strength, Elastic Modulus, and KJclimit Values: Estimated temperature range of usable data:
Yield strength values measured at several test temperatures are
5T 0 ~ est! #2 650°C
summarized in the table below along with elastic modulus (E)
5292.1°C # T i # 17.9°C
values and KJclimit values calculated at the same temperatures.
E values have been obtained from Eq 7 while KJclimit have been Now conduct additional testing within this range for To
obtained from Eq 1 for 1/2T and 1T specimens with a/W = 0.5 determination. Test data is presented in Table X3.1 along with
and ν = 0.3. the KJclimit values, determined in X3.3, for each test tempera-
ture and specimen size. Note. that for this data ductile tearing
Test Yield Elastic KJclimit
Temperature Strength modulus (MPa=m)
limits are not exceeded until T ≥ -5 °C. At these temperatures
(°C) (MPa) KJclimit < KJca, so only KJclimit censoring is required as per
(GPa) 1/2T 1T
8.9.2 and 10.2.1.
–130 840 212 288.1 407.4
–80 740 209 268.3 379.4
–50 700 207 260.0 367.6 X3.7 Calculation of ToQ —Based on To(est)#2, data is valid
−10 663 205 251.3 355.4 between −92.1 °C and 7.9 °C. Using data from -80°C to 0°C,
0 655 204 249.4 352.8
23 641 203 245.8 347.6
with N = 53 and r = 49, Eq 20 yields ToQ = -48.1 °C. Based on
this result, the valid test temperature range is –98.1 °C to
Values of yield strength at intermediate test temperatures to 1.9 °C. Since calculations were performed with data within this
those in this table are determined by linearly interpolating from range, no iteration is required.
the values in the table that bound the test temperature as per
7.5. E values at intermediate test temperatures are obtained X3.8 Qualified Data Summation:
from Eq 7 while the KJclimit values are obtained from Eq 1. (T − To) range Number of Weighting ri · n i
(°C) uncensored results, ri factor, ni
X3.4 Slow-stable Crack Growth Limits: The maximum al- 50 to −14 43 1/6 7.2
lowable crack growth is 0.64 mm for 1/2T specimens (8.9.2) −15 to −35 6 1/7 0.9
−36 to −50 0 1/8 0
and equates to the following censoring value that has been
determined from a J-R curve for the material that has been X3.9 Validity Check:
calculated as per Test Methods E1820: KJc(0.64 mm) = KJca =
255.0 MPa√m. Σr i n i 5 8.1.1.0
Therefore, T oQ 5 T o
X3.5 Estimation Procedure #1 from Charpy Curve:

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
25
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X3.1 Data Tabulation
KJc
Test Specimen
KJclimit (MPa=m)
temperature, δj
(MPa=m) Raw 1T
(°C) Type Size
data equivalent
−130 C(T) 1/2T 288.1 59.5 53.2 1
85.1 74.7 1
55.3 49.7 1
56.4 50.6 1
−80 C(T) 1/2T 268.3 51.3 46.3 1
87.9 77.1 1
113.4 98.5 1
−65 SE(B) 1T 373.5 73.9 73.9 1
126.8 126.8 1
−55 C(T) 1/2T 261.4 167.7 144.2 1
88.5 77.6 1
115.2 100.0 1
81.4 71.6 1
121.9 105.7 1
145.0 125.1 1
104.2 90.8 1
64.4 57.3 1
96.8 84.6 1
114.5 99.5 1
107.4 93.5 1
81.0 71.3 1
70.0 62.0 1
131.8 114.0 1
69.5 61.6 1
67.5 59.9 1
−30 C(T) 1/2T 255.6 102.3 89.2 1
194.0 166.3 1
170.4 146.5 1
129.5 112.1 1
118.2 102.6 1
147.9 127.5 1
178.8 153.5 1
95.9 83.8 1
−20 SE(B) 1T 358.5 135.1 135.1 1
108.9 108.9 1
177.1 177.1 1
141.7 141.7 1
174.4 174.4 1
84.8 84.8 1
132.1 132.1 1
−10 C(T) 1/2T 251.3 211.4 180.9 1
179.9 154.5 1
171.8 147.6 1
153.0 131.8 1
236.9 202.4 1
156.8 135 1
−5 C(T) 1/2T 250.4 121.5 105.3 1
194.2 166.5 1
110.4 96.0 1
197.0 168.8 1
134.7 116.5 1
(264.4) (213.7) 0
0 C(T) 1/2T 249.4 (277.8) (212.9) 0
218.9 187.2 1
107.7 93.7 1
(269.3) (212.9) 0
(327.1) (212.9) 0
23 C(T) 1/2T 245.8 (325.0)A (209.9) 0
(328.0)A (209.9) 0
227.0 194.1 1
A
R-curve (no cleavage instability).

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
26
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

X4. CALCULATION OF TOLERANCE BOUNDS11

X4.1 As an example, the 5 % and 95 % bounds on the X4.2.1 An equivalent value of the median toughness for a
Appendix X2 master curve with To = -80.3 °C, are (from Eq 30 data set, K Jceq~ med! , is defined by Eq 32 in 10.10.1 (31):
in 10.9):
1 r
eq
K Jc~ med! 5 (
r i51
30170 exp~ 0.019 ~ T i 2 T o !!

F S
K Jc~ 0.05! 5201 1n
1
120.05 DG 1⁄4

$ 11 1 77 exp @ 0.019~ T180.3! # % Where Ti are the individual test temperatures. When K Jceq~ med!
is equal to or greater than 83 MPa√m, β = 18°C (32). If the
F S
K Jc~ 0.95! 5201 1n
1
120.95 DG 1⁄4

$ 11 1 77 exp @ 0.019~ T180.3! # %

eq
K Jc~ med! is below 83 MPa√m, values of β must be increased
according to the following schedule (from table in 10.10.1):
eq
K Jcs medd
β
These tolerance bounds are illustrated in Fig. X4.1 and are 1T equivalentA (°C)
(MPa√m)
also provided in Table X4.1.
83 to 66 18.8
X4.1.1 The potential error due to finite sample size can be 65 to 58 20.1
considered, in terms of To, by calculating a margin adjustment, A eq
Round off K Jcs medd to nearest whole number.
as described in 10.10. For the dataset in Appendix X2, which was conducted at a
single test temperature of -75 °C, K Jceq~ med! = 107.4 MPa√m which
X4.2 Margin Adjustment—The margin adjustment is an
is greater than 83 MPa√m. Therefore, β = 18°C. and the
upward temperature shift of the tolerance bound curve. Margin standard deviation (σ) on the estimate of To obtained from six
is added to cover the uncertainty in To that is associated with (6) uncensored specimens is
the use of only a few specimens to establish To. Additional
uncertainty can result from other sources including material
inhomogeneity and experimental errors. The standard devia- σ5 Œ β2
r
1σ 2exp 5 Œ 182
6
14 2 5 8.4
tion on the estimate of To, which incorporates sample size and X4.2.2 To estimate the uncertainty in To, a standard two-tail
experimental uncertainties is given by Eq 31 in 10.10: normal deviate, Z, should be taken from statistical handbook
tabulations. The selection of the confidence limit for To
σ5 Œ β2
r
1σ 2exp adjustment is a matter for engineering judgment. The following
example calculation is for 85 % confidence (two-tail) adjust-
where: ment to the dataset in Appendix X2 with the standard deviation
β = sample size uncertainty factor given above.
r = total number of uncensored data used to establish the ∆ T o 5 σ ~ Z 85! 5 8.4 ~ 1.44! 5 12°C
value of To. T o ~ margin! 5 T o 1∆T o 5 280.3°112.0° 5 268.3°C
σexp = contribution of experimental uncertainties. If standard
Then the margin-adjusted 5 % tolerance bound in X4.1 (that
calibration practices are followed, σexp = 4 °C may
is, KJc(0.05)) is:
be used.
K Jc~ 05! F S
5 20 1 1n
1
1 2 0.5 DG 1⁄4

$ 11 1 77 exp @ 0.019~ T 1 68.3! # %

This margin-adjusted 5% tolerance bound is plotted in Fig.
X4.2 as the dashed line (LB) and is also provided in Table
X4.2.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
27
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

FIG. X4.1 Master Curve With Upper and Lower 95 % Tolerance

Bounds

TABLE X4.1 5 % and 95 % Tolerance Bounds

KJc(0.05) KJc(0.95)
T (°C)
(MPa=m) (MPa=m)
-130 39.5 73.9
-125 40.9 77.8
-120 42.5 82.1
-115 44.2 86.9
-110 46.1 92.1
-105 48.1 97.8
-100 50.4 104.1
-95 52.9 111.1
-90 55.7 118.7
-85 58.7 127.1
-80 62.1 136.3
-75 65.8 146.5
-70 69.8 157.6
-65 74.2 169.9
-60 79.1 183.4
-55 84.5 198.3
-50 90.4 214.6
-45 96.9 232.5
-40 104.0 252.3
-35 111.9 274.0
-30 120.5 297.9

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
28
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

FIG. X4.2 Master Curve Showing the Difference Between 5 % Tol-

erance Bound and Lower Bound That Includes 85 % Confidence
Margin on To

TABLE X4.2 Margin-adjusted 5% Tolerance Bound

T (°C) KJc(0.05)
–130 36.6
-125 37.7
-120 38.9
-115 40.3
-110 41.8
-105 43.5
-100 45.3
-95 47.3
-90 49.5
-85 51.9
-80 54.5
-75 57.5
-70 60.7
-65 64.2
-60 68.1
-55 72.4
-50 77.1
-45 82.2
-40 87.9
-35 94.1
-30 101.0

X5. TREATMENT OF POTENTIALLY INHOMOGENEOUS DATA SETS

X5.1 Data sets that fail the screening criterion of 10.6.3
(that is, Toscrn- To > 1.44√β2/r) can be further evaluated to
better characterize the material using the methods in this
appendix. Section X5.2 provides a simplified method (37) for
evaluating material inhomogeneity and determining a gener-
ally conservative estimate of the material’s reference tempera-
ture (ToIN), that shall be used in lieu of To. Section X5.3
provides additional methods for assessing material
brittle transition regime than the methods in X5.2.
X5.2 Simplified Method:
X5.2.1 Small data sets (that is, N ≤ 9):
X5.2.1.1 For every 1T-adjusted KJc value (KJci) for which
δi = 1 (that is, non-censored), calculate the single-data Toi
estimate, using:

F S D G
0.25
N
inhomogeneity, but they are applicable only for data sets with 1
~ K Jci 2 20! ln2 211
N ≥ 20. These methods provide a more accurate determination T oi 5 T i 2
0.019
ln
77
(X5.1)
of the likelihood that the material is inhomogeneous than the
method in 10.6. Further, if the material is likely to be where N is the total number of KJc values in the data set and
inhomogeneous, the methods in X5.3 provide a more accurate Ti is the test temperature corresponding to the KJci data point.
characterization of material performance within the ductile-to- X5.2.1.2 The maximum value of Toi for the data set is Tomax.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
29
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
X5.2.1.3 If: N

T omax 2 T oscrn . 8 °C (X5.2)

lnL 5 ( @ δ ln f
i51
i i 1 ~ 1 2 δ i ! lnS i # (X5.3)

where Toscrn is the screening reference temperature as where:

determined in 10.6.3, then ToIN = Tomax. Otherwise, ToIN = N = number of specimens tested,
Toscrn. δi = 1.0 if the datum is uncensored or zero if the datum is a
X5.2.2 Larger data sets (that is, N ≥ 10): censored value,
fi = the datum failure probability given by Eq X5.4 below,
ToIN = Toscrn, where Toscrn is the screening reference as
and
determined in 10.6.3.
Si = the datum survival probability given by Eq X5.5 below.
X5.2.3 ToIN shall be used as the reference temperature for
the data set instead of To in all subsequent calculations and
reporting. For example, the transition temperature curve in
~ K Jci 2 K min! 3
f i 5 4p A
~ A~ i!
K T 2 K min!
4 exp 2F S
K
K Jci 2 K min 4
A ~ T i ! 2 K min
DG1

10.7 shall be calculated using ToIN instead of To. Tolerance

bounds and margin adjustment for the data set are obtained
4 ~ 1 2 p A!
~ K Jci 2 K min! 3
~ K B ~ T i ! 2 K min! 4 F S
exp 2
K Jci 2 K min 4
K B ~ T i ! 2 K minDG
using 10.9 and 10.10, respectively, by also replacing To with (X5.4)
ToIN.
X5.2.4 An example problem depicting this evaluation is
S i 5 p A exp 2F SK Jci 2 K min 4
K A ~ T i ! 2 K min
1 DG (X5.5)
provided at the end of this appendix.
F S K 2 K min 4
~ 1 2 p A ! exp 2 K ~Jci
B T i ! 2 K min
DG
X5.3 Methods for Evaluating Large Data Sets (N ≥ 20):
where:
X5.3.1 If the data set contains at least 20 KJc values, both
the likelihood that this data set is inhomogeneous and its KJci = either the uncensored or censored (as defined in
reference temperature can be more accurately determined using 8.9.2) KJc datum that has been converted to 1T size
the procedures in this section. X5.3.2 can be used for a data set equivalence as defined in 10.2.2,
that exhibits a bimodal toughness distribution, while X5.3.3 Kmin = 20 MPa√m,
can be used for a data set that exhibits a multimodal distribu- KA(Ti) = Weibull scale parameter for population A at test
tion (38). However, when a data set is found to likely be temperature Ti, given by:
inhomogeneous according to the criteria provided in this
section, it is not possible to analytically determine the nature of 30170 exp@ 0.019 ~ T i 2 T A ! # 2 K min
K A~ T i! 5 1K min
the inhomogeneity (that is, if the data set represents a bimodal ~ ln 2 ! 1⁄4
or multimodal material). Information on the origin of the (X5.6)
specimen data set, fabrication and location of the test
specimens, and processing of the material may suggest which KB(Ti) = Weibull scale parameter for population B at test
inhomogeneity model is most likely to apply. However, if a temperature Ti, given by:
physical basis to determine the nature of the inhomogeneity is
lacking, it is recommended that both the bimodal analysis 30170 exp@ 0.019 ~ T i 2 T B ! # 2 K min
(X5.3.2) and the multimodal analysis (X5.3.3) be performed K B~ T i! 5 1K min
~ ln2! 1⁄4
and the result that leads to the most conservative assessment is (X5.7)
chosen.
X5.3.2 Bimodal Method:
X5.3.2.1 The bimodal toughness distribution applies to data X5.3.2.4 The standard deviations of the obtained parameters
sets that contain two toughness populations; the first population depend on the number of data available and are given by:
is more brittle than the second. Typically, such toughness
22 °C
distributions are encountered in heat-affected zone (HAZ) σ TA 5 (X5.8)
materials where the crack tip can sample the low- or high- =N·p A 2 2
toughness material. The combined bimodal toughness distribu- 16 °C
tion is fully defined by three parameters: the reference tem- σ TB 5 (X5.9)
perature of population A, TA, the reference temperature of
=r 2 N·p A 2 2
population B, TB, and the probability of sampling a specimen 0.35
σ pA 5 (X5.10)
from population A, pA. The probability of sampling a specimen =N·p A 2 2
from population B is equal to 1-pA. Populations A and B are
defined using the convention TB ≤ TA. where r is the number of valid data as defined in 10.3.
X5.3.2.2 A minimum of 20 data points are required within If any of the square root terms in Eq X5.8-X5.10 is less than
the temperature range defined by [To-50 °C, To+50 °C], where or equal to zero, the corresponding standard deviation is not
To has been determined in 10.2, to perform this evaluation. defined and the data set cannot be analyzed using this method.
X5.3.2.3 TA, TB and pA are determined using an appropriate X5.3.2.5 The measure of the likelihood that the data set is
solver algorithm that maximizes the logarithm of the inhomogeneous, MLNH, is assessed using the following equa-
likelihood, given by: tion:

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
30
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

MLNH 5
?T A 2 TB ? (X5.11)
= σ T2 A 1σ T2 B 1σ 2exp
where σexp is the experimental uncertainty of the reference
temperature, which may be taken as 4 °C (see 10.10). The data
set may be inhomogeneous if MLNH is greater than the MLNH
exceedance criterion (MLNHec), defined in Table X5.1 as a
function of N. Linear interpolation shall be used to determine
MLNHec for values of N between the values listed in Table
X5.1.
X5.3.2.6 Monte Carlo based analysis (39) has shown that
the confidence associated with the accuracy of MLNH to
correctly identify a material as inhomogeneous is a function of
N, TA – TB, and pA. The confidence in the accuracy of MLNH
is poor if TA – TB ≤ 30 °C, if pA ≥ 0.8, or if pA ≤ 0.2. Table X5.2
provides the percent confidence in the MLNH evaluation
(MLNHconf). If MLNH exceeds the MLNHec value defined in
Table X5.1, then the data set is likely to be inhomogeneous
with the percent confidence defined in Table X5.2. The values
in Table X5.2 shall be linearly interpolated to determine
MLNHconf for data sets with intermediate values of N, TA – TB,
and pA.
X5.3.2.7 Tolerance Bounds—There is no exact analytical
expression for the tolerance bounds. To obtain the tolerance
bounds for each temperature within the valid temperature range
(X5.3.2.2), find the value of KJc(0.xx) that satisfies the following
equation:
S 5 1 2 0.xx
where xx represents the selected cumulative failure probabil-
KJc(1T) = 100 MPa√m (T0.05), calculated using X5.3.2.7, is
compared in Fig. X5.1 with the tolerance bound calculated
using 10.9 as a function of ∆T = TA – TB and pA.
(3) Option 3 —Accept the results of the bimodal analysis
and utilize TA, as determined in X5.3.2.3, to represent the data
set instead of To in all subsequent calculations and reporting.
For example, the transition temperature curve in 10.7 shall be
calculated using TA instead of To. Tolerance bounds and margin
adjustment for the data set are obtained using 10.9 and 10.10,
respectively, with the substitution of TA for To. This option is
more conservative than Option 2 in X5.3.2.8(2).
X5.3.3 Multimodal Method:
X5.3.3.1 The multimodal toughness distribution applies to
data sets that contain randomly distributed toughness popula-
tions. Typically, such cases are encountered in heterogeneous
ferritic steels, for which the macroscopic heterogeneities are
randomly distributed, or data sets of similar materials that are
combined together. The overall data set is composed of several
populations that individually follow the master curve distribu-
tion. The combined distribution is fully defined by two
parameters: the mean reference temperature of all populations
(Tm), and the standard deviation around the mean (σTm).
X5.3.3.2 To perform this evaluation, a minimum of 20 data
points are required within the temperature range defined as [Tm
- 50 °C, Tm + 50 °C].
X5.3.3.3 Tm and σTm are determined using an appropriate
solver algorithm that maximizes the logarithm of the likelihood
(X5.12) given by the following equation:
N

ity in percent, and where: lnL 5 ( @ δ ln f

i51
i i 1 ~ 1 2 δ i ! lnS i # (X5.14)

FS ~
S 5 p A exp -
K Jc ~ 0.xx! 2 K min 4
K A T i ! 2 K min
1 DG (X5.13)
where:
N = number of specimens tested,
! F S DG
K Jc ~ 0.xx! 2 K min 4 δi = 1.0 if the datum is uncensored or zero if the datum is a
~ 1 2 p A exp - K ~ T ! 2 K
B i min censored value,
X5.3.2.8 If the confidence in the MLNH evaluation is low fi = the datum cumulative failure probability as given below
and testing additional specimens is either not possible or not by Eq X5.15, and
Si = the datum cumulative survival probability as given
likely to increase this confidence based on the results in Table
below by Eq X5.16.
X5.2, the following optional procedures are provided.
(1) Option 1 —Investigate the possibility that the data set `

exhibits multimodal inhomogeneity, using X5.3.3. fi 5 *f

2`
T f τ 0 ~ i ! dτ 0 (X5.15)
(2) Option 2 —Accept the results of the bimodal analysis
`
and use the corresponding tolerance bounds in X5.3.2.7 for the
application of interest. This choice provides a more conserva- Si 5 *f
2`
T S τ 0 ~ i ! dτ 0 (X5.16)
tive result than using the tolerance bounds associated with the
assumption of homogeneity implicit in To as calculated in 10.9. with:
The additional temperature shift in the 5 % tolerance bound at
fT 5
1
σ Tm =2π
exp H 2 ~ τ 0 2 T m! 2
2
2σ Tm J (X5.17)
TABLE X5.1 MLNH Exceedance Criterion (MLNHec) as a Function
of N

FS DG
N MLNHec
~ K Jci 2 K min! 3 K Jci 2K min 4
20 2.4 f τ 0~i ! 5 4 4 exp - (X5.18)
32 3.0 ~ τ0 i
K ~ T ! 2K min! K τ 0 ~ T i ! 2K min
64 3.7

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
31
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X5.2 Confidence that MLNH has Correctly Identified a Bimodal Inhomogeneous Data Set
MLNHconf (%) MLNHconf (%) MLNHconf (%) MLNHconf (%)
pA
TA – TB = 20 °C TA – TB = 30 °C TA – TB = 40 °C TA – TB = 50 °C
0.2 34 50 60 70
0.3 42 65 82 91
N=20 0.4 49 76 93 98
0.6 51 80 97 100
0.8 41 64 78 88
0.2 43 65 79 87
0.3 54 81 94 97
N=32 0.4 57 89 98 99
0.6 64 92 100 100
0.8 51 73 89 94
0.2 69 90 97 99
0.3 79 97 100 100
N=64 0.4 81 99 100 100
0.6 83 99 100 100
0.8 68 92 99 100

FIG. X5.1 Additional Shift in the 5% Tolerance Bound on KJc(1T) due to Bimodal Inhomogeneity as a Function of ∆T = TA – TB and pA

FS
S τ 0 ~ i ! 5 exp -
K Jci 2 K min
K τ 0 ~ T i ! 2 K min DG
4
(X5.19)
Kmin
Kτo(Ti)
= 20 MPa√m, and
= Weibull scale parameter for a population charac-
terized by a reference temperature τ0 at the test
where:
temperature Ti, given by:
τ0 = the temperature region for calculating the cumula-
tive failure and survival densities, 30170 exp @ 0.019 ~ T i 2 τ 0 ! # 2K min
K τ 0~ T i ! 5 1K min (X5.20)
KJci = either the uncensored or censored (as defined in ~ ln 2 ! 1⁄4
8.9.2) KJc datum that has been converted to 1T size
equivalence as defined in 10.2.2,

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
32
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.

Note that when using numerical methods to calculate the
infinite integrals in Eq X5.15 and Eq X5.16, evaluation over
the range of -200 °C < τ0 < 200 °C typically provides sufficient
accuracy.
X5.3.3.4 The measure of the likelihood that the data set is
inhomogeneous, MLNH, is assessed using the following equa-
tion:
σ Tm
MLNH 5
Œ β2
r
1σ 2exp
where β and r are calculated using To in 10.10 and 10.3,
respectively, and σexp is the experimental uncertainty on the
reference temperature which may be taken as 4 °C (10.10). The
MLNH exceedance criterion (MLNHec) is 2 for all values of N.
Therefore, the data set may be considered to be inhomoge-
neous if MLNH > MLNHec.
X5.3.3.5 The confidence associated with the accuracy of
MLNH (Eq X5.21) to correctly identify a material as inhomo-
geneous (that is, MLNH > MLNHec > 2) is a function of N and
σTm. The confidence in the accuracy of MLNH is poor if σTm <
10 °C. Table X5.3 provides the percent confidence in the
MLNH evaluation (MLNHconf). If MLNH > 2, then the data set
is likely to be inhomogeneous with the percent confidence
defined in Table X5.3. The values in Table X5.3 shall be
linearly interpolated to determine MLNHconf for data sets with
intermediate values of N and σTm.
X5.3.3.6 Tolerance Bounds—There is no exact analytical
expression for the tolerance bounds. To obtain the tolerance
bounds, for each temperature within the valid temperature
range (X5.3.3.2), find the value of KJc(0.xx) that satisfies the
following equation:
S 5 1 2 0.xx
where xx represents the selected cumulative probability in
percent, and where:
E1921 − 19
data set instead of To. For example, the transition temperature
curve in 10.7 shall be calculated using Tm instead of To. The
corresponding tolerance bounds in X5.3.3.6 shall be used for
the application of interest instead of the tolerance bounds
calculated in 10.9. Additionally, the margin adjustment for the
data set (10.10) shall be calculated with Tm instead of To. The
use of Tm instead of To and the tolerance bounds of X5.3.3.6
will result in a more conservative assessment than using To and
(X5.21) the associated tolerance bounds as calculated in 10.9. The
additional temperature shift in the 5 % tolerance bounds at
KJc(1T) = 100 MPa√m (T0.05), calculated using X5.3.3.6, is
compared in Fig. X5.2 with the tolerance bounds calculated
using 10.9, as a function of σTm.
X5.4 Report:
X5.4.1 If the data set has been analyzed according to X5.2,
Tomax and ToIN shall be reported.
X5.4.2 If the data set has been analyzed according to
X5.3.2, the following information shall be reported:
X5.4.2.1 MLNH, MLNHec, pa, TA, and TB.
X5.4.2.2 MLNHconf.
X5.4.2.3 KJc(0.05) and KJc(0.95) from -50 °C ≤ T – To ≤ 50 °C
using a maximum temperature increment of 20 °C.
X5.4.3 If the data set has been analyzed according to
X5.3.3, the following information shall be reported:
X5.4.3.1 MLNH, Tm, and σTm.
X5.4.3.2 MLNHconf.
X5.4.3.3 KJc(0.05) and KJc(0.95) from -50 °C ≤ T – Tm ≤ 50 °C
using a maximum temperature increment of 20 °C.
X5.5 Example Problems:
NOTE X5.1—In these example problems, the precision provided for the
(X5.22) results is not supported by the methods in this appendix, which, as
documented in the appendix, have larger uncertainties. However, this
precision is provided so that analysis procedures can be verified. These
results are obtained by rounding values only when the results are reported

J FS DG
and not during intermediate steps in the calculation.
expH
` 4
1 2~τ 2 T m! 2 K Jc ~ 0.xx ! 2 K min
S5 *σ =
2` 2π 2σ
0
2
Tm
exp 2
K τ 0 ~ T ! 2 K min
dτ 0 X5.5.1 Evaluating Material Inhomogeneity Using the Sim-
Tm
plified Method (X5.2)—For this problem, the user’s results
(X5.23)
should either match or, at most, fall within 6 1 significant digit
Note that when using numerical methods to calculate the for the results provided. The material information and data
infinite integral in Eq X5.23, evaluation over the range of –200 associated with the -55 °C test temperature for the example
°C < τo < 200 °C typically provides sufficient accuracy. problem in Appendix X3 are used. The pertinent information
X5.3.3.7 If the confidence in the MLNH evaluation is low from Appendix X3 is summarized as follows.
and testing additional specimens is either not possible or not X5.5.1.1 Material and specimen geometries:
likely to increase this confidence based on the results in Table Material:
X5.3, the following optional procedures are provided. A533 Grade B plate
(1) Option 1—Investigate the possibility that the material Quenched and tempered
demonstrates bimodal inhomogeneity, using X5.3.2. 900 °C WQ (water quench); and 440 °C (5 h) temper
(2) Option 2—Accept the results of the multimodal analy- Specimen Geometry: 0.5T C(T) with ao/W = 0.5
sis and utilize Tm, as determined in X5.3.3.3, to represent the X5.5.1.2 Mechanical properties and censoring limits:

TABLE X5.3 Confidence that MLNH has Correctly Identified a Multimodal Inhomogeneous Data Set
MLNHconf (%) MLNHconf (%) MLNHconf (%) MLNHconf (%)
N
σTm=10 °C σTm=15 °C σTm=20 °C σTm=30 °C
16 26 58 78 96
32 39 81 95 100
64 50 96 100 100

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
33
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19

FIG. X5.2 Additional Shift in the 5% Tolerance Bound on KJc(1T) due to Multimodal Inhomogeneity as a Function of σTm

Yield strength (at -55 °C): 708 MPa X5.5.1.4 Calculate To for data set—While either the multi-
Elastic modulus (at -55 °C): 207 GPa or single-temperature formulation can be used to determine To,
KJclimit (at -55 °C) for 1/2T C(T) specimen: 261 MPa√m the single-temperature equations are used in this example
KJca = 255 MPa√m problem. Therefore, Ko = 100.8 MPa√m from Eq 22, KJcmed =
X5.5.1.3 KJc data—All raw KJc data and the size-adjusted 93.7 MPa√m from Eq 23, and ToQ = To = -50.0 °C from Eq 24.
values are summarized in Table X5.4. Note that because all raw X5.5.1.5 Perform material homogeneity screening evalua-
data is less than KJclimit and KJca, no censoring is required, and
tion (10.6.3):
r = N = 16 for this dataset.
(1) Step 1:
(a) To(step1) = To = -50.0 °C
TABLE X5.4 KJc Data Tabulation
(b) KCENS1 = KJcmed (from Eq 23) = 93.7 MPa√m (can
Test temperature, KJc (MPa=m)
δj also obtained from Eq 25 with To(step1) = To = -50.0 °C)
(°C) Raw data 1T equivalent
(2) Steps 2 through N:
–55 167.7 144.2 1
88.5 77.6 1 (a) Next, censor the KJc data in Table X5.4 using KCENSi
115.2 100.0 1 as the censoring limit. N = 16 but now r = 10. The remaining
81.4 71.6 1
121.9 105.7 1
parameters become Ko = 93.0 MPa√m from Eq 22, KJcmed =
145.0 125.1 1 86.6 MPa√m from Eq 23, and To(step2) = To = -43.8 °C from Eq
104.2 90.8 1 24. Next, KCENS2 = KJcmed = 86.6 MPa√m (from Eq 25) and
64.4 57.3 1
96.8 84.6 1 this process is repeated until To(stepn) – To(stepn-1) < 0.5 °C. For
114.5 99.5 1 the existing problem, the results of each stepwise calculation
107.4 93.5 1
81.0 71.3 1
are included in Table X5.5.
70.0 62.0 1 (b) The iteration stops at step 5 as To(step5) – To(step4) =
131.8 114.0 1 -1.4 °C < 0.5 °C.
69.5 61.6 1
67.5 59.9 1 (3) Evaluate screening criterion:
(a) Toscrn = To(step4) = -41.8 °C

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
34
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X5.5 Results of Screening Iteration Calculations
Step 1 Step 2 Step 3 Step 4 Step 5
N 16 16 16 16 16
r 16 10 8 8 7
Ko (MPa=m) 100.8 93.0 91.6 90.6 92.2
KJcmed (MPa=m) 93.7 86.6 85.4 84.4 85.9
To(stepi) (°C) -50.0 -43.8 -42.6 -41.8 -43.2
KCENSi (MPa=m) 93.7 86.6 85.4 84.4 N/A

(b) From 10.10, K eq

Jcmed = 93.7 MPa√m which is greater (b) Because Tomax is not significantly different than Toscrn,
than 83 MPa√m. Therefore, β = 18 and r = 16. ToIN = Toscrn = -41.8 °C, and ToIN replaces To in the transition
temperature (10.7), tolerance bound (10.9), and margin adjust-
(c) Evaluate T oscrn 2T o ~ step1! # 1.44 Œ β2
r
ment (10.10).
X5.5.2 Evaluating material inhomogeneity for a large data
set (N ≥ 20): The material information and data associated with
T oscrn 2 T o(step1) 5 -41.8 2 -50.0 5 8.2 the -80 °C to 0 °C test temperature range from example
problem Appendix X3 are used. See example problem Appen-
dix X3 for pertinent mechanical properties, censoring limits,
1.44 Œ β2
r
5 1.44 Π182
16
5 6.5
raw data, and the 1T size-adjusted data. The user’s results
should either match or, at most, fall within 6 1 significant digit
for the results provided in this section.
X5.5.2.1 Calculate To for data set—As summarized in

Since 8.2 . 6.5, T oscrn 2 T o(step1) . 1.44 Œ β2

r
example problem Appendix X3, the pertinent results needed to
evaluate material inhomogeneity are as follows: r = 49, N = 53,
and To = -48.1 °C.
and, as per 10.6, this material may be inhomogeneous. X5.5.2.2 Perform material homogeneity screening evalua-
X5.5.1.6 Determine alternative reference temperature—The tion.
user decides to calculate an alternative measure of To to (1) Step 1:
account for possible material inhomogeneity. Because N < 20, (a) To(step1) = To = -48.1 °C
only the simplified method in X5.2 can be used. (b) Calculate KCENS1 from Eq 25 for each individual test
(1) Calculate Toi and determine Tomax : temperature and specimen size with To(step1) = To = -48.1 °C)
(a) Calculate Toi (using Eq X5.1) for each uncensored (2) Steps 2 through N:
data point in Table X5.4. For this data, all points are uncen- (a) Next, censor the individual 1T size-adjusted KJc data
sored and Ti = -55 °C for all data. The results of these in Table X5.4 using KCENS1 as the censoring limit. N = 53, but
calculations are summarized in Table X5.6. now r = 32. For step 2, the censored KJc values and the values
(b) Tomax = max[Toi] = -50.6 °C from Table X5.6. for KCENSi are summarized in Table X5.7.
(2) Calculate ToIN: (b) Next, calculate To(step2) for the data in Table X5.7
(a) Use Eq X5.2 to determine if Tomax is significantly using Eq 20 in 10.2.3: To(step2) = -42.7 °C. Next, using To(step2),
different from Toscrn: the KCENS2 values are determined, the 1T KJci values are
T omax2T oscrn 5250.62-41.8528.8 °C, 8 °C censored using these new KCENS2 values and the process is
repeated until To(stepn) – To(stepn-1) < 0.5 °C. For this example,
four iterations are necessary to satisfy this condition and the
TABLE X5.6 Results of Inhomogeneity Evaluation
To(stepi) values are summarized in Table X5.8.
Test Temperature 1T Equivalent KJc Toi
(°C) (MPa=m) (°C)
(c) The iteration stops at step 4 as To(step4) – To(step3) =
-55 144.2 -119.3
-0.8 °C < 0.5 °C
-55 77.6 -76.2 (3) Evaluate screening criterion:
-55 100.0 -94.9 (a) Toscrn = To(step3) = -40.6 °C
-55 71.6 -69.9
-55 105.7 -98.8
(b) From 10.10, K eqJcmed is calculated for this data using Eq
-55 125.1 -110.1 32, and K eq Jcmed = 128.7 MPa√m, which is greater than 83
-55 90.8 -88.0 MPa√m. Therefore, β = 18 and r = 16.
-55 57.3 -50.6
-55 84.6 -82.8
-55
-55
-55
99.5
93.5
71.3
-94.5
-90.1
-69.5
(c) Evaluate T oscrn 2T o(step1) # 1.44 Œ β2
r
-55 62.0 -57.8
-55 114.0 -103.9
-55 61.6 -57.2 T oscrn 2 T o(step1) 5 2 -40.6 2 -48.1 5 7.5
-55 59.9 -54.7

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
35
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X5.7 Results of Step 2 of the Screening Iteration

Test
Calculation
Censored 1T
Œ
Since 7.5 . 3.7, T oscrn 2 T o(step1) . 1.44
β2
r
KCENS1
Temperature KJci δi
(°C) (MPa=m)
(MPa=m) and, as per 10.6, this material may be inhomogeneous.
-80 46.3 1 68.2 X5.5.2.3 Determine alternative reference temperature—The
-80 68.2 0 68.2
-80 68.2 0 68.2
user decides to perform a more accurate assessment of material
-65 73.9 1 80.8 inhomogeneity and, if necessary, calculate an alternative mea-
-65 80.8 0 80.8 sure of To. Because N ≥ 20, either the bimodal or multimodal
-55 91.4 0 91.4
-55 77.6 1 91.4 evaluation method in X5.3 or the simplified method in X5.2
-55 91.4 0 91.4 can be used. However, for illustrative purposes, both the
-55 71.6 1 91.4 bimodal and multimodal evaluations are performed in this
-55 91.4 0 91.4
-55 91.4 0 91.4 example. Note that due to the iterative nature of the bimodal
-55 90.8 1 91.4 and multimodal evaluation techniques, more uncertainty is
-55 57.3 1 91.4 expected in the reported results. However, results within 6 2
-55 84.6 1 91.4
-55 91.4 0 91.4 significant digits for the results provided in this section are
-55 91.4 0 91.4 achievable.
-55 71.3 1 91.4 (1) Evaluate data set using bimodal method (X5.3.1).
-55 62.0 1 91.4
-55 91.4 0 91.4 (a) Calculate pA, TA, and TB—In the bimodal method, pA,
-55 61.6 1 91.4 TA, and TB in Eq X5.4-X5.7 are iteratively changed until ln(L)
-55 59.9 1 91.4 in Eq X5.3 is maximized. For the data set in this problem, the
-30 89.2 1 128.8
-30 128.8 0 128.8 following final values are obtained after this process is com-
-30 128.8 0 128.8 pleted: pA = 0.56, TA = -33.8 °C, TB = -58.3 °C, and max[ln(L)]
-30 112.1 1 128.8 = -243.5
-30 102.6 1 128.8
-30 127.6 1 128.8 (b) Calculate σpA, σTA, and σTB—Eq X5.8-X5.10 are used
-30 128.8 0 128.8 with N = 53 and r = 49 to obtain the following values: σpA =
-30 83.8 1 128.8 0.066, σTA = 4.17 °C, and σTB = 3.85 °C.
-20 135.1 1 149.5
-20 108.9 1 149.5 (c) Calculate MLNH, MLNHec, and evaluate material
-20 149.5 0 149.5 homogeneity—Eq X5.11 is used to calculate MLNH with the
-20 141.7 1 149.5
values for TA, TB , σTA, and σTB provided previously. The value
-20 149.5 0 149.5
-20 84.8 1 149.5 for σexp is set equal to 4 °C. The result of this calculation is
-20 132.1 1 149.5 MLNH = 3.52.
-10 174.5 0 174.5
-10 154.5 1 174.5
Next, MLNHec is determined from Table X5.1. For this
-10 147.6 1 174.5 example, N = 53 so MLNHec is interpolated between the values
-10 131.8 1 174.5 for N = 32 and N = 64. The result of this calculation is MLNHec
-10 174.5 0 174.5
-10 135.0 1 174.5
= 3.46. Finally, MLNH – MLNHec = 3.52 – 3.46 = 0.06 and
-5 105.4 1 188.9 because MLNH > MLNHec, the material is also determined to
-5 166.5 1 188.9 be inhomogeneous using the bimodal method.
-5 96.0 1 188.9
-5 168.8 1 188.9
The confidence in the accuracy of the MLNH evaluation is
-5 116.5 1 188.9 obtained through linear interpolation of the values in Table
-5 188.9 0 188.9 X5.2. For this example, TA – TB = -33.8 - (-58.3) = 24.5 °C, N
0 204.7 0 204.7
0 187.3 1 204.7
= 53, and pA = 0.56. This will require three linear interpolations
0 93.7 1 204.7 of Table X5.2 to determine the confidence. The portion of
0 204.7 0 204.7 Table X5.2 pertaining to this problem is recreated below along
0 204.7 0 204.7
with the results of the three separate interpolations for TA – TB,
pA and N.
TABLE X5.8 Summary of To(stepi) Results
The information in bold has been linearly interpolated. As
seen in Table X5.9, for the data set in this example problem and
Step 1 Step 2 Step 3 Step 4
TA – TB = 24.5 °C, N = 53, and pA = 0.56, there is an 85%
N 53 53 53 53
r 49 32 29 25 confidence that the material is inhomogeneous, with a bimodal
To(stepi) (°C) -48.1 -42.7 -40.6 -41.4 toughness distribution, as predicted by MLNH.
(d) Determine bimodal tolerance bounds—The 5% and
95% bimodal tolerance bounds are calculated next using Eq
X5.12 and Eq X5.13. From Eq X5.12, 0.xx = 0.05 and S = 0.95
1.44 Œ β2
r 51.44 Π182
49
5 3.7
for calculating the 5% tolerance bound, while 0.xx = 0.95 and
S = 0.05 for calculating the 95% tolerance bound.
In Eq X5.13, the 5% tolerance bound is represented by the
KJc(0.05) values, while the 95% tolerance bound is represented

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
36
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X5.9 % Confidence Values from Table X5.2 and (b) Calculate MLNH and evaluate material
Interpolated Results homogeneity—MLNH is evaluated using Eq X5.21 with, as
% Confidence % Confidence % Confidence stated earlier in the problem, β = 18, r = 49, and σexp = 4 °C.
PA
TA- TB = 20 °C TA- TB = 24.5 °C TA- TB = 30 °C Therefore, with σTm = 13.2 °C, Eq X5.21 becomes:
0.4 57 71 89
N=32 0.56 76 σ Tm 13.2
MLNH 5 5 5 2.77
N=53
0.6
0.56
0.4
64

81
77
85
89
92

99
Œ β
r
2
1σ 2exp Œ 182
49
14 2
N=64 0.56 90
0.6 83 90 99 Therefore, the data set may be considered to be
inhomogeneous, with a multimodal toughness distribution,
since MLNH = 2.77 > 2.
by the KJc(0.95) values. These values are determined iteratively The confidence that MLNH has correctly identified this
at selected test temperatures between -50 °C ≤ Ti – To ≤ 50 °C material as inhomogeneous can be determined using Table
using Eq X5.13. KA(T) and KB(T) in Eq X5.13 are determined X5.3. The values in Table X5.3 should be interpolated for both
from Eq X5.6 and Eq X5.7 at each selected evaluation N and σTm. The portion of Table X5.3 pertaining to this
temperature, T. problem is recreated in Table X5.11 along with the results of
In this example, pA = 0.56 in Eq X5.13 and TA = -33.8 °C interpolations for N and σTm. In this case, two separate
and TB = -58.3 °C are used in Eq X5.6 and Eq X5.7, interpolations are required. The information in bold has been
respectively. The values for KA(T) and KB(T) along the 5% and linearly interpolated.
95% tolerance bounds determined from Eq X5.13 are summa- As seen in Table X5.11, for the data set in this example
rized in Table X5.10 at 5 °C temperature intervals. Since To = problem with N = 53, and σTm. = 13.2 °C, there is a 75%
-48.1 °C for the data set, the bounds are calculated between -98 confidence that the material is inhomogeneous as calculated by
°C and 2 °C. MLNH.
(2) Evaluate data set using multimodal method (X5.3.2). (c) Determine multimodal tolerance bounds—The 5%
(a) Calculate Tm and σTm—The values for Tm and σTm are and 95% multimodal tolerance bounds are calculated next
determined iteratively by maximizing the likelihood function using Eq X5.22 and Eq X5.23. From Eq X5.22, 0.xx = 0.05 and
(that is, max[ln(L)]) of Eq X5.14 with variables as defined in S = 0.95 for calculating the 5% tolerance bound, while 0.xx =
Eq X5.15-X5.20. For the data set in this problem, the following 0.95 and S = 0.05 for calculating the 95% tolerance bound. In
final values are obtained after this process is completed: Tm = Eq X5.23, the 5% tolerance bound is represented by the
-44.1 °C, σTm = 13.2 °C and max[ln(L)] = -243.5. KJc(0.05) values, while the 95% tolerance bound is represented
by the KJc(0.95) values. These values are determined iteratively
TABLE X5.10 Bimodal 5% and 95% Tolerance Bounds at selected test temperatures between -50 °C ≤ Ti – Tm ≤ 50 °C
T i – To Ti KA(T) KB(T) KJc(0.05) KJc(0.95) using Eq X5.23. Kτo(T) in Eq X5.23 is calculated using Eq
(°C) (°C) (MPa=m) (MPa=m) (MPa=m) (MPa=m) X5.20 for each selected evaluation temperature, T. In this
-50 -98 53.6 67.1 37.7 77.1 example, Tm = -44.1 °C and σTm = 13.2 °C are used in Eq
-45 -93 55.9 70.7 38.9 81.5 X5.23.
-40 -88 58.4 74.6 40.2 86.3
-35 -83 61.1 79.0 41.7 91.6 The values for Kτo(T) and the 5% and 95% tolerance bounds
-30 -78 64.1 83.7 43.3 97.4 determined from Eq X5.23 are summarized in Table X5.12 at
-25 -73 67.4 89.0 45.1 103.8 5 °C temperature intervals, starting at T = -90 °C. Since Tm =
-20 -68 71.1 94.8 47.1 110.8
-15 -63 75.1 101.2 49.2 118.5 -44.1 °C for the data set, the bounds are applicable between -94
-10 -58 79.4 108.2 51.6 127.0 °C and 6 °C.
-5 -53 84.3 115.8 54.1 136.3
0 -48 89.6 124.3 57.0 146.6
5 -43 95.4 133.6 60.1 157.9
10 -38 101.9 143.8 63.6 170.3
15 -33 108.9 155.1 67.4 184.0 TABLE X5.11 % Confidence Values from Table X5.3 and
20 -28 116.7 167.5 71.5 199.0 Interpolated Results
25 -23 125.2 181.1 76.1 215.5
30 -18 134.6 196.0 81.1 233.7 % Confidence % Confidence % Confidence
N
35 -13 145.0 212.5 86.7 253.6 σTm=10 °C σTm=13.2 °C σTm=15 °C
40 -8 156.3 230.6 92.8 275.6 32 39 81
45 -3 168.8 250.4 99.5 299.7 53 46 75 91
50 2 182.6 272.3 106.8 326.3 64 50 96

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
37
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
TABLE X5.12 Multimodal 5% and 95% Tolerance Bounds
T i – Tm Ti KJc(0.05) KJc(0.95)
(°C) (°C) (MPa=m) (MPa=m)
-46 -90 39.4 84.1
-41 -85 40.7 89.2
-36 -80 42.2 94.7
-31 -75 43.9 100.8
-26 -70 45.7 107.6
-21 -65 47.7 115.0
-16 -60 49.9 123.1
-11 -55 52.3 132.1
-6 -50 54.9 142.0
-1 -45 57.8 152.8
4 -40 61.0 164.7
9 -35 64.5 177.9
14 -30 68.4 192.3
19 -25 72.6 208.1
24 -20 77.3 225.6
29 -15 82.4 244.8
34 -10 88.1 265.9
39 -5 94.3 289.1
44 0 101.1 314.6

REFERENCES

(1) Gao, X, and Dodds, R. H., “Constraint Effects on the Ductile-to-
Cleavage Transition Temperature of Ferritic Steels: A Weibull Stress
Model,” International Journal of Fracture, 102, 2000, pp. 43-69.
(2) Wallin, K., Planman, T., Valo, M., and Rintamaa, R., “Applicability of
Miniature Size Bend Specimens to Determine the Master Curve
Reference Temperature To,” Engineering Fracture Mechanics, Vol 68,
2001, pp. 1265-1296.
(3) Joyce, J. A., and Tregoning, R. L., “Investigation of Specimen
Geometry Effects and Material Inhomogeneity Effects in A533B
Steel,” ECF 14—Fracture Mechanics Beyond 2000, Proceedings of
the 14th European Conference on Fracture, Cracow, September 2002.
(4) Anderson, T. L., Steinstra, D., and Dodds, R. H., “A Theoretical
Framework for Addressing Fracture in the Ductile-Brittle Transition
Region,” Fracture Mechanics, 24th Volume, ASTM STP 1207, ASTM,
1994, pp. 185-214.
(5) Ruggeri, C., Dodds, R.H., and Wallin, K.,“Constraint Effects on
Reference Temperature, To, for Ferritic Steels in the Transition
Region,” Engineering Farcture Mechanics, 60( 1), 1998, pp. 19-36.
(6) Paris, P. C., “Fracture Mechanics in the Elastic-Plastic Regime,” Flaw
Growth in Fracture, ASTM STP 631, ASTM, August 1976, pp. 3-27.
(7) Turner, C. E., “The Eta Factor,” Post Yield Fracture Mechanics,
Second Ed., Elsevier Applied Science Publishers, London and New
York, 1984, pp. 451-459.
(8) McCabe, D. E., Evaluation of Crack Pop-ins and the Determination
of their Relevance to Design Considerations, NUREG/CR-5952
(ORNL/TM-12247), February 1993.
(9) Wallin, K., “The Scatter in Klc Results,” Engineering Fracture
Mechanics, 19( 6) (1984), pp. 1085-1093.
(10) McCabe, D. E., Zerbst, U., and Heerens, J., “Development of Test
Practice Requirements for a Standard Method of Fracture Toughness
Testing in the Transition Range,” GKSS Report 93/E/81, GKSS
Forschungszentrum, Geesthacht, GmBH, Germany, 1993.
(11) Steinstra, D. I. A., “Stochastic Micromechanical Modeling of Cleav-
age Fracture in the Ductile-Brittle Transition Region,” MM6013-90-
11, Ph.D. Thesis, Texas A & M University, College Station, TX,
August 1990.
(12) Wallin, K., “A Simple Theoretical Charpy V-Klc Correlation for
Irradiation Embrittlement,” ASME Pressure Vessels and Piping
Conference, Innovative Approaches to Irradiation Damage and
Fracture Analysis, PVP-Vol 170 , American Society of Mechanical
Engineers, New York, July 1989.
(13) Heerens, J., Read, D. T., Cornec, A., and Schwalbe, K.-H., “Inter-
pretation of Fracture Toughness in the Ductile-to-Brittle Transition
Region by Fractographical Observations,” Defect Assessment in
Components - Fundamentals and Applications, J. G. Blauel and K.
H. Schwalbe, eds., ESIS/EGF9, Mechanical Engineering
Publications, London, 1991, pp. 659-78.
(14) Landes, J. D., and McCabe, D. E., “Effect of Section Size on
Transition Temperature Behavior of Structural Steels,” Fracture
Mechanics: Fifteenth Symposium, ASTM STP 833, ASTM, 1984, pp.
378-392.
(15) Wallin, K., “Recommendations for the Application of Fracture
Toughness Data for Structural Integrity Assessments,” Proceedings
of the Joint IAEA/CSNI Specialists Meeting on Fracture Mechanics
Verification by Large-Scale Testing, NUREG/CP-0131 (ORNL/TM-
12413), October 1993, pp. 465-494.
(16) Petti, J.P., and Dodds, R.H., Jr., “Constraint Comparisons for
Common Fracture Specimens: C(T)s and SE(B)s,” Engineering
Fracture Mechanics, Vol. 71, 2004, pp. 2677 – 2683.
(17) Joyce, J. A., and Tregoning, R. L., “Determination of Constraint
Limits for Cleavage Initiated Toughness Data,” Engineering Frac-
ture Mechanics, Vol. 72, Issue 10, July 2005, pp. 1559-1579.
(18) Nanstad, R.K., et. al., “IAEA Coordinated Research Project on
Master Curve Approach to Monitor Fracture Toughness of RPV
Steels: Final Results of the Experiment Exercise to Support Con-
straint Effects,” Proceedings of PVP2009, 2009 ASME Pressure
Vessels and Piping Division Conference, July 26-30, 2009, Prague,
Czech Republic.
(19) Zhu, X. K., Leis, B.N., and Joyce, J.A., “Experimental Estimation of
J-R Curves for Load-CMOD Records for SE(B) Specimens” Journal
of ASTM International, Vol 5 No.5 March 2008.
(20) Scibetta, M., “3-D Finite Element Simulation of the PCCv Specimen
Statically Loaded in Three-Point Bending,” report BLG-860, SCK·
CEN Mol, Belgium, March 2000.
(21) Schwalbe, K. H., Hellmann, D., Heerens, J., Knaack, J., and
Mueller-Roos, J., “Measurement of Stable Crack Growth Including
Detection of Initiation of Growth Using Potential Drop and Partial
Unloading Methods,” Elastic-Plastic Fracture Test Methods, Users
Experience, ASTM STP 856, ASTM, 1983, pp. 338-62.
(22) Landes, J. D., “J Calculation from Front Face Displacement Mea-
surements of a Compact Specimen,” International Journal of
Fracture, Vol 16, 1980, pp. R183-86.

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
38
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.
 E1921 − 19
(23) Madison, R.B. and Irwin, G.R., “Dynamic Kc Testing of Structural
Steel,” Journal of the Structural Division, Proceedings of the
American Society of Civil Engineers, Vol. 100, No. ST7, July 1974,
pp. 1331 – 1349.
(24) Sokolov, M. A., and Nanstad, R. K., “Comparison of Irradiation-
Induced Shifts of KJc and Charpy Impact Toughness for Reactor
Pressure Vessel Steels,” in Effects of Radiation on Materials: 18th
International Symposium, ASTM STP 1325, R. K. Nanstad, M. L.
Hamilton, F. A. Garner, and A. S. Kumar, Eds., American Society for
Testing and Materials, West Conshohocken, PA, 1999, pp. 167-190.
(25) Joyce, J. A., Tregoning, R.L., and Roe, C., “On Setting Testing Rate
Limitations for the Master Curve Reference Temperature, To, of
ASTM E1921,” Journal of Testing and Evaluation, Mar 2006, Vol.
34, No. 2.
(26) Pisarki, H.G., “Measurement of Heat Affected Zone Fracture
Toughness,” Paper TS31, Steel in Marine Structures, C. Noordhoek
and J. De Back, eds., Proceedings of 3rd International ECSC
Conference, Delft, June 15-18, 1987, Elsevier Science Publishers
B.V., Amsterdam, pp. 647 - 656.
(27) Wallin, K., “Validity of Small Specimen Fracture Toughness Esti-
mates Neglecting Constraint Corrections,” in Constraint Effects in
Fracture: Theory and Applications, ASTM STP 1244, M. Kirk and A.
Bakker, eds., ASTM, 1994, pp. 519-537.
(28) Merkle, J. G., Wallin, K., and McCabe, D. E., Technical Basis for an
ASTM Standard on Determining the Reference Temperature, To for
Ferritic Steels in the Transition Range, NUREG/CR-5504 (ORNL/
TM-13631) November 1998.
(29) Wallin, K., “Inhomogeneity Screening Criterion for the ASTM
E1921 To Estimate Based on the SINTAP Lower-Tail Methodology,”
Journal of Testing and Evaluation, Vol. 40, No. 6, 2012
(30) ASME Boiler and Pressure Vessel Code. An American National
Standard, Sect. XI, “Rules for Inservice Inspection of Nuclear Power
Plant Components,” Article A-4000, American Society of Mechani-
cal Engineers, New York, 1993.
(31) Lucon, E., Scibetta, M. and van Walle, E., “Assessment of the Master
Curve approach on three reactor pressure vessel steels”, Interna-
tional Journal of Fracture 119: 161-178, 2003.
(32) Wallin, K., “Master Curve Analysis of Ductile to Brittle Transition
Region Fracture Toughness Round Robin Data (The Euro Fracture
Toughness Curve)”, VTT Technical Document 367.58P, Espoo,
Finland, 1998.
(33) IAEA-TECDOC-1631, “Master Curve Approach to Monitor Frac-
ture Toughness of Reactor Pressure Vessels in Nuclear Power
Plants,” International Atomic Energy Agency, Vienna (Austria),
October 2009.
(34) Wallin, K., “Effect of Strain Rate on the Fracture Toughness
Reference Temperature To for Ferric Steels”, in Recent Advances on
Fracture,” R.K. Mahidhara, A.B. Geltmacher and K. Sadananda,
eds., The Mineral, Metals, & Materials Society, 1997
(35) Irwin, G.R., “Crack-Toughness Testing of Strain Rate Sensitive
Materials”, Transitions of the ASME Journal of Engineering for
Power, Oct. 1964, pp. 444–450.
(36) Shoemaker, A.K., “Factors Influencing the Plane-Strain Crack
Toughness Values of a Structural Steel,” Transactions of the ASME,
Journal of Basic Engineering, Sept. 1969, pp. 506–511.
(37) Pisarski H and Wallin K., “The SINTAP fracture toughness estima-
tion procedure,” Engineering Fracture Mechanics, Volume 67, Issue
6, December 2000, pp. 613–624.
(38) Wallin, K., Nevasmaa, P. Laukkanen A., and Planman, T., "Master
Curve Analysis of Inhomogeneous Ferritic Steels," Engineering
Fracture Mechanics, Volume 71, Issues 16-17, November 2004, pp.
2329- 2346.
(39) Joyce, J.A. and Tregoning R, “Evaluation of a Method to Charac-
terize Material Inhomogeneity in Ferritic Steels within the Ductile-
to-Brittle Transition Regime,” Engineering Fracture Mechanics,
.Volume 78, Issue 17, December 2011, pp. 2870 – 2884.

ASTM International takes no position respecting the validity of any patent rights asserted in connection with any item mentioned
in this standard. Users of this standard are expressly advised that determination of the validity of any such patent rights, and the risk
of infringement of such rights, are entirely their own responsibility.

This standard is subject to revision at any time by the responsible technical committee and must be reviewed every five years and
if not revised, either reapproved or withdrawn. Your comments are invited either for revision of this standard or for additional standards
and should be addressed to ASTM International Headquarters. Your comments will receive careful consideration at a meeting of the
responsible technical committee, which you may attend. If you feel that your comments have not received a fair hearing you should
make your views known to the ASTM Committee on Standards, at the address shown below.

This standard is copyrighted by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959,
United States. Individual reprints (single or multiple copies) of this standard may be obtained by contacting ASTM at the above
address or at 610-832-9585 (phone), 610-832-9555 (fax), or service@astm.org (e-mail); or through the ASTM website
(www.astm.org). Permission rights to photocopy the standard may also be secured from the Copyright Clearance Center, 222
Rosewood Drive, Danvers, MA 01923, Tel: (978) 646-2600; http://www.copyright.com/

Copyright by ASTM Int'l (all rights reserved); Thu Mar 21 04:53:44 EDT 2019
39
Downloaded/printed by
Kongju National University (Kongju National University) pursuant to License Agreement. No further reproductions authorized.