STP 1576-2015

ASTM INTERNATIONAL
Selected Technical Papers
Small Specimen
Test Techniques:
6th Volume
STP 1576
Editors
Mikhail A. Sokolov
Enrico Lucon
SELECTED TECHNICAL PAPERS
STP1576
Editors: Mikhail A. Sokolov, Enrico Lucon
Small Specimen Test Techniques:

6th Volume
ASTM Stock #STP1576
ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959.
Printed in the U.S.A.
Library of Congress Cataloging-in-Publication Data
ISBN: 978-0-8031-7597-6
ISSN: 1949-4572
Copyright © 2015 ASTM INTERNATIONAL, West Conshohocken, PA. All rights reserved. This material may
not be reproduced or copied, in whole or in part, in any printed, mechanical, electronic, film, or other
distribution and storage media, without the written consent of the publisher.
Photocopy Rights
Authorization to photocopy items for internal, personal, or educational classroom use, or the internal,
personal, or educational classroom use of specific clients, is granted by ASTM International provided that
the appropriate fee is paid to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923,
Tel: (978) 646-2600; http://www.copyright.com/
The Society is not responsible, as a body, for the statements and opinions expressed in this publication.
ASTM International does not endorse any products represented in this publication.
Peer Review Policy

Each paper published in this volume was evaluated by two peer reviewers and at least one editor. The
authors addressed all of the reviewers’ comments to the satisfaction of both the technical editor(s) and
the ASTM International Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors and the
technical editor(s), but also the work of the peer reviewers. In keeping with long-standing publication
practices, ASTM International maintains the anonymity of the peer reviewers. The ASTM International
Committee on Publications acknowledges with appreciation their dedication and contribution of time
and effort on behalf of ASTM International.
Citation of Papers
When citing papers from this publication, the appropriate citation includes the paper authors, “paper title”,
STP title, STP number, book editor(s), page range, Paper doi, ASTM International, West Conshohocken, PA,
year listed in the footnote of the paper. A citation is provided on page one of each paper.
Printed in Bay Shore, NY

May, 2015
Foreword
This Compilation of Selected Technical Papers, STP1576, Small Specimen Test
Techniques: 6th Volume, contains 15 peer-reviewed papers presented at a sympo-
sium held January 29–31, 2014 in Houston, TX. The symposium was sponsored by
ASTM International Committee E10 on Nuclear Technology and Applications and
Subcommittee E10.02 Behavior and Use of Nuclear Structural Materials.
The Symposium Chairpersons and STP Editors are Mikhail A. Sokolov, Oak
Ridge National Lab, Oak Ridge, TN, USA and Enrico Lucon, National Institute of
Standards and Technology, Boulder, CO, USA.
Contents
Tensile Testing
Influence of Surface Roughness on Tensile Strength of Reduced-Activation

Ferritic/Martensitic Steels Using Small Specimens 3
Shigekazu Suzuki, Shinnosuke Sato, Miho Suzuki, Hiroshi Kinoshita,
Shinji Sato, Shiro Jitsukawa, and Hiroyasu Tanigawa
Micro-Tensile Test Technique Development and Application to Mechanical

Property Determination 12
J. Džugan, R. Procházka, and P. Konopı́k
Role of Scale Factor During Tensile Testing of Small Specimens 31

Maxim N. Gussev, Jeremy T. Busby, Kevin G. Field, Mikhail A. Sokolov,
and Sean E. Gray
Fracture Toughness Testing
International Round Robin Test on Master Curve Reference Temperature

Evaluation Utilizing Miniature C(T) Specimen 53
Masato Yamamoto, Kunio Onizawa, Kentaro Yoshimoto, Takuya Ogawa,
Yasuhiro Mabuchi, Matti Valo, Marlies Lambrecht, Hans-Werner Viehrig,
Naoki Miura, and Naoki Soneda
J-R Curve Determination for Disk-Shaped Compact Specimens Based on the

Normalization Method and the Direct Current Potential Drop Technique 70
Xiang Chen, Randy K. Nanstad, and M. A. Sokolov
Extended Mechanical Testing of RPV Surveillance Materials Using Reconstitution

Technique for Small Sized Specimen to Assist Long Term Operation 88
J. May, J. Rouden, P. Efsing, M. Valo, and H. Hein
Rotation Point and KJC Estimations for Miniature CT-Specimens Based on

Off-Load Line Displacement 110
Matti J. Valo, Tapio K. Planman, and Kim R. Wallin
Small Specimen Reuse Technique to Evaluate Fracture Toughness of High
Dose HT9 Steel 121
T. S. Byun, S. A. Maloy, and J. H. Yoon
Small Punch Testing
Small-Punch Testing for Tensile and Fracture Behavior: Experiences

and Way Forward 145
Karel Matocha
Application of Miniature Small Punch Test Specimen in Determination of

Tensile Properties 160
R. Kopriva, M. Brumovsky, M. Kytka, M. Lasan, J. Siegl, and K. Matocha
Evaluation of Damage in Materials Due to Fatigue Cycling Through Static

and Cyclic Small Punch Testing 168
Raghu V. Prakash and S. Arunkumar
Other Mechanical Tests

Certified KLST Miniaturized Charpy Specimens for the Indirect Verification
of Small-Scale Impact Machines 189
E. Lucon, C. N. McCowan, R. L. Santoyo, and J. D. Splett
Effect of Hydrogen on Crack Growth Behavior in F82H Steel Using

Small-Size Specimen 209
Yuzuru Ito, Masahiro Saito, Katsunori Abe, and Eiichi Wakai
Non-Destructive Techniques
Nonlinear Ultrasonic Characterization of Radiation Damage Using
Charpy Impact Specimen 227
Kathryn H. Matlack, Jin-Yeon Kim, James J. Wall, Jianmin Qu,
and Laurence J. Jacobs
In-Situ Synchrotron X-Ray Study of the Elevated Temperature Deformation

Response of SS 316L Pressurized Creep Tubes 244
Kun Mo, Hsiao-Ming Tung, Xiang Chen, Di Yun, Yinbin Miao,
Weiying Chen, Jonathan Almer, April Novak, and James F. Stubbins
Author Index 257
Subject Index 259

TENSILE TESTING
SMALL SPECIMEN TEST TECHNIQUES: 6TH VOLUME 3
STP 1576, 2015 / available online at www.astm.org / doi: 10.1520/STP157620140019
Shigekazu Suzuki,1 Shinnosuke Sato,1 Miho Suzuki,1

Hiroshi Kinoshita,1 Shinji Sato,1 Shiro Jitsukawa,1
and Hiroyasu Tanigawa2
Influence of Surface Roughness

on Tensile Strength of Reduced-
Activation Ferritic/Martensitic
Steels Using Small Specimens
Reference
Suzuki, Shigekazu, Sato, Shinnosuke, Suzuki, Miho, Kinoshita, Hiroshi, Sato, Shinji, Jitsukawa,
Shiro, and Tanigawa, Hiroyasu, “Influence of Surface Roughness on Tensile Strength of
Reduced-Activation Ferritic/Martensitic Steels Using Small Specimens,” Small Specimen Test
Techniques: 6th Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 3–11,
doi:10.1520/STP157620140019, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
Small specimen test technology (SSTT) is a key technique in evaluating the
irradiation performance of reduced-activation ferritic/martensitic (RAF/M) steels
used in fusion demonstration plants. Because SSTT results are rather sensitive to
surface finishing conditions (mechanical damage, roughness, and dimensional
accuracy), the effect of surface finishing on tensile data was examined. Sheet
tensile specimens were 5 mm long, 1.5 mm wide, and 0.76 mm thick gage
sections (SS-J3 type) prepared from 15-mm-thick plates of RAF/M steel F82H
(F82H-B07 heat). SS-J3 specimens were obtained from the plates by wire
electro-discharge machining (WEDM). The surfaces of the specimens were
finished using several techniques: polishing with abrasive paper, chemical
polishing, and polishing with alumina suspensions. With these techniques, four
types of surface-finished specimens were prepared: (1) WEDM specimens, (2)
specimens finished by abrasive paper polishing, (3) specimens finished by
Manuscript received February 27, 2014; accepted for publication April 8, 2015; published online May 5, 2015.
1
Fukushima National College of Technology, 30 Nagao Kamiarakawa Taira, Iwaki, Fukushima, 970-8034 Japan
(Corresponding author).
2
Japan Atomic Energy Agency, Rokkasho, Aomori, 039-3212 Japan.
3
ASTM Sixth International Symposium on Small Specimen Test Techniques on January 29–31, 2014 in
Houston, TX.
Copyright V
C 2015 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959.
4 STP 1576 On Small Specimen Test Techniques
abrasive paper polishing followed by alumina suspension polishing, and (4)

specimens finished by abrasive paper polishing followed by chemical and
alumina suspension polishing. Tensile tests were conducted at a nominal strain
rate of 3.33 104 /s. The effects of surface finishing were not strong.
Specimens finished by method 4 exhibited the highest yield stress level. Those
finished by method 3 exhibited similar but slightly smaller strength values. The
tensile results were fairly uniform along the thickness of the plates. The ultimate
tensile strength values obtained from the specimens finished using method 3
were 645 MPa (standard deviation of 16 MPa) for the 15-mm-thick plates. The
area reduction of the 15-mm-thick plates was 0.76 in natural strain.
Keywords
sheet tensile specimens, surface roughness, surface finishing
Introduction
Small specimens are used in the development of fusion and nuclear reactor materi-
als; irradiation tests are conducted using an accelerator and test reactor with the
developed material. Small specimens are used to examine the effect of neutrons and
ions on the material. In addition, they are used in the remaining life assessment of
aging power generation equipment. Small specimens have been made from waste
material and pieces extracted from in-service members. Previous studies reported
on the relation between tensile specimens and surface roughness [1–3]. However,
there is no clear set of rules describing the manufacturing conditions of small
specimens. Furthermore, fatigue test specimens are primarily used to elucidate
the relation between the mechanical properties and surface roughness of a small
specimen [4].
To design a demonstration reactor, the fundamental mechanical properties of
F82H steel must be determined. The effects of specimen size on material properties
will also be important to examine. Previous studies have primarily focused on
strength. Thus, in this research, the effects of varying the small specimen prepara-
tion conditions on both the tensile strength and elongation are examined.
Experimental Conditions
F82H STEEL
The reduced-activation ferritic/martensitic (RAF/M) steel used in this study was
F82H steel provided by the Japan Atomic Energy Agency (JAEA). The composition
is Fe, 8 % Cr, 2 % W, 0.2 % V, and 0.04 % Ta (Fe–8Cr–2W–0.2V–0.04Ta). Sheet
tensile specimens were 5 mm long, 1.5 mm wide, and 0.76 mm thick gage sections
(SS-J3 type) prepared from plates of an RAF/M F82H steel (F82H-B07 heat) with a
thickness of 15 mm. The F82H plates were normalized for 60 min at 1323 K and
tempered for 90 min at 1023 K.
SUZUKI ET AL., DOI 10.1520/STP157620140019 5
FIG. 1 Shape and dimensions of small specimen (SS-J3 type).
SMALL SPECIMEN PROCESSING METHOD

A preliminary experiment was conducted to examine the specimen preparation
conditions. We believed that polishing the side surface would be effective in reduc-
ing the dispersion of the results. We examined the optimal roughness.
Test specimens were prepared using the following procedure.
1. The cutting surface for the general-purpose milling machine was approxi-
mately 2 mm because of the oxide film on the surface.
2. The pilot hole for the wire electric discharge machining (WEDM) and the jig
pin were processed in the general-purpose milling machine.
3. The specimen block was made by WEDM.
4. The sides of the specimen block were polished with abrasive paper.
5. The specimen block was sliced to a thickness of 1.15 mm by WEDM. Then the
polishing margin was set at 0.2 mm on each side for abrasive paper polishing.
6. Alumina suspension polishing was conducted on the surface finish for thick-
ness adjustment and removal of the working layer.
We also prepared SS-J3 type specimens (Fig. 1) with different top surface finish
conditions. The methods of obtaining the different finish conditions and their cor-
responding top surface roughness values (Ra) were as follows.
1. WEDM. Ra ¼ 2.46 lm.
2. Chemical polishing using hydrofluoric acid after WEDM. Ra ¼ 0.56 lm.
3. Abrasive paper polishing (240–4000 grit) after WEDM. Ra ¼ 0.08 lm.
4. Chemical polishing after method 3. Ra ¼ 0.15 lm.
TENSILE TEST CONDITIONS

Tensile tests were conducted at ambient temperature using an Instron 33R 4482
tensile testing machine at a nominal strain rate of 3.3 104. The Instron advanced
video extensometer (AVE), a high-performance, non-contacting extensometer, was
used to obtain the strain distribution in the specimen during testing. For this
reason, the gage point was marked on the surface of the specimen.
For evaluation after the test, the fracture area was measured by observing the
fracture surface with a microscope.
FIG. 2 Jig for tensile test: (a) sample mounted on holder, (b) sample on holder fixed in
testing equipment, and (c) test initiation (holder removed).
A jig was prepared (Fig. 2) to allow the specimen to be mounted on the test
machine without being deformed. The jig and holder were made of stainless steel
and brass, respectively. The specimen was set in the jig on the table. The specimen
was attached to the testing machine while fixed in the holder. Finally, the holder
was removed, and the specimen was pinned to the bottom of the testing equipment.
SPECIMEN SAMPLING POSITIONS

We evaluated the dispersion in the mechanical properties of F82H steel. Specimens
were extracted from different parts of the steel sheet and oriented in different
directions (as shown in Fig. 3) because F82H steel is a rolled material. Samples were
distinguished based on whether they were extracted from the sheet’s center or edge.
In addition, specimens were extracted from each location in orientations that were
perpendicular and parallel to the rolling direction, yielding the four specimen sam-
pling positions and orientations shown in Fig. 3. We took 10 specimens along the
thickness direction.
Results
RESULTS OF PRELIMINARY EXPERIMENT
The specimens prepared using different methods were examined to determine the
effects of the preparation conditions on the material properties. Figure 4 presents
the relation between the surface roughness, tensile strength, and uniform elongation
of the specimens. If the surface was rough, the tensile strength and uniform
elongation tended to be small. Surface corrosion by chemical polishing resulted in
FIG. 3 Specimen sampling positions and orientations.
increased roughness. It is believed that the work-hardened layer near the surface
was removed; therefore, the tensile strength and uniform elongation increased.
From the results of the preliminary experiment, it is seen that variation
decreased when the side surface was polished. Therefore, the following steps were
followed to achieve optimum specimen finish conditions.
1. Extract the sample using WEDM.
2. Polish the side surface with abrasive paper.
3. Slice the specimen to a predetermined thickness.
4. Polish the top surface with abrasive paper (240–2400 grit).
5. Polish the top surface with alumina suspension.
FIG. 4 Surface roughness and tensile properties.

FIG. 5 Typical stress–strain curve.
RESULTS OF TENSILE TEST

Figure 5 presents a typical stress–strain curve. Strain in the uniformly deformed
region, fracture strain, and strain in the locally deformed region are defined as eu, ef,
and De, respectively.
Figure 6 presents the dispersion of the strain in a 15-mm-thick F82H plate. eu
was almost constant at approximately 0.05 across the plate. The dispersion of the
strain was very small. In this case, the standard deviation was 0.008. However, the
average De was 0.147, and its standard deviation was 0.016, corresponding to a
dispersion that was approximately twice that of eu.
Figure 7 presents the distribution of the tensile strength of the 15-mm-thick
F82H plate. It was not definitively confirmed that the sampling position and orien-
tation significantly influence the tensile strength. The average tensile strength was
FIG. 6 Dispersion of strain.

FIG. 7 Dispersion of tensile strength.
645 MPa, with a standard deviation of 16 MPa. It is understood that the material
exhibits very small tensile strength dispersion.
To examine the effect of roughness, specimens with different surface finishes
were prepared. The surface roughness values were Ra ¼ 0.055 and 0.032 lm at 800
and 2400 grit, respectively.
No significant effect of surface finishing on the measured tensile properties was
observed. However, microcracks were observed at the surface of the gage part of the
tested and deformed specimens that were finished with 800 grit abrasive paper, as
shown in Fig. 8. The cracks were approximately 20 lm wide with an opening dis-
placement of approximately 10 lm at the surface. These microcracks covered at
least 20 % of the deformed region surfaces. The surfaces of the tested specimens
FIG. 8 SEM observation of the surface of the gage part.

that were finished with 2400 grit abrasive paper were much smoother, as shown in
Fig. 8(b).
In cases where the observed microcracks were supposed to be half-penny
shaped cracks with a 10 lm depth (a), the stress intensity factor in linear elastic
pffiffiffiffiffi
fracture mechanics may be found from 8 MPam1/2 (¼ 1:12r pa) using values that
were obtained when the specimens fractured (interaction between the microcracks
was ignored). The obtained tensile stress was approximately 1300 MPa when frac-
turing occurred. This possibly suggests that such cracks have no significant effect
on the fracture condition of most structural steels because the steels have a higher
fracture toughness even after neutron irradiation and thermal aging. Therefore, sur-
face finishing with 800 grit abrasive paper seems to be effective for structural steel
specimens. However, finer finishing may be required for materials with lower
fracture toughness.
Summary
1. The optimum specimen finish conditions are as follows:
(a) Extract the specimen using WEDM
(b) Polish the side surface with abrasive paper
(c) Slice the sample to a predetermined thickness
(d) Polish the top surface with alumina suspension
2. Elongation and tensile strength are not affected by the specimen sampling
position.
3. The strain in regions experiencing uniform deformation showed a constant
value of approximately 0.05. The standard deviation was 0.008.
4. The tensile strength of the 15-mm-thick F82H plate was approximately
645 MPa. The standard deviation was 16 MPa.
5. The effects of surface finishing (Ra 0.1) were not strong.
ACKNOWLEDGMENTS
This research was supported by the framework of the Agreement between The
Government of Japan and the European Atomic Energy Community for The Joint
Implementation of The Broader Approach Activities in the Field of Fusion Energy
Research.
References
[1] Schmieder, A. K., “Effect of Specimen Taper and Machining Processes on Tension Test
Results,” Reproducibility and Accuracy of Mechanical Tests, ASTM STP 626, J. M. Holt,
Ed., ASTM International, West Conshohocken, PA, 1977, pp. 3–20.
[2] Sharpe, W. N., Danley, D., and LaVan, D. A., “Microspecimen Tensile Tests of A533-B
Steel,” Small Specimen Test Techniques, ASTM STP 1329, W. R. Corwin, S. T. Rosinski and
E. van Walle, Eds., ASTM International, West Conshohocken, PA, 1998, pp. 497–512.
[3] Byun, T. S., Kim, J. H., Chi, S. H., and Hong, J. H., “Effect of Specimen Thickness on the
Tensile Deformation Properties of SA508 Cl.3 Reactor Pressure Vessel Steel,” Small
Specimen Test Techniques, ASTM STP 1329, W. R. Corwin, S. T. Rosinski, and E. van Walle,
Eds., ASTM International, West Conshohocken, PA, 1998, pp. 575–585.
[4] Kim, S. W., Tanigawa, H., Hirose, T., and Kohyama, A., “Effects of Surface Morphology
and Distributed Inclusions on the Low Cycle Fatigue Behavior of Miniaturized Specimens
of F82H Steel,” J. ASTM Int., Vol. 5, No. 8, 2008, JAI101104.
J. Dzugan,1 R. Procházka,2 and P. Konopı́k2
Micro-Tensile Test Technique

Development and Application
to Mechanical Property
Determination
Reference
Dzugan, J., Procházka, R., and Konopı́k, P., “Micro-Tensile Test Technique Development and
Application to Mechanical Property Determination,” Small Specimen Test Techniques: 6th
Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 12–30, doi:10.1520/
STP157620140022, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
The evaluation of the actual mechanical properties of in-service structures after a
period of operation or the determination of local properties for detailed finite
element method (FEM) simulation requires relevant material data obtained with high
accuracy from a small volume of the experimental material. Therefore, non-
destructive or semi-destructive techniques using small samples are being developed.
The use of small-scale samples also enables the evaluation of material properties in
various locations of tested components—for example, the mechanical properties of
isolated regions of welds. One of the widely used methods in miniature specimen
testing is the small punch test (SPT). The SPT is usually based on conversion of the
obtained results into conventional parameters such as tensile properties, creep,
fatigue, notch toughness, transition temperature, or fracture toughness parameters,
but it requires known correlation parameters determined for the specific material. It
was considered whether there is really a need to convert values from an SPT if there
is not a chance to perform, for example, tensile tests directly on the same volume of
experimental material as used in the SPT. Preliminary FEM analysis was performed
with subsequent experimental verification of the possibility of testing small tensile
Manuscript received February 28, 2014; accepted for publication July 3, 2014; published online September
26, 2014.
1
COMTES FHT a.s., Průmyslová 995, 334 41 Dobřany, Czech Republic, e-mail: jan.dzugan@comtesfht.cz
2
COMTES FHT a.s., Průmyslová 995, 334 41 Dobřany, Czech Republic.
3
Houston, TX.
Copyright V

DZUGAN ET AL., DOI 10.1520/STP157620140022 13
samples with the following dimensions: active part length, 3 mm; width, 1.5 mm; and
thickness, 0.5 mm. On the basis of promising preliminary results, testing fixtures for
micro-tensile samples were developed together with a testing procedure. The
performance of the micro-tensile test method relative to standard tensile tests is
shown here together with its application to weld characterization.
Keywords
micro tensile tests, small sample techniques, dynamic testing
Introduction
The residual lifetime assessment and potential for possible failure of service components
is a critical issue in safety and reliability analyses of industrial plants. A component’s re-
sidual life can be evaluated with standard mechanical test techniques, such as the tensile
test, uniaxial creep test, or Charpy or fracture toughness test. However, for these tests
there is usually insufficient material to sample the component non-invasively.
Therefore, non-destructive techniques are being developed, as well as testing
methods using mini-samples. One of the methods of greatest interest is the small
punch test (SPT). The SPT method uses small disc-shaped samples, usually 8 mm in
diameter and 0.5 mm thick. The testing itself involves the penetration of a hard ball
through the tested sample while the force and ball displacement are measured at
either room temperature or non-ambient temperature. The results of the SPT are of-
ten converted into conventional parameters such as tensile properties, fracture tough-
ness values, or transition temperature determined on the basis of Charpy impact
tests. A broad range of correlations of the SPT with the mentioned parameters has
been proposed on the basis of empirical correlations [1–5]. However, these correla-
tions do not have general validity and must be verified for each new material.
Among the most converted values from SPT results are tensile properties. In
this paper, the authors suggest micro-tensile specimens on the basis of the SPT
specimen size that can be used for real tensile tests. This kind of test would have
minimal material requirements and the same loading mode for samples as in stand-
ard tensile tests, and it enables direct result determination in standard terms with-
out the necessity of any kind of correlation determination. The performance of the
proposed testing procedure is presented in comparison with that of standard-size
specimens for several materials. Additionally, application of the method to the local
property determination of weld zones is shown here. Application of the developed
procedure is also demonstrated for dynamic property measurement. The newly
proposed specimen size is very small relative to the standard specimen size, and
thus the new procedure is named the micro-tensile test (M-TT).
Development of M-TT
A lot of attention has been paid to the development of the SPT within the past two dec-
ades. Sampling devices were developed, as well as testing procedures. Recently, thanks
to the proposed code of practice, the sample size was also standardized, and the
recommended disc geometry is 8 mm in diameter and 0.5 mm thick. There are already
many applications using this sample size (for example, surveillance programs), and
thus this geometry was used as a base for the newly developed testing procedure. The
target was to run a real tensile test, free of any necessary correlations, on specimens
using the same material volume as in the SPT. Some finite element method (FEM) cal-
culations were performed, and the sample geometry shown in Fig. 1 was proposed.
Higher stress concentrations were seen in the transition from gauge section to specimen
shoulders; however, later experiments did not show any problems with expected crack
initiation at these points. This is probably due to the small sample thickness resulting
in prevailing plane stress, allowing stress redistribution over a bigger area.
The specimen did not fit into any kind of testing fixture, and thus the next step was
testing fixture development. In the beginning, the very simple testing setup shown in Fig.
2 was used for the first M-TT trials. The fixture was a simple clamp without any kind of
alignment. During these first attempts, an actuator linear variable differential transformer
(LVDT) was used for strain measurement. The first results were very promising: the
stress levels obtained in M-TTs were very close to those obtained for standard-size speci-
mens. A further step was then improvement of the strain measurement. Because of the
minute specimen size, it was not possible to attach a standard extensometer directly on
the test piece. The extensometer was attached to grips, as can be seen in Fig. 3.
At this stage, it was just a slightly improved crosshead measurement. The
resulting curves of these measurements for different heat-treated states of
34CrNiMo6 steel are displayed in Fig. 4. The obtained curves are very similar to the
curves obtained for standard-size samples, except the yielding part of the curve.
The sensitivity of the strain measurement with the extensometer attached to the
grips was significantly lower than in the case of direct strain measurement on the
sample, which is probably the reason for the different shapes of these curve parts
for M-TTs. The shape of the obtained M-TT curves clearly shows that a different
indexing offset has to be used for comparison with standard yield stress values. On
FIG. 1 M-TT specimen.

FIG. 2 Model of testing grips for M-TT.
FIG. 3 Testing setup with mechanical extensometer.

FIG. 4 Comparison of tensile curves obtained for standard-size specimens and M-TT
specimens with a mechanical extensometer for strain measurement.
the basis of comparison of several offset values, an offset of 0.7 % was used for the
yield stress determination. Very good agreement between M-TT results and results
obtained on standard specimens was found. Already at this stage, use of the pro-
posed sample geometry with widely available testing equipment was able to provide
very nice repeatable results for tensile property determination.
As the target was to obtain, if possible, curves and results very similar to those
for standard-size specimens, further effort was applied to improve the strain measure-
ment. Because of the M-TT specimen size, it was clear that a mechanical extensome-
ter could hardly be used, and therefore optical systems were going to be employed.
As a first candidate, a videoextensometer was used. The testing setup is shown in
Fig. 5. The applied extensometer allowed simultaneous measurement of longitudinal
and transversal strain that could be used, for example, for Poisson ratio calculation,
even with this kind of specimen. The results obtained for a low-carbon steel (LCS)
with this kind of testing setup are depicted in Fig. 6. They are displayed there together
with curves for standard-size tensile specimens with the strain measured by a me-
chanical extensometer. Almost identical curves can be observed for M-TT specimens
and standard specimens up to the ultimate tensile strength (UTS). After the UTS, the
curves’ deviation from each other can be observed; this is due to the various positions
of the extensometer with regard to the necking area. However, the initial gauge length
marking is rather inconvenient because of the small specimen size, and a more so-
phisticated method for the strain measurement is going to be applied.
In the last stage of development of the testing procedure for strain measure-
ment for M-TTs, the digital image correlation (DIC) system ARAMIS was used. In
order for strain measurement with DIC to be performed, a random speckle pattern

FIG. 5 Testing setup with print screen of videoextensometer measurement.
has to be applied on the specimen surface. An example of the DIC strain measure-
ment during the tensile test can be found in Fig. 7.
We carried out tests on several materials with a wide range of tensile properties.
Namely, Al-alloy, Ni-alloy, Titanium Gr. 5, and several steels were compared in
tests on standard-size specimens and with the use of the M-TT. Standard-size sam-
ples were round with diameters ranging from 5 to 10 mm, and in one case (steel) a
segment of pipe was tested. The resulting curves from these tests are graphically
FIG. 6 Comparison of tensile curves obtained on standard-size specimens and M-TT

specimens with videoextensometer.
FIG. 7 DIC strain measurement during different stages of tensile tests for M-TT
specimens.
summarized in Fig. 8. There was excellent agreement for all materials investigated
between standard-size specimens and M-TT specimens for the whole range of
strength levels from about 250 MPa up to 1250 MPa.
On the basis of the positive results attained with the use of DIC for the strain mea-
surement, this system was used for further strain measurement in the course of M-TTs.
Determination of Cross-section Reduction

and Elongation
In addition to the strength properties, properties describing material plasticity such
as cross-section reduction (CSR) and elongation (A) were evaluated. Because of the
FIG. 8 Comparison of records obtained with DIC for M-TT measurement of various
metallic materials.

small size of M-TT specimens, the standard kind of measurement is rather compli-
cated; therefore, alternative methods were tried for the CSR determination. Three
methods were investigated: minimal neck dimensions (CSRM), direct measurement
of fracture area via optical microscope (CSRD), and measurement with circle fitting
(CSRSM).
In the case of CSRM, the minimal specimen dimensions according to Figs. 9(a)
and 9(b) were determined. The area in this case was calculated via the simple multi-
plication of minimal measured dimensions. In the second case, CSRD is the cross-
section area measured with the use of an optical microscope by means of digital
image processing of the fracture area, as depicted in Fig. 9(c). The last applied
method was CSRSM in which a circle is fitted on the basis of minimal and maximal
neck thickness and minimal neck width [Fig. 9(d)]. On the basis of these parameters,
the necking area was subsequently calculated.
These three approaches were applied to a set of specimens, and the results were
compared with the values determined for the CSR of standard-size specimens. A
summarization is shown in Table 1. The best agreement was found for the simplified
measurement CSRM when minimal neck dimensions were simply multiplied.
FIG. 9 Cross-section reduction determination for M-TT specimens.

TABLE 1 Comparison of cross-section reduction measurement with M-TT and standard-size

specimens.
CSRS, % CSRM, % CSRD, % CSRSM, % (CSRM CSRS)2 (CSRD CSRS)2 (CSRSM CSRS)2
53.4 53.2 49.9 46.2 0.04 12.25 51.84

57.1 59.3 54.2 51.7 4.84 8.41 29.16
56.8 58 51.7 49.6 1.45 26.1 51.84
62.6 60.8 54.1 53.6 3.25 72.25 81.00
63.5 66.5 60.9 56.6 9.00 6.77 47.61
59.1 63.2 62.4 56.4 16.81 10.89 7.30
62.1 61.3 61.1 57.8 0.65 1.00 18.49
64.1 64.5 64.3 60.4 0.17 0.05 13.69
63.2 66.9 66.1 60.3 13.69 8.41 8.42
45.7 45.1 49.4 46.4 0.37 13.69 0.49
44.7 45.1 40.8 40.3 0.16 15.21 19.37
46.5 46.5 43.6 41.1 0 8.41 29.16
Sum 50.38 183.33 358.34
Elongation is the last typical value evaluated from tensile tests. Because of the
small specimen dimensions, it is very sensitive topic with M-TT. Considering the
uniform part of a gauge length 2.6 mm the obtained elongation would not be the
standard A5, but A2.65. This would give a hint that elongations measured via M-TTs
should be significantly greater. This kind of measurement requires marking of the
gauge length on the sample. Because of the small specimen dimensions, this is
rather difficult, as the marking accuracy and marking linewidth in relation to the
gauge length lead to sizeable uncertainty. Instead of this, we decided to consider as
a gauge length for the elongation measurement the whole narrow part of the speci-
men including the radius between the gauge section and specimen shoulders
(Fig. 10). The initial gauge length was uniform, thanks to the milling cutter used for
specimen machining in a single pass. This dimension is also rather easy to measure
prior to and after the test. This approach was applied to a set of specimens. The
results are summarized in Table 2. Good agreement across the measured data popu-
lation can be observed. The differences in elongation values between specimen
dimensions are within the standard data scatter attained for this value, and thus the
proposed simple method of elongation calculation from the distance between
shoulder edges seems to be applicable.
Determination of the uniform elongation Ag with the use of strain measure-
ment by a standard extensometer and DIC for M-TT specimens exhibited some dif-
ferences. However, the differences seemed to be on the edge of the measurement
accuracy. The only significant difference was attained for copper, for which very flat
curves were obtained and thus even little differences in peak stress determination
could lead to large differences in the evaluated Ag. In the tensile test records in
Fig. 8, almost identical curves are found for copper.

FIG. 10 Elongation determination for M-TT.
Weld Investigation
The investigated material was an experimental dissimilar weld consisting of two
chromium–molybdenum heat-resistant steels and weld metal. This type of weld is
widely used in the oil and gas industries and in fossil fuel and nuclear power plants.
A weld macro is depicted in Fig. 11.
Sample extraction out of the weld was performed in the direction transverse to
the weld axis. A segment was extracted and grooved with a milling cutter so that
the specimen gauge section was created in a single pass. This semi-product was sub-
sequently sliced on an accurate metallographic saw with a step of 1 mm. Samples
were ground to their final thickness with a grinding machine. The specimen orien-
tation was in the thickness direction. One specimen per location and loading rate
was tested.
The investigation was performed across all weld zones: basic material 1 (BM-
1), heat-affected zone 1 (HAZ 1), weld metal (WM), heat-affected zone 2 (HAZ 2),
and basic material 2 (BM-2). Prior to tests, gauge length dimensions were measured
and recorded. All tests were performed under quasi-static loading conditions in
agreement with standard requirements. The results are summarized in Fig. 12. Hard-
ness values also are shown for comparison, as this is a traditional measure used for
property characterization across a weld. There is good agreement in trends among
all evaluated quantities. The M-TT proved sensitive to local property changes and
provided significantly more information about material behavior such as elonga-
tion, CSR, and strain hardening than hardness tests or SPTs.
22
STP 1576 On Small Specimen Test Techniques
TABLE 2 Summarization of tensile test results for standard and M-TT specimens.
Tensile Tests results
Material Specimen D0, mm a0, mm b0, mm E, GPa Rp0,2, MPa Rm, MPa Ag, % A, % Z, %
Experimental LCS_Standard 4.84 209.4 1143.4 1258.0 3.6 15.6 71.1

low-carbon LCS_M-TT 0.50 1.45 203.8 1133.0 1255.3 2.4 13.9 74.4
steel (LCS)
LCS Steel_Standard 30.00 2.00 12.51 205.1 964.1 1051.6 4.0 15.9 62.7
(segments) Steel_M-TT 0.67 1.57 213.4 990.1 1061.9 2.7 20.3 64.2
Titanium Gr.5 Ti_Standard 7.99 119.6 913.5 965.1 5.2 16.3 42.0
Ti_M-TT 0.55 1.43 114.2 892.3 981.2 5.9 18.6 35.8
X14CrMoVNb- COST F 9.88 218.3 622.1 762.3 8.0 20.7 63.9
N10-1 _Standard
COST F _M-TT 0.58 1.55 228.7 592.3 769.2 7.0 21.7 62.8
P91 P91_Standard 5.01 209.6 540.7 696.2 9.3 20.0 74.3
P91_M-TT 0.55 1.49 146.7 520.9 710.4 9.5 22.8 74.4
Aluminum Al_Standard 2.45 4.96 68.1 264.4 309.3 7.6 11.3 13.5
alloy EN AW Al_M-TT 0.47 1.57 60.7 264.1 311.0 5.6 8.2 11.4
6005 T6
Copper Cu_standard 4.02 110.7 234.6 260.0 16.3 34.2 73.7
99.99 % Cu_M-TT 0.56 1.58 97.4 235.7 254.5 4.4 35.4 70.5

FIG. 11 Macro of the investigated heterogeneous weld.
Strain Rate Sensitivity Measurement

Another application that is going to demonstrate the capabilities of the developed
testing procedure is testing to assess the strain rate sensitivity of materials. As an
experimental material, the same kind of weld was used with similar sampling and
sample preparation procedures. Specimen sampling is shown in Fig. 13. We
extracted four blocks in order to provide samples for different strain rates from the
same locations with regard to the weld. In this case nine different locations across
the weld were investigated. One specimen per location and loading rate was tested.
The sampling orientation was in the plate length direction in this case.
Tests were carried out at room temperature at four different loading velocities,
resulting in the following initial strain rates: 0.003, 0.3, 3, and 33 s 1. Testing strain
rates are designated in graphs as 1 to 4, where 1 means the slowest and 4 the fastest.
The testing setup can be seen in Fig. 14. Examples of the resulting curves are depicted in
Fig. 15. As could be expected, we found an increase in strength properties from the base
metal toward the weld. Because of the high gradient of the property changes in the
transition from the base metal across the heat-affected zone to the weld and multi-pass
welding technology, some property scatter can be expected. Because of these facts, sam-
pling is crucial, as a small deviation in specimen positioning can reveal a different
microstructure. Nevertheless, the obtained results yielded reasonable trends.
Strain rate sensitivity evaluation is shown in Fig. 16. Parallel increasing trends were
found for both yield stress and tensile strength. For both sampling locations, there was
slight deviation from common trends in the values for a strain rate of 3 s 1. This might
have been due to small deviation from the expected sampling location. Nevertheless,
very good results were attained in this case with the use of M-TTs. Considering the fact
that there was just one sample per location and condition, it is clear that the developed
testing procedure provides reliable results.
24
FIG. 12 Results of M-TT and hardness measurement across weld.

FIG. 13 Sampling scheme for strain rate sensitivity tests. Gray lines depict the weld, and
black lines represent sampling locations.
Discussion
Complete development of the M-TT technique is summarized in this paper, from
the specimen geometry proposal to the testing fixture design and strain measure-
ment possibilities through the CSR and elongation determination. The proposed
procedure allows determination of all standard parameters on small-size samples.
FIG. 14 Testing setup for dynamic M-TT with DIC strain measurement.
26
FIG. 15 Comparison of tensile curves at different strain rates for different locations across the weld.
FIG. 16 Strain rate sensitivity evaluation for different weld locations. YS, yield strength; UTS, ultimate tensile strength.
DZUGAN

27 ET AL., DOI 10.1520/STP157620140022
The first issue concerning testing was the strain measurement. At the first stage,
a mechanical extensometer for more accurate crosshead displacement measurement
was employed. The results confirmed the applicability of the proposed measurement
technique with the use of the M-TT. In the second step, a videoextensometer was
used. The results were already providing very good agreement with standard tensile
curves. However, because of the small size of the sample, markings of the gauge
length were rather inconveniently big relative to the specimen size. In the end,
shoulder edges were used as a gauge length mark. When this kind of measurement
was used, deformation of the transition region between the gauge length and should-
ers was included in the strain measurement, which can result in some inaccurate
results. Therefore, a DIC system was used in the final stage of the strain measurement
procedure search. In the case of DIC, local strains can be evaluated including true
stress–true strain measurements. DIC systems allow the user to define the gauge
length; in our case the selected value was within the specimen gauge length. M-TTs
with the use of DIC for the strain measurement were executed for a wide range of
materials: copper alloy, aluminum alloy, titanium Gr. 5, and several different steels.
The results obtained for standard-size samples and M-TT samples exhibited excellent
agreement for both considered strength parameters, yield stress and tensile strength.
The next stage for the M-TT development was procedures for measurement for
elongation and CR determination. Determination of these values was not possible
with a standard micrometer or warier caliper; instead optical measurement with a
microscope had to be carried out. In the case of the CSR, three methods of area
evaluation were investigated. The first one was simple measurement of the neck
minimum dimensions. The second one was measurement of the real cross-section
area via digital image processing. The last one was fitting a circle into the neck sil-
houette on the basis of the minimal width measurement and minimal and maximal
thickness. These three values of the CSR were subsequently compared with values
obtained for standard round specimens. Based on this comparison, the best agree-
ment was obtained with the simple method using just minimal neck dimensions.
Concerning the elongation, the main problem is the accuracy of any kind of
gauge length marking taking into account the 2.6-mm-long gauge length. Mechani-
cal marking or paint marking was disregarded because of problems with application
and subsequent gauge length measurement. The parallel distance between shoulder
edges was used for the elongation evaluation. Elongation values determined via this
kind of measurement exhibited fair agreement with standard sample results.
The developed testing procedure was subsequently applied to the evaluation of
local properties across the weld and the determination of dynamic tensile behavior.
Both of these examples show the potential of the developed method. The method
developed here allows reliable tensile property determination with the use of sam-
ples using the same material volume as used for SPT samples. However, the
reported values obtained in M-TTs are significantly higher, as no correlations are
applied and complete stress-strain curves are available from these measurements as
in the case of standard tests. The M-TT method allows the determination of local

properties and property evolution over some component region and a determina-
tion of anisotropy, and the results shown here also point out the applicability of the
method for dynamic material property determination.
However, when the method proposed here is to be applied, attention should be
paid to the size effect when testing coarse-grained materials. In order for realistic
results to be obtained, the specimen must contain several hundreds of grains over
its cross-section, as was found, for example, in Refs 6 and 7. Significantly smaller
specimens than discussed in this work also can be used for tensile testing in appro-
priate applications, as mentioned in, for example, Refs 8–10.
Conclusions
The development of a micro-tensile test (M-TT) is shown here. A small tensile
specimen size is proposed here on the basis of the small punch test (SPT) specimen
dimensions. The testing procedure for M-TTs is demonstrated here. The strain
measurement is recommended by a digital image correlation (DIC) system, and an
approach for the elongation and the cross-section reduction (CSR) is also proposed.
The proposed procedure allows the determination of all typical tensile test measures
for standard-size samples with the use of micro-samples. Tests were carried out on
several materials with a wide range of tensile properties, namely, Al-alloy, Ni-alloy,
Titanium Gr. 5, and several steels. We found excellent agreement for all materials
investigated between standard-size specimens and M-TT specimens for the whole
range of strength levels from about 250 MPa up to 1250 MPa.
The M-TT performance is demonstrated on the evaluation of local properties
across dissimilar welds. Trends of all evaluated parameters followed trends of hard-
ness measurement but provided much more complex information about material
behavior than hardness tests. The M-TT was also utilized for material property
evaluation under dynamic loading conditions. Realistic strain rate hardening curves
were attained. This creates the potential for dynamic property measurements with
the use of “slower testing systems,” as a strain rate of about 100 s 1 can be reached
with M-TT samples on a testing system with a velocity of about 300 mm/s. This is
significantly different from standard-size samples, for which a loading velocity of
about 5 m/s would be required. Next to the use of less expensive testing systems,
slower loading velocities also diminish the problem of oscillations in the course of
the dynamic tests.
A new testing procedure for small-size tensile samples was proposed. The
method allows direct determination of standard tensile test values without any kind
of correlation. It allows the determination of local properties such as anisotropy
and strain rate sensitivity. There is a wide range of potential applications for the M-
TT method, from evaluation of residual service life to local property determination
as input data for FEM simulations. Verification of the influence of grain size on the
results has to be performed, and validity limits have to be established. Thus further
work on this method is expected in the near future.
ACKNOWLEDGMENTS
This work was done within the work on the project “Creation of an International
Team of Scientists and Participation in Scientific Networks in the Sphere of Nanotech-
nology and Unconventional Forming Material” (CZ.1.07/2.3.00/20.0038), sponsored
by the European Social Fund and the national budget of the Czech Republic.
References
[1] Konopı́k, P. and Dzugan, J., “Determination of Tensile Properties of Low Carbon Steel
and Alloyed Steel 34CrNiMo6 by Small Punch Test and Micro-Tensile Test,” 2nd Interna-
tional Conference SSTT, Ostrava, Czech Republic, Oct 2–4, Material and Metallurgical
Reseach, Ltd., Ostrava, Czech Republic, 2012, pp. 319–328.
[2] Konopı́k, P. and Dzugan, J., “Determination of Fracture Toughness and Tensile Properties
of Structural Steels by Small Punch Test and Micro-Tensile Test,” Proceedings of the
22nd International Conference on Metallurgy and Materials, Brno, Czech Republic, May
15–17, Tanger Ltd., Slezsk
a Ostrava, Czech Republic, 2013.
[3] Konopı́k, P., Dzugan, J., and Prochazka, R., “Evaluation of Local Mechanical Properties of
Steel Weld by Miniature Testing Technique,” MS&T ’13: Materials Science & Technology
2013 Conference & Exhibition, Montreal, QC, Canada, Oct. 27–31, The Minerals, Metals &
Materials Society (TMS), Warrendale, PA, 2013.
[4] Procházka, R., 2013, “Determination of Mechanical Properties of Structural Materials by

Using Miniature Specimen Test Techniques,” M.Sc. thesis, West Bohemian University, Pil-
sen, Czech Republic.
[5] Konopı́k, P., 2013, “Mechanical Properties Results Compatibility for Structural Steels,”
Ph.D. thesis, West Bohemian University, Pilsen, Czech Republic.
[6] Mark Henning, M. and Vehoff, H., “Statistical Size Effects Based on Grain Size and Tex-
ture in Thin Sheets,” Mater. Sci. Eng. A, Vol. 452–453, 2007, pp. 602–613.
[7] Poling, W. A., 2012, “Grain Size Effects in Micro-Tensile Testing of Austenitic Stainless
Steel,” M.Sc. thesis, Colorado School of Mines, Golden, CO.
[8] Zhang, D. and Chu, J., “Mechanical Characterization of Post-buckled Micro-bridge

Beams by Micro-tensile testing,” Microsyst. Technol., Vol. 16, No. 3, 2010, pp. 375–380.
[9] Auhorn, M., Kasanická, B., Beck, T., Schulze, V., and Löhe, D., “Mechanical Strength and
Microstructure of Stabilor-G and ZrO2 Microspecimens,” Microsyst. Technol., Vol. 12, No.
7, 2006, pp. 713–716.
[10] Ilzhöfer, A., Schneider, H., and Tsakmakis, C., “Tensile Testing Device for Microstructured
Specimens,” Microsyst. Technol., Vol. 4, No. 1, 1997, pp. 46–50.
Maxim N. Gussev,1 Jeremy T. Busby,1 Kevin G. Field,1

Mikhail A. Sokolov,1 and Sean E. Gray2
Role of Scale Factor During

Tensile Testing of Small
Specimens
Reference
Gussev, Maxim N., Busby, Jeremy T., Field, Kevin G., Sokolov, Mikhail A., and Gray, Sean E.,
“Role of Scale Factor During Tensile Testing of Small Specimens,” Small Specimen Test
ABSTRACT
The influence of scale factor (tensile specimen geometry and dimensions) on
mechanical test results was investigated for different widely used types of small
specimens (SS-1, SS-2, SS-3, and SS-J3) and a set of materials. It was found that
the effect of scale factor on the accurate determination of yield stress, ultimate
tensile stress, and uniform elongation values was weak; however, clear
systematic differences were observed and should be accounted for during
interpretation of results. In contrast, total elongation values were strongly
sensitive to variations in specimen geometry. Modern experimental methods like
digital image correlation allow the impact of scale factor to be reduced. Using
these techniques, it was shown that true stress–true strain curves describing
strain-hardening behavior were very close for different specimen types. The
limits of miniaturization are discussed, and an ultra-miniature specimen concept
was suggested and evaluated. This type of specimen, as expected, may be
suitable for scanning electron microscopy and transmission electron microscopy
in situ testing.
Manuscript received February 7, 2014; accepted for publication May 13, 2014; published online August 29,
2014.
1
Oak Ridge National Laboratory, Oak Ridge, TN, 39831, United States of America.
2
The Univ. of Michigan, Ann Arbor, MI, 48109, United States of America.
3
Houston, TX.
Copyright V
Keywords
miniature samples, scale factor, digital image correlation, SS-1, SS-2, SS-3, SS-J3,
true curves
Introduction
Small specimens are widely used for investigation of mechanical properties and
deformation-hardening behavior of irradiated metals and alloys [1,2], nanostruc-
tured materials, and different composites. A survey of the opened literature shows
that a variety of different sizes and geometries are used in nuclear material science
[1–3], and the choice of geometry depends on the availability of materials and pecu-
liarities of test methods.
It is also well known that dimensions and geometry of the sample can influence
test results [1,3]. In recent decades, numerous industrial standards have been developed
to help obtain consistent results. However, it is not always possible to follow these
standards, especially in the nuclear industry, where specimen miniaturization is impor-
tant because of limited specimen volume and safety requirements post-irradiation.
As was observed in Ref 4, data scattering and a wide variety of specimen geome-
tries may be an issue during literature data survey and data mining in analytical
research. One of the goals of the present work is to compare mechanical properties
and true stress–true strain curves obtained using commonly found tensile specimens
types in nuclear materials literature. Analysis of the literature also revealed that
researchers are moving toward the use of even smaller specimens; so, checking the
limits of miniaturization was also of interest. In the present work, the concept of new
small specimen type is introduced, and its advantages and disadvantages are analyzed.
Experimental Procedures
Conventional well-known materials, as well as advanced steels, were studied in
the present work; the goal was to examine materials with different structures (face-
centered-cubic versus body-centered-cubic) and different deformation-hardening
behaviors. Austenitic AISI 316L and ferritic A533B steels were selected as being rep-
resentative of widely used materials. F82H steel is an example of modern ferritic
material, and modified cast high-nitrogen stainless steel (designated as 213L) [5]
represented newly developed austenitic materials (see Table 1).
316L stainless steel was supplied as an annealed flat bar with an average grain
size of 50 lm. The bar was 20 % cold rolled and used for specimen machining.
Part of the specimens was used in a cold-worked (CW) condition; other specimens
were annealed at 1323 K for 30 min (cooling with furnace, the final average grain
size: 34 lm) and at 1163 K (30 min, also cooling with furnace, the final average
grain size: 24 lm) to provide different strength levels. Cold work and intermediate
annealing at 1163 K were used to simulate a radiation-hardening effect. In the
GUSSEV ET AL., DOI 10.1520/STP157620140013 33
TABLE 1 Element composition of the investigated materials.
Composition (wt. %)
Material Type C Mn Si S P Cr Ni Cu Mo N Others
316L Commercial 0.0180 1.740 0.420 0.260 0.04 16.9 10.1 0.34 2.05 0.08
213L Cast steel 0.01 5.1 0.45 0.01 0.03 17.7 12.6 2.8 2.0 0.32 W:1 %.
F82H Ferritic steel 0.09 0.21 0.10 n/d n/d 7.46 – – – 0.006 1.96 W
A533B RPV steel 0.21 1.31 0.24 0.012 0.009 0.08 0.65 0.11 0.54 0.011
context of the presented work, possible negative effects caused by annealing at

1163 K, such as sensitization, were neglected.
An advanced steel, designated in the present work as 213L, was developed as an
improved cast material for the international thermonuclear experimental reactor
(ITER) shield modules [5]. The steel contained high levels of Mn, N, and W, and,
because of casting, had the largest grain size (1–2 mm) of any material investigated.
Nitrogen is known as a powerful solid solution strengthener [6], and a high Mn
content should increase the solubility of N and provide better phase stability [6]. As
expected, W in combination with Mo promotes additional strength and may
improve pitting corrosion resistance [5].
The purpose of F82H steel, developed by the Japan Atomic Energy Research
Institute (JAERI) and the NKK Corporation [7], was to produce reduced activation
ferritic–martensitic steel by substituting W and V for Mo and Nb. After
standard heat treatment (austenization for 30 min at 1313 K followed by an air
cold þ tempering for 1 h at 1013 K), this steel had a 100 % martensite structure [7].
Steel A533B is widely used for the construction of both nuclear and non-
nuclear pressure vessels. The material used in the present work was normalized at
1186 K and then annealed at 871 K for 4 h followed by quenching in water. The
material was then tempered at 936 K for 4 h (cooling in furnace). Stress-relief
annealing was conducted at 894 K for 40 h followed by cooling in the furnace.
Figure 1 shows the geometry and dimensions of the specimen types used in the
present work, as well as their dimensional tolerances. Prior to the tensile test, the
thickness and width of each specimen were measured with an accuracy of 10 lm or
better, so it was believed the dimension inaccuracy did not impact the strength
data; gauge length variations (60.02 mm) were too small to influence plasticity
results.
The SS-1 specimen type was developed in the 1980s for material science irradi-
ation programs in the EBR-2 reactor [8]. Later, to reduce the volume occupied
by an individual sample (Table 2) and to decrease residual activity levels, the SS-2
specimen type was developed [9]; because it is a thin specimen, SS-2 is especially
suitable for ion irradiation experiments, but it is also the most sensitive to occa-
sional damage because of sample handling.
FIG. 1 Geometry and dimensions of tested miniature specimens and ASTM subsize
sample. All dimensions are in mm; “T” designates thickness. Dimensional
tolerances for small specimens were as following: gauge length 60.02 mm,
gauge width 60.02 mm, specimen thickness 60.01 mm, and other dimensions
60.04 mm. For ASTM subsize specimen: 60.1 mm.
Later, in the middle of the 1980s for irradiation in the fast flux test facility, the
SS-1 specimen design was changed, and SS-3 specimen with a much shorter gauge
was offered [10]. In the 1990s, the SS-J3 specimen, smaller than SS-3, was offered
by Japanese researchers. As was noted in the introduction, the increasing trend in
the specimen geometry evolution has been to decrease length, volume, and weight.
Table 2 shows some important geometric parameters that could be relevant to
the data analysis described below. These parameters are thickness-to-width ratio
(T/W), length-to-width ratio (L/W), shoulder curvature (R), and shoulder
curvature-to-width ratio (R/W). The T/W parameter affects neck geometry and the
angle between the neck and load axis [11]; the latter two parameters influence stress
concentration near the shoulders and can cause a change in the yield stress value
obtained from the engineering curves. Shoulder curvature can also affect the geom-
etry of the triangle-like “dead zone,” which often forms near the grips [3]. As noted
TABLE 2 Dimensions and base geometry parameters of specimens.
Occupied Gauge Cross- T/W L/W Shoulder

Volume Length section, (thickness/ (length/ Curvature R/W
Type (mm3) (mm) (mm2) width ratio) width ratio) R (mm) Ratio
SS-1 168 20.32 1.16 0.5 (thick) 13.5 (long) 6.4 4.2
SS-2 21 12.70 0.25 0.2 (thin) 12.7 (long) 2.4 2.4
SS-3 96 7.62 1.16 0.5 (thick) 5 (short) 4.0 2.7
SS-J3 48 5.00 0.90 0.6 (thick) 4.5 (short) 1.4 1.3
ASTM 6000 28 18 0.5 4.7 6 1
subsize
in Ref 3, such a zone can demonstrate decreased or even zero strain levels compared
to the rest of the gauge. In simplest form, the specimen geometries can be classified
based on their thickness and length. First, SS-2 is described as being a thin speci-
men compared to SS-1 and SS-3, which are considered to be thick (Table 2), and
specimens SS-1 and SS-2 are long (L/W > 10) compared to SS-3 and SS-J3, which
are described as being short.
According to ASTM E8/E8M-15, uniform elongation should include both plastic
and elastic elongation; uniform elongation shall be determined using autographic
methods with extensometers. However, if no extensometer was used during the ten-
sile experiment and if strain was calculated from the beam-displacement value, then
elastic elongation included not only the elastic deformation of the specimen but also
the elastic contribution of the loading system (grips, beams, etc.). In this case, calcula-
tion of deformation values according to ASTM practice can lead to systematic error,
which will depend on tensile machine stiffness, grips, and many other factors. There-
fore, in the present work, the elastic portion of the total and uniform elongation val-
ues was excluded to eliminate this source of systematic inaccuracy. In other words, in
the present work, plasticity values show only plastic deformation.
Specimens for mechanical testing were produced from bulk material using an elec-
tric discharge machine (EDM). Some of the specimens were also mechanically ground
by a vendor to fit the dimensions. All specimens were produced by the same vendor.
EDM cutting and subsequent grinding may lead to a damaged surface layer of up to
10–20 lm [12]. Therefore, electro-polishing was conducted prior to mechanical testing
to eliminate the effect of different surface conditions. Chemistry solution compositions
for different steels were obtained from the literature, and electro-polishing modes (volt-
age, duration) were adjusted as necessary to get deformation-free surfaces. The thick-
ness of the layer removed by electro-polishing varied from 20 to 30 lm.
To conduct a comparative analysis, ASTM subsize specimens (Fig. 1) were pro-
duced from F82H steel by EDM; no additional surface treatment was performed on
this specimen. Similar subsize specimens with the same gauge dimensions were
used in Ref 13.
Figure 2 shows the scheme and general view of grips used in the present work.
The grips were configured to provide shoulder loading of the specimens during
mechanical testing. This style of grips and the same loading scheme were used for
all miniature specimens. Grips varied by shoulders curvature, and grips were used
for different specimen types. ASTM subsize specimens were tested using pneumatic
grips. Miniature specimens were tested on an MTS “Insight 2-52” screw-driven
machine with a force capacity of 2 kN. ASTM subsize specimens were tested on an
MTS servo-hydraulic machine with a 407 controller. All tensile experiments were
conducted at room temperature with strain rate 0.001 s1. Variations in strain rate
for different specimen types did not exceed 2 %–3 %. For each data point, three
identical specimens were tested.
All tensile experiments were conducted using optic non-contact measurements
and digital image correlation (DIC) algorithms. Optic strain measurements are
FIG. 2 Scheme and image of grips with loaded specimen. Note that shoulder curvature
of the grips varies for different specimen types.
based on comparing images of an object (tensile specimen) before and after defor-
mation. Details of the optic extensometry method and DIC are discussed elsewhere
[14–16]. An Allied Vision Technology GX3500 digital camera was used to obtain
images during tensile experiments; the resolution was 10–15 lm per pixel, depend-
ing on specimen type. To calculate strain fields and true stress–strain curves, VIC-
2D commercial software and in-house-developed software were used. For all tested
specimens, true stress–true strain curves were calculated for the middle third part
of the gauge. These experimental true curves were compared with true curves calcu-
lated from engineering diagrams. Such calculations [17] are a commonly used
method to extract additional data from tensile tests.
Experimental Results and Discussion

ENGINEERING DIAGRAMS OF TESTED MATERIALS
Figure 3 shows typical engineering tensile curves obtained for the different specimen
geometries and alloys investigated. One can see that scattering is relatively small for
specimens of the same type, and that the engineering curves are located close to
each other. At the same time, different specimen types demonstrate clear systematic
differences in elongation value and strength level. In particular, the SS-2 specimen
type shows the lowest total elongation value for all materials except cast 213 steel
(Fig. 3).
It is interesting that the SS-2 specimen type did not show Lüders deformation
and yield plateau for ferritic A533B steel, which were visible on the other specimen
types for A533B steel. This unexpected behavior could have been caused by a rela-
tively small T/W ratio for the SS-2 specimen.
FIG. 3 Engineering tensile curves “engineering strain–retain engineering strain:” 1–SS-1,

2–SS-2, 3–SS-3, 4–SS-J3. For two materials (213 and 316L annealed at 1323 K),
three curves are shown; for the other materials, the one most typical curve per
specimen type is presented.
INFLUENCE OF SPECIMEN TYPE ON MECHANICAL PROPERTIES

Figure 4 shows standard mechanical properties—yield stress (YS), ultimate tensile
strength (UTS) stress, uniform elongation (UE), and total elongation (TE). The
data in Fig. 4 were sorted by increasing T/W ratio (SS-2, SS-1, SS-3, SS-J3); this
ordering also allowed the long (SS-1, SS-2) and the short (SS-3, SS-J3) specimens to
be compared.
Based on Fig. 4, the YS values for F82H and cast 213 steels can be considered as
almost geometry-insensitive values (except data for the ASTM subsize specimen
with a different gripping scheme). Error bars for these materials were close each to
other. For other materials, YS tended to increase with an increase in T/W ratio (see
data for SS-2, SS-1, and SS-3 specimens), but the SS-J3 specimen, in many cases,
especially for austenitic steels, showed a lower YS value than that of SS-3.
FIG. 4 Yield and ultimate stress and uniform and total elongation for the investigated
materials. Error bars show one standard deviation value. Specimen types are
sorted by T/W ratio, from the smallest to the largest (except SS-mini).
UTS, in general, demonstrated the same patterns as YS. For A533B steel, UTS
tended to increase with increasing T/W ratio, but this was not a general tendency.
For non-irradiated and irradiated JPCA (modified 316) and JFMS (modified 9Cr
ferritic) steels, it was previously shown [18] that YS and UTS values increased with
increasing T/W ratio; if the T/W value exceeds 0.18, YS and UTS values saturate.
A similar result was obtained in Ref 19.
Uniform elongation values increased with an increase in the T/W parameter.
The exceptions were cast 213 steel with a high grain size and cold-worked 316L
steel; the latter demonstrates unexpectedly high scattering of data and large error
bars. The SS-J3 specimen demonstrated a smaller UE value than SS-3 if UE
exceeded 20 %. There was no pronounced difference between SS-3 and SS-J3
types for the low-ductile materials (e.g., F82H). Note that the SS-J3 and SS-3
specimens are similar enough from a geometry point of view; the only significant
difference is the shoulder curvature R (Table 2). In the most cases (except CW
316L), the short specimens (SS-3, SS-J3) show higher UE than the long specimens
(SS-1, SS-2).
ASTM subsize specimen (Fig. 4, data for F82H steel) demonstrated lower YS
compared to miniature specimens and higher UTS; plasticity parameters were very
close to SS-3 and SS-J3 samples.
In general, YS, UTS, and UE were relatively insensitive to the specimen geome-
try for all investigated materials, except cast steel. F82H steel was relatively insensi-
tive to specimen geometry, whereas cast 213 and annealed 316L steels were
relatively sensitive. The difference between specimen geometry types did not exceed
15 %.
TOTAL ELONGATION VALUE

Total elongation value (Fig. 4) was found to be strongly sensitive to specimen type
and geometry. For the TE value, the data demonstrated a pronounced trend: TE
increased as the L/W value decreased; in other words, the shorter the specimen, the
larger the TE. It is possible that the reason for such behavior is more geometrical
than physical. Elongation parameters (both UE and TE) are relative values; one
uses initial gauge length (L) of the sample to calculate UE and TE. If L decreases
but the neck geometry and the duration of neck development (D) change insignifi-
cantly, then the relative contribution of necking (D/L) to the TE increases greatly.
Clearly, in this case, during the discussion of TE data, the results do not show a
pure scale factor effect but some artifact caused by the calculation method.
LOCALIZED ELONGATION AND NECKING

During irradiation, because of radiation-induced embrittlement, UE may disappear
after some damage dose [17]. In this case, material will demonstrate necking imme-
diately after yield stress. Hence, it was interesting to analyze the effect of scale factor
on the development of localized deformation and necking.
Figure 5 shows the value of localized elongation for different specimen types. In
the framework of the present research, localized elongation (D) can be defined as a
time interval (or mobile beam displacement value) that corresponds to the develop-
ment of the local neck. To compare different specimen geometries, localized elonga-
tion was measured in millimeters (physical displacement of the mobile beam of the
tensile frame), not in %, which allowed the difference in gauge length to be
eliminated.
As shown in Fig. 5, localized elongation, as a rule, increased with increasing
gauge length. The reason for this phenomenon could be complex. It is probable
that some stress concentration exists in the short specimens, and an increase in the
gauge length decreases it, allowing the neck to develop longer. It is also possible the
FIG. 5 Average localized elongation (difference between uniform elongation and total
elongation, see detail in text). Specimen types are in order of increasing gauge
length L.
effect might be caused by the relative shoulder curvature (R/W); the row SS-J3–SS-
3–SS-1 also corresponds to an increase in R/W parameter (Table 2).
A trend was exhibited for thick specimens with a T/W ratio of 0.5 or more
(Table 2), but the thin SS-2 sample failed to follow this rule and showed smaller
localized elongation. As described in Ref 11, total elongation and localized elonga-
tion decrease as the T/W ratio decreases; this effect is associated with the difference
in neck stress states for thin and thick specimens.
IMPACT OF SCALE FACTOR ON TRUE CURVES OBTAINED

BY OPTIC EXTENSOMETRY
Figure 6 shows true stress–true strain curves obtained by optic extensometry for
tested materials. Comparing true curves obtained for different specimen types, one
can see that there are no significant systematic differences between different speci-
men geometries. The results agree well except for cast steel, which will be discussed
below; the deviation between the curves for the different specimens, in general,
remains the same (3 %–5 % of stress value) as strain increases and can be explained
by an inaccuracy in cross-section measurement and the natural inhomogeneity of
specimens.
Analytical examination of true curves was beyond the scope of the present
research. Nevertheless, it should be noted that true curves for steel 316L in all struc-
ture states demonstrated parabolic hardening (Fig. 6). In general, these curves can
be described by a Swift equation with two free parameters (r ¼ k(e–r02/k2)0.5). This
equation was discussed in detail in Ref 17. True curves for ferritic steels show more
complex behavior. The whole curves, as a rule, cannot be fitted by a simple
FIG. 6 Typical true strain–true stress curves obtained with DIC. The curves were
obtained for the middle third part of the specimen gauge. 1–SS-1, 2–SS-2,
3–SS-3, 4–SS-J3.
parabolic equation. Some aspects of true curves examination for ferritic steels are
analyzed in Refs 20 and 21.
For the tested materials (except cast steel), a notable difference between data
for different specimens exists only for SS-2 specimens of A533B steel (Fig. 6). For
this specimen geometry, the true curves, also as engineering diagrams, do not dem-
onstrate the Lüders deformation in the area of small strains.
True curves for cast 213L stainless steel demonstrated very high scattering
(Fig. 7); in many cases, scattering of the stress values at the same strain level reached
25 %–30 %. This fact can be explained by large grain size. The gauge of the cast steel
FIG. 7 True strain–true stress curves for cast 213 stainless steel. 1–SS-1, 2–SS-2, 3–SS-3,
4–SS-J3. “E” indicates easy slip stage; “L” shows stage of “linear hardening.”
tensile specimen in most cases consists of a few grains, and the specimen, in terms
of Ref 22, can be considered multi- or quasi-single crystalline. During a tensile
experiment, such objects demonstrate formation of strong deformation relief, bend-
ing, etc. The terms “true strain” and “true stress” should be applied to this class of
objects very carefully. Detailed investigation of their plastic behavior is an interest-
ing and important task (see, for example, Ref 23), but it is out of scope of the pres-
ent work.
Some of the true curves for cast steel are similar to those for single-crystal metal
materials [24]. By analogy with single-crystal samples of similar stainless steel, it is
possible to allocate the stage of “easy slip” and stage of “linear hardening” (Fig. 7).
For a single-crystal sample, the stress and strain at this stage demonstrate a linear
relationship [24], but for the case of cast steel, this stage is not purely linear.
Because only a few grains are present in the gauge, the acting stress increases
according to a more complex law, which could be fitted by an empiric equation
with an extra parabolic term: r k1e þ k2e2. Also, for a single-crystal sample, the
parabolic hardening stage (r e0.5) should follow the linear hardening stage; for
the case of cast steel, parabolic hardening does not occur for the whole gauge, but,
as optic measurements show, parabolic hardening takes place only in the neck if the
local strain value exceeds 0.30–0.35.
COMPARISON OF CALCULATED AND EXPERIMENTAL TRUE CURVES

As was noted above, in most cases, tensile tests of small irradiated specimens
reported in the literature are being conducted without the use of an extensometer of
any kind; deformation value is being recorded as displacement of mobile beam.
This is not an issue as provided tensile diagrams and engineering mechanical prop-
erties are the only objective of the research.
However, engineering diagrams are often used to calculate “true stress r–true
strain e” curves [17,25], which are more informative [17] and can be useful in
finite-element analysis [26]. In this case, the difference between calculated and
measured true curves should be analyzed in detail.
Usually, the calculation of “r–e” curves are being performed using “constant
volume criteria” [27]. When doing this calculation, it is important to check how the
scale factor and specimen geometry influence such calculated curves. Also, it is
necessary to estimate the difference between the calculated and experimental true
curves.
Figure 8 shows experimental and calculated true curves for the investigated
materials. The calculation works only before the ultimate stress point [17]; there-
fore, the calculated curves were limited by the uniform elongation part of the tensile
curve. One can see a clear difference between the calculated and experimental true
FIG. 8 Comparison of experimental true curves obtained using DIC (thin lines with
markers) and true curves calculated from engineering diagrams (solid lines).
1–cast 213 SS; 2–annealed 316L; 3–316L annealed at 1163 K; 4–CW 316L; 5–F82H;
6–A533B.
curves, and this difference changes with material strength level and strain degree.
For instance, soft, coarse-grain cast 213L steel with a low deformation-hardening
rate (especially in the area of small strains) demonstrates very high divergence
between calculated and experimental true curves (up to 25 %, Fig. 8). For soft,
annealed 316 steel, the difference reaches a maximum (10 %) at e 0.25.
The most pronounced difference exists for SS-3 and SS-J3 specimens; for the
same material, SS-1 and SS-2 specimens show a much smaller divergence. This
result can be explained by larger gauge length value in the SS-1 and SS-2 samples.
Larger gauge length decreases the relative input of heads in the strain value meas-
ured from engineering diagrams.
Also, it is possible to see that the difference between calculated and experimen-
tal true curves decreased as material yield stress increased. For 316L steel with a
moderate strength level (annealed at 1163 K), the difference was less than that for
fully austenized 316L steel, and the difference did not exceed 2 %–3 % for CW 316L
steel.
These results allow one to conclude that calculated true curves for annealed,
soft materials may include significant inaccuracy. However, if yield stress exceeds
450 MPa and the material still has some uniform elongation, calculated curves
should be accurate enough.
The Concept of an Ultra-Miniature Specimen

In some cases, the amount of available material may be not enough to produce even
a SS-J3 specimen, or the activity level may require volumes less than that of SS-3
geometry. In such cases, researchers use smaller non-standard specimens produced
by various methods such as EDM, machining, or punching. For instance, Fukuya
et al. [28] used specimens with a 4-mm gauge punched out using a specific die.
Also, ultra-miniature specimen geometries that can facilitate in situ test
methods using analytical electron microscopy equipment such as scanning electron
microscopy (SEM) and transmission electron microscopy (TEM) have been devel-
oped and are widely used by many researchers. Such methods allow one to observe
dislocation movement and dislocation-defect or dislocation-grain boundary inter-
action directly, which provides further insights into the behavior of materials under
a stress environment [29].
The behavior of many polycrystalline materials is strongly influenced by grain
orientation and grain boundary character distribution [30,31]. One area of interest
is the development of deformation localization caused by grain-to-grain interaction
under straining [31]. Analysis of stress and strain localization based on grain orien-
tation and material properties requires the development special specimen geome-
tries capable of being utilized in analytical electron microscopy equipment. Several
considerations must be investigated in developing new specimen geometries includ-
ing electron equipment load capacity and factors such as electron beam shadowing
as a result of specimen morphology.
FIG. 9 General view and dimensions of SS-mini specimen type used in the present
work. All dimensions are in mm. Dimensional tolerances for this geometry:
gauge length 60.02 mm, gauge width 60.02 mm, and gauge thickness
60.01 mm.
Based on these considerations, a modified ultra-miniature specimen (SS-mini)

geometry was developed and a preliminary analysis was conducted (Fig. 9). This
specimen may be easily produced from the heads of SS-1 or SS-3 specimens using
EDM or by machining. The geometry significantly reduces any material limitation
issues. The reduced volume also has a marked benefit of reducing the residual sam-
ple activity level by nearly two orders of magnitude. At the same time, the gauge
contains enough grains to accumulate statistics and to study sensitivity-to-grain ori-
entation, e.g., twinning [30] or phase transformations [6].
Figure 10 shows the specimen loaded in a copper carrier coupon, which is then
loaded in a Gatan Model 671.DH single title liquid-nitrogen-cooled straining
holder. The copper carrier coupon serves as the straining holder grips by providing
shoulder loading and interfacing to the straining holder. The spring-like shape
allows for full extension of the holder without failure, limiting the possibility of
specimen loss during straining.
Configurations such as those shown in Fig. 10 allow for unique experimentation
such as pre-straining and post-straining grain orientation analysis using electron
backscattering diffraction (EBSD) or focused ion beam (FIB) TEM specimen prepa-
ration techniques on a special grain configuration (e.g., high Schmid factor or grain
FIG. 10 (a) Optical micrograph showing the ultra-miniature specimen loaded in a Gatan
Model 671.DH holder; and (b) screen capture of a slip line interacting with a
grain boundary utilizing the in situ test configuration on an annealed 316SS.
boundary inclined 90 to the load direction). A screen capture showing slip line
and dislocation pile-up using the in situ TEM configuration is shown in Fig. 10(b).
The specimen was prepared using an EBSD-assisted FIB sample preparation. The
image shown in Fig. 10(b) is an example of the type of information that can be
obtained; the nature of the channeling dynamics will not be discussed here. Such
techniques can provide valuable insights into the deformation mechanics of speci-
mens. The geometry also lends itself to larger scale in situ observations such as the
use of a scanning electron microscope.
Limited mechanical properties data obtained using the proposed SS-mini-speci-
men type are shown in Fig. 4. One can see that the SS-mini-type provides smaller
yield and ultimate stress data than SS-3 (at 10 %) and demonstrates significantly
smaller ductility. In general, the results are close to the SS-2 specimen-type data.
One may expect that plasticity values may be improved by some increase in speci-
men thickness.
Conclusions
In the present work, the influence of scale factor (tensile specimen dimensions and
geometry) on mechanical test results was investigated for different widely used
types of the small specimens (SS-1, SS-2, SS-3, and SS-J3) and a set of materials
(annealed and cold-worked 316L, F82H, A533B, etc.). The effect of scale factor on
yield and ultimate stress and uniform elongation value was found to be relatively
weak. However, a clear systematic difference existed and should be considered.
Total elongation value and localized deformation were strongly sensitive to the scale
factor.
Modern experimental methods like digital image correlation and optic extens-
ometry allow the effect of scale factor to be reduced by eliminating the contribution
of the out-of-gauge deformation into the elongation value. True stress–true strain
curves describing strain-hardening behavior were shown to be very close for differ-
ent specimen types.
ACKNOWLEDGMENTS
This research is supported by the U.S. Department of Energy, Office of Nuclear
Energy, for the Light Water Reactor Sustainability Research and Development Effort,
and through a user project supported by ORNL’s Center for Nanophase Materials Sci-
ences (CNMS), which is sponsored by the Scientific User Facilities Division, Office of
Basic Energy Sciences, U.S. Department of Energy. The writers would like to thank
Dr. Wang Yanli (ORNL) for help with tensile test of ASTM samples and D.P. Stevens
(ORNL) for valuable help with manuscript preparation. This manuscript was auth-
ored at the Oak Ridge National Laboratory, and managed by UT-Battelle LLC under
Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The pub-
lisher, by accepting the article for publication, acknowledges that the U.S. Government
retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce
the published form of this manuscript, or allow others to do so, for U.S. Government
purposes.
References
[1] Klueh, R. L., “Miniature Tensile Test Specimens for Fusion Reactor Irradiation Studies,”
Nucl. Eng. Des. Fusion, Vol. 2, 1985, pp. 407–416.
[2] Wakai, E., Nogami, S., Kasada, R., Kimura, A., Kurishita, H., Saito, M., Ito, Y., Takada, F.,
Nakamura, K., Molla, J., and Garin, P., “Small Specimen Test Technology and Methodol-
ogy of IFMIF/EVEDA and the Further Subjects,” J. Nucl. Mater., Vol. 417, Nos. 1–3, 2011,
pp. 1325–1330.
[3] Pierron, O. N., Koss, D. A., and Motta, A. T., “Tensile Specimen Geometry and the Consti-
tutive Behavior of Zircaloy-4,” J. Nucl. Mater., Vol. 312, Nos. 2–3, 2003, pp. 257–261.
[4] Busby, J. T., Hash, M. C., and Was, G. S., “The Relationship Between Hardness and Yield
Stress in Irradiated Austenitic and Ferritic Steels,”J. Nucl. Mater., Vol. 336, Nos. 2–3,
2005, pp. 267–278.
[5] Busby, J. T., Maziasz, P. J., Rowcliffe, A. F., Santella, M., and Sokolov, M., “Development of
High Performance Cast Stainless Steels for ITER Shield Module Applications,” J. Nucl.
Mater., Vol. 417, Nos. 1–3, 2011, pp. 866–869.
[6] Lee, T.-H., Oh, C.-S., Kim, S.-J., and Takaki, S., “Deformation Twinning in High-Nitrogen
Austenitic Stainless Steel,” Acta Mater., Vol. 55, No. 11, 2007, pp. 3649–3662.
[7] Jitsukawa, S., Tamura, M., van der Schaaf, B., Klueh, R. L., Alamo, A., Petersen, C., Schirra, M.,
Spaetig, P., Odette, G. R., Tavassoli, A. A., Shiba, K., Kohyama, A., and Kimura, A.,
“Development of an Extensive Database of Mechanical and Physical Properties for Reduced-
Activation Martensitic Steel F82H,” J. Nucl. Mater., Vols. 307–311, 2002, pp. 179–186.
[8] Bajaj, R., Shogan, R. P., Deflitch, C., Fish, R. L., Paxton, M. M., and Bleiberg, M. L., “Tensile
Properties of Neutron-Irradiated Nimonic PE16,” Effects of Radiation on Materials Tenth
Conference, ASTM STP 725, D. Kramer, H. R. Brager, and J. S. Perrin, Eds., ASTM Interna-
tional, West Conshohocken, PA, 1981, pp. 326–351.
[9] Tanaka, M. P., Bloom, E. E., and Horak, J. A., “Tensile Properties and Microstructure of
Helium Injected and Reactor Irradiated V-20Ti,”J. Nucl. Mater., Vols. 103–104, 1981, pp.
895–900.
[10] Bauer, R. E. and Truitt, R. W., “Materials, Open Test Assembly (MOTA) Irradiation in
FFTF,” Effects of Radiation on Materials: 12th Conference, ASTM STP 870, F. A. Garner
and J. S. Perrin, Eds., ASTM International, West Conshohocken, PA, 1985.
[11] Byun, T. S., Kim, J. H., Chi, S. H., and Hong, J. H., “Effect of Specimen Thickness on the
Tensile Deformation Properties of SA508 Cl.3 Reactor Pressure Vessel Steel,” ASTM STP
1329, 1998, pp. 575–587.
[12] Bleys, P., Kruth, J.-P., Lauwers, B., Schacht, B., Balasubramanian, V., Froyen, L., and
Humbeeck, J. V., “Surface and Sub-Surface Quality of Steel After EDM,” Adv. Eng. Mater.,
Vol. 8, Nos. 1–2, 2006, pp. 15–25.
[13] Garware, M., Kridli, G. T., and Mallick, P. L., “Tensile and Fatigue Behavior of Friction-Stir
Welded Tailor-Welded Blank of Aluminum Alloy 5754,” J. Mater. Eng. Perform., Vol. 19,
No. 8, 2012, pp. 1161–1171.
[14] Suzuki, K., Jitsukawa, S., Okubo, N., and Takada, F., “Intensely Irradiated Steel Compo-
nents: Plastic and Fracture Properties, and a New Concept of Structural Design Criteria
for Assuring the Structural Integrity,” J. Nucl. Eng. Des., Vol. 240, No. 6, 2010,
pp. 1290–1305.
[15] Schalk, P., Fauster, E., and O’Leary, P., “High-Temperature Video-Extensometry for Mate-
rial Testing of Refractories,” Proceedings of SPIE-IS&T Electronic Imaging, Vol. 5679,
SPIE, Bellingham, WA, 2005, pp. 129–139.
[16] Pan, B., Qian, K., Xie, H., and Asundi, A., “Two-Dimensional Digital Image Correlation for
In-Plane Displacement and Strain Measurement: A Review,” Measure. Sci. Technol., Vol.
20, No. 6, 2009, 062001.
[17] Gussev, M. N., Byun, T. S., and Busby, J. T., “The Relationship Between Hardness and
Yield Stress in Irradiated Austenitic and Ferritic Steels,” J. Nucl. Mater., Vol. 427, Nos.
1–3, 2012, pp. 62–68.
[18] Kohno, Y., Kohyama, A., Hamilton, M. L., Hirose, T., Katoh, Y., and Garner, F. A., “Specimen
Size Effects on the Tensile Properties of JPCA and JFMS,” J. Nucl. Mater., Vols. 283–287,
2000, pp. 1014–1017.
[19] Kohyama, A., Hamada, K., and Matsui, H., “Specimen Size Effects on Tensile Properties
of Neutronirradiated Steels,” J. Nucl. Mater., Vols. 179–181, 1991, pp. 417–420.
[20] Vanaja, J., Laha, K., Sam, S., Nandagopal, M., Panneer Selvi, S., Mathew, M. D., Jayakumar,
T., and Rajendra Kumar, E., “Influence of Strain Rate and Temperature on Tensile Proper-
ties and Flow Behavior of a Reduced Activation Ferritic–Martensitic Steel,” J. Nucl.
Mater., Vol. 424, Nos. 1–3, 2012, pp. 116–122.
[21] Spätig, P., Baluc, N., and Victoria, M., “On the Constitutive Behavior of the F82H Ferritic/
Martensitic Steel,” Mater. Sci. Eng. A, Vols. 309–310, 2001, pp. 425–429.
[22] Keller, C., Hug, E., and Feaugas, X., “Microstructural Size Effects on Mechanical Proper-
ties of High Purity Nickel,” Int. J. Plastic., Vol. 27, No. 4, 2011, pp. 635–654.
[23] Kim, K.-H., Kim, H.-K., and Oh, S.-I., “Deformation Behavior of Pure Aluminum Specimen
Composed of a Few Grains During Simple Compression,” J. Mater. Proc. Technol., Vol.
171, No. 2, 2006, pp. 205–213.
[24] Karaman, I., Sehitoglu, H., Maier, H. J., and Chumlyakov, Y. I., “Competing Mechanisms
and Modeling of Deformation in Austenitic Stainless Steel Single Crystals With and
Without Nitrogen,” Acta Mater., Vol. 49, No. 19, 2001, pp. 3913–3933.
[25] Fabritsiev, S. A. and Pokrovsky, A. S., “The Effects of Grain Size and Helium Accumula-
tion on Radiation Hardening and Loss of Ductility of Pure Copper for ITER Application,”
Fusion Eng. Des., Vol. 65, No. 4, 2003, pp. 545–559.
[26] Kim, J. W. and Byun, T. S., “Analysis of Tensile Deformation and Failure in Austenitic
Stainless Steels: Part II—Irradiation Dose Dependence,” J. Nucl. Mater., Vol. 396, No. 1,
2010, pp. 10–19.
[27] Gussev, M. N., Maksimkin, O. P., Osipov, I. S., and Garner, F. A., “Application of Digital
Marker Extensometry to Determine the True Stress-Strain Behavior of Irradiated Metals
and Alloys,” J. ASTM Int., Vol. 5, No. 4, 2008, JAI101073.
[28] Fukuya, K., Nishioka, H., Fujii, K., Miura, T., and Kitsunai, Y., “Local Strain Distribution
Near Grain Boundaries Under Tensile Stresses in Highly Irradiated SUS316 Stainless
Steel,” J. Nucl. Mater., Vol. 432, Nos. 1–3, 2013, pp. 67–71.
[29] Kacher, J. and Robertson, I. M., “Quasi-Four-Dimensional Analysis of Dislocation Interac-

tions with Grain Boundaries in 304 Stainless Steel,” Acta Mater., Vol. 60, No. 19, 2012,
pp. 6657–6672.
[30] Gussev, M. N., Byun, T. S., Busby, J. T., and Parish, C. M., “Twinning and Martensitic Trans-
formations in Nickel-Enriched 304 Austenitic Steel During Tensile and Indentation
Deformations,” Mater. Sci. Eng. A, Vol. 588, 2013, pp. 299–307.
[31] McMurtrey, M. D., Was, G. S., Patrick, L., and Farkas, D., “Relationship Between Localized
Strain and Irradiation Assisted Stress Corrosion Cracking in an Austenitic Alloy,” Mater.
Sci. Eng. A, Vol. 528, Nos. 10–11, 2011, pp. 3730–3740.
FRACTURE TOUGHNESS TESTING
Masato Yamamoto,1 Kunio Onizawa,2

Kentaro Yoshimoto,3 Takuya Ogawa,4 Yasuhiro Mabuchi,5
Matti Valo,6 Marlies Lambrecht,7 Hans-Werner Viehrig,8
Naoki Miura,1 and Naoki Soneda1
International Round Robin Test

on Master Curve Reference
Temperature Evaluation Utilizing
Miniature C(T) Specimen
Reference
Yamamoto, Masato, Onizawa, Kunio, Yoshimoto, Kentaro, Ogawa, Takuya, Mabuchi, Yasuhiro,
Valo, Matti, Lambrecht, Marlies, Viehrig, Hans-Werner, Miura, Naoki, and Soneda, Naoki,
“International Round Robin Test on Master Curve Reference Temperature Evaluation Utilizing
Miniature C(T) Specimen,” Small Specimen Test Techniques: 6th Volume, STP 1576, Mikhail A.
Sokolov and Enrico Lucon, Eds., pp. 53–69, doi:10.1520/STP157620140020, ASTM
International, West Conshohocken, PA 2015.9
ABSTRACT
The Master Curve (MC) method is a promising technique for evaluating the
fracture toughness of ferritic steels. It enables the determination of the reference
temperature, To, of a probabilistic fracture toughness curve using small
specimens. The Central Research Institute of Electric Power Industry (CRIEPI)
investigated this capability of the MC method using different size of C(T)
specimens, and it was found that 0.16T-C(T) specimen with the dimensions of 4
Manuscript received February 28, 2014; accepted for publication June 24, 2014; published online August 29,
2014.
1
Central Research Institute of Electric Power Industry, Yokosuka, Kanagawa 240-0196, Japan.
2
Japan Atomic Energy Agency, Tokai, Ibaraki 319-1184, Japan.
3
Mitsubishi Heavy Industries LTD., Takasago, Hyogo 676-8686, Japan.
4
Toshiba Co., Yokohama, Kanagawa 235-8523, Japan.
5
Hitachi-GE Nuclear Energy LTD., Hitachi, Ibaraki 317-0073, Japan.
6
VTT Technical Research Centre of Finland, Metallimiehenkuja, Espoo FI-02044, Finland.
7
SCKCEN, Belgian Nuclear Research Centre, Boeretang, Mol 2400, Belgium.
8
Helmholtz-Zentrum Dresden-Rossendorf, Bautzner Landstraße, Dresden 01328, Germany.
9
Houston, TX.
Copyright V
by 10 by 9.6 mm, hereafter called “Mini-CT specimen,” can be used to obtain

valid To values. The advantage of this Mini-CT specimen technique is that
multiple specimens can be machined from one of the broken halves of Charpy
size specimens, which are used in a standard surveillance capsule of a reactor
pressure vessel (RPV). In order to ensure the robustness of this technique, a
round-robin test was planned. The idea is to perform MC tests using Mini-CT
specimens by different investigators to see if consistent To values can be
obtained. All the specimens used were machined and pre-cracked by one
fabricator from unique RPV material to avoid any possible effect of specimen
preparation on To values. Seven institutes participated in this exercise, and
obtained valid To values. No specific difficulty was found in the MC tests
performed in accordance with the ASTM E1921-10el protocol. The scatter of the
obtained To values was well within the uncertainty range defined in Appendix
X4.2 of ASTM E1921, indicating the robustness of the Mini-CT specimen technique
in terms of the testing procedure. Throughout this activity, we could obtain
182 KJc (1Teq) for a single material. We investigated the statistics of this large
database, and found that there is no remarkable difference not only in the To
values, but also in the fracture toughness distribution between the Mini-CT
specimen and the standard 1T-C(T) specimen results.
Keywords
the Master Curve method, reactor pressure vessel material, fracture toughness
evaluation
Introduction
The Master Curve (MC) method [1,2] is promising as the fracture toughness
evaluation technique of reactor pressure vessels (RPVs) for two reasons: (1) that the
transition of fracture toughness can be evaluated by limited number of specimens
tested in one or a few temperature conditions, and (2) the fracture toughness value
can be converted to that of the different specimen size by means of the weakest
linking theory. This means that fracture toughness values can be evaluated by
means of small size specimens. If the MC evaluation can be done using very small
C(T) specimens, which can be machined from broken halves of Charpy size speci-
mens of surveillance material, additional estimation of fracture toughness in the
transition region can be made coexistent with the existing surveillance program.
ASTM E1921-10e1 [3] specifies the reference temperature evaluation procedure for
ferritic steels following to the MC method. Since ASTM E1921-10e1 does not limit
the available specimen size, very small C(T) specimens can be used as is standar-
dized. However, most of the back data, which had been used as the technical basis
of the standard, are of larger size, i.e., of 0.4T or larger C(T) specimens.
Scibetta et al. [4] firstly showed that such small C(T) specimens can be used for
the MC evaluation as well as the sub-size pre-cracked Charpy specimens. Central
Research Institute of Electric Power Industry (CRIEPI) also proposed a MC-based
fracture toughness evaluation procedure for very small size C(T) specimen with the
YAMAMOTO ET AL., DOI 10.1520/STP157620140020 55
dimension of 4 by 10 by 9.6 mm, hereafter called “Mini-CT specimen,” and verified

the reference temperature To values against up to 4T-C(T) specimens (101.6 mm)
of three different heats of RPV materials [5].
A round robin program was organized with the participation of academia,
industries, and government institutes. The program aims to verify the reliability
and robustness of the evaluation method based on experimental data measured
on Mini-CT. The interim reports with results from 4 organizations in Japan
(PVP2012-78661 [6], PVP2013-97936 [7]) suggested that the Mini-CT test tech-
nique is fairly robust in regard to differences in the testing machine and operator,
and the obtained To gives a similar loading rate dependency to the larger C(T)
specimens.
The round robin program achieved the extensive participation of up to 14
organizations in an international collaborative framework. The following 7 organi-
zations have submitted their results:
• Central Research Institute of Electric Power Industry (Japan)
• Helmholtz-Zentrum Dresden-Rossendorf (Germany)
• Japan Atomic Energy Agency (Japan)
• Mitsubishi Heavy Industry (Japan)
• SCK CEN (Belgium)
• Toshiba/Hitachi group (Japan)
• VTT Technical Research Center of Finland (Finland)
We identify the organizations by letters A, B, C, D, E, F, and G in the following.

It is noted that the order of the above list does not correspond to the order of ids
A to G.
In this paper, the authors first addressed the comparison of To evaluation results
among the organizations. Here, we used a small number, i.e., 8–12, of specimens for
the To evaluation. Second, we investigated the statistics of a whole database, which
includes 182 points of fracture toughness data from 7 organizations. The similarity of
Weibull plots between those of Mini-CTs and 1T-C(T)s is discussed. In addition, the
effect of the number of specimens on To was investigated by the randomly pick up of
7, 8, 10, 13, 17, or 20 data points out of the whole database.
Material and Experimental Procedure

The tested material is a Japanese RPV steel forging, SFVQ1A [8], whose chemical
composition is shown in Table I. Eq. (1) and (2) give the Young’s modulus E (GPa)
and Yield stress ry (MPa) versus temperature behavior from room temperature to
150 C investigated on several Japanese RPV materials including SFVQ1A [9].
(1) E ¼ 210 0:063ðT 20Þ

(2) ry ¼ 84 þ 271 expf110=ðT þ 273:15Þg
Here, T is temperature in C. The material was sampled from the broken half of a
4T-C(T) specimen, which was T–L orientation specimen taken from center part of
TABLE 1 Chemical composition of tested material, unit in wt. %.
Material (unit C Si Mn P S Ni Cr Mo V
in wt. %)
SFVQ1A Target 0.25 0.40 1.20 1.50 0.025 0.025 0.40 1.00 0.25 0.45 0.60 0.05
value
Product 0.18 0.18 1.46 0.002 <0.001 0.90 0.12 0.52 <0.01
value
160 (t) by 1050 (w) by 3000 (L) mm forging material (Fig. 1(a)). Blocks of 12 by 12
by 100 mm were machined from the broken half. The block was sliced into 20 pieces
of 4 by 10 by 9.6 mm (Fig. 1(b)). Every piece was numbered as specimen 01 to 20
according to its location in block. Here, the specimens 01 to 04 and 17 to 20 are
with an accuracy of 610 mm within 1=4 and 3=4 thickness location of the plate,
respectively. Each small piece was machined into Mini-CT specimen in T–L
FIG. 1 Configuration of tested material and specimen preparation.

FIG. 2 Geometry of Mini-CT specimen.
orientation. The geometry of Mini-CT (Fig. 2) follows the specifications in ASTM

E1921-10e1 with an exception in the width of wire-cut slit. Because of the limitation
in available minimum wire diameter (0.1 mm) used in slit machining, the slit
width became around 0.2 mm instead of 0.01 W (0.08 mm in Mini-CT) in ASTM
E1921-10e1. Here, W is specimen width and W ¼ 8 mm in the Mini-CT specimen.
The specimens were fatigue pre-cracked up to an a/W ratio of 0.5, whereby the
length of the fatigue crack was 0.8 mm optically measured on the side surfaces. The
total crack length with the machined notch length on the side surface was main-
tained to be 3.8 mm to consider some amount of tunneling toward the thickness
direction. After the testing the averaged crack length was measured with 9 points
method according to ASTM E1820-09e1 [10]. All the averaged crack length fell
within the range of 0.5 W 6 0.05 W and the difference between each 9 points mea-
surement and averaged crack length were less than 5 % of the averaged crack length
or 0.5 mm, as required in ASTM E1921-10e1. CRIEPI machined and pre-cracked
all the specimens before the distribution of them to the participated organizations.
All the 20 specimens from the single block were sent to a single organization as one
set of specimens.
Tensile or fatigue test machines between 10 to 100 kN capacity were used for
the fracture toughness tests. The liquid nitrogen spray in the thermostatic chamber
cooled specimen, clevis, and loading rods. Instead of direct temperature measure-
ment of specimens, organizations A, C, D, E, F, and G attached their thermocouples
to the clevis beside the specimen and controlled the test temperature. In those cases,
the temperature difference between specimen and clevis was determined before a
series of tests using a specimen with a thermocouple attached on it. Table 2 shows
an example of temperature comparison among clevis, specimen side surface near
the crack tip, and specimen back surface. The temperature difference between
TABLE 2 Example of temperature comparison among clevis, specimen back surface and specimen
side surface near crack tip.
Target Temperature Clevis (Control), Specimen Back Surface Specimen Side Surface

C C C Near Crack Tip C
100 100 101 100

110 110 111 111
120 120 121 120
130 130 131 130
140 140 141 140
150 150 150 150
specimen and clevis was at most 1 C. Organization B monitored and controlled

the specimen temperature with thermocouples attached to the back surface of the
specimen by means of a spring fixture mechanism. The crack mouth opening dis-
placement was measured at the front face of the specimen by means of a clip-on
gage with 3.0 mm gage length. The measure of the clip-on gage was converted into
the load line displacement by multiplying factor of 0.75 under the assumption of
that specimen rotation center during the loading deformation is at the middle point
of ligament length of C(T) specimen [5]. The converted load line displacement was
used for determination of J integral value at fracture.
Table 3 shows the test conditions together with the results. The round robin
tests were conducted in conformity with ASTM E1921-10e1. The test temperature
was specified to 120 C. Loading rate at the initial elastic region was selected from
the range of 0.1 to 2 MPaHm/s. Eight specimens of a single block numbered as 01
to 04 and 17 to 20, which were located near 1=4 and 3=4 thickness location of the
plate, were tested as the mandatory requirement in the round robin test. Additional
tests were carried out when the number of fracture toughness data points was not
sufficient to obtain valid To using specimens taken from the mid-thickness location
of the same block. It was separately confirmed that no remarkable difference in
fracture toughness can be seen between 1=4 T and the center location [6]. As a result,
8–12 specimens in each dataset were tested. The 16 datasets named A1 to G2,
which appear above the doublet in Table 2, are those matched to the round robin
requirement. Datasets A1a, A2a, A3a, and A3b are additional data and used only in
the statistical analysis of the whole dataset described later. It is noted that all the To
data in the paper were calculated in CRIEPI based on fracture toughness data pro-
vided by each organization.
Comparison of To Among the Organizations

Columns N and r of Table 3 shows the number of tested specimens and number
of valid data. Some of the fracture toughness data KJc(1Teq), which were
TABLE 3 Test conditions and results.
dK/dt Test
Dataset Block MPa Temperature, Validity r

Number Number Organization Hm/s C N r To (TQ) of To b (Eq 4)
A1 F225 A 0.3 120 11 8 107 Valid 18 7.52

A2 F220 A 0.5 120 11 7 110 Valid 18 7.89
A3 F204-1 A 2 120 12 8 107 Valid 18 7.52
B1 F219,227 B 0.1 120 11 8 109 Valid 18 7.52
B2 F222 B 0.5 120 10 9 103 Valid 18.8 7.43
B3 F219 B 1.9 120 11 11 99 Valid 18.8 6.94
C1 F228 C 0.5 120 12 8 111 Valid 18 7.52
C2 F223 C 0.7 120 10 10 98 Valid 18.8 7.17
D1 F229 D 0.5 120 8 7 95 Valid 18.8 8.15
D2 F224 D 1.4 120 8 8 (88) Invalid 18.8 7.76
(load
control)
E1 F233 E 0.5 120 8 4 (118) Invalid 18 9.85
(Lower
number
of valid
data)
E2 F232 E 0.5 120 8 4 (118) Invalid 18 9.85
(Lower
number
of valid
data)
E3 F205 E 0.5 120 8 7 100 Valid 18.8 8.15
F1 F245 F 0.5 120 10 9 95 Valid 18.8 7.43
G1 F238 G 0.3 120 10 8 102 Valid 18.8 7.76
G2 F239 G 0.3 120 10 9 99 Valid 18.8 7.43
A1a F225 A 0.3 120 7 7 — — — —
A2a F220 A 0.5 120 7 6 — — — —
A3a F204-1 A 2 120 4 3 — — — —
A3b F204-2 A 1.3 120 4 3 — — — —
Note: Last 2 digits in block number are corresponding to the block number in Fig. 1(a) and showing
specimen location in the original 4T-C(T) specimen.
converted to equivalent to 1T-C(T) specimen size, were determined as invalid

due to exceeding the limit value KJc(limit), which is given in ASTM E1921-10e1 as
follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Eb0 ry
(3) KJcðlimitÞ ¼
30ð1 2 Þ
where:
b0 ¼ ligament length between crack front and specimen back surface, W-a0,
and
¼ Poisson’s ratio (¼ 0.3).
The invalid data were censored (replaced by KJc(limit)) and used for the reference
temperature determination. Among the 16 datasets from A1 to G2, the reference
temperature by D2 was determined as invalid because the test was conducted in
constant loading rate control instead of stroke rate control. This determination was
simply made because of the violation of control mode specification in ASTM
E1921-10e1. The reference temperature by E1 and E2 were also determined as
invalid because of low number of valid data. For the other 13 datasets, To was deter-
mined as valid. ASTM E1921-10e1 Appendix X4.2 gives the standard deviation on
the estimate on To due to material inhomogeneity and experimental errors as
follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
b2
(4) r¼ þ r2exp
r
Here, b is sample size uncertainty factor, which varies with the magnitude of the
median value of KJc, KJc(med) (18 C for KJc(med) = 83 MPaHm, 18.8 C for 83 MPaHm
> KJc(med) = 66 MPaHm and 20.1 C for 65 MPaHm > KJc(med) = 58 MPaHm). rexp
is the contribution of experimental uncertainties (rexp ¼ 4 C). The standard devia-
tion, r, for each of the dataset is shown in Table 3.
Figure 3 shows the Weibull plot for each organization. The MC concept is based
on the weakest link theory and the scatter of fracture toughness is considered to fol-
low the Weibull distribution with three parameters, Kmin (location parameter, fixed
to 20 MPaHm), Weibull exponent n (shape parameter, fixed to 4), and K0 (scale
parameter). p is the median rank as the cumulative failure probability.
(5) p ¼ ði 0:3Þ=ðN þ 0:4Þ
Here, i is the rank and N is the number of specimens. In Fig. 3, Kmin is fixed to
20 MPaHm, while n is not fixed. As shown in Fig. 3, Weibull exponents, n, were
determined from 2.2 to 4.4.
Figure 4 shows the relationship between the number of specimens and the
Weibull exponent in comparison with data points in literature [1,11]. Solid lines
show the 2.5 and 97.5 % tolerance bounds, which are theoretically obtained with
the assumption of Kmin ¼ 20 MPaHm. The results of the present study were in the
range of variation in previous studies, and fall between the 2.5 and 97.5 % tolerance
bounds. The results by Figs. 3 and 4 confirmed that the distribution of KJc values
obtained by Mini-CT specimens conformed to the assumption of the MC method,
i.e., Kmin ¼ 20 and n ¼ 4.
Figure 5 compares the KJc (med), which is the median value of KJc (1Teq) in a data-
set. The error bar shows the standard deviation of valid KJc(1Teq). The datasets D2,
FIG. 3 Weibull plot of each dataset.

FIG. 4 Distribution of Weibull exponent by each dataset.
E1, and E2, whose reference temperatures were determined as invalid, showed the
lowest (D2) and the highest (E1 and E2) KJc (med). All the KJc (med) of the 13 datasets
excluding D2, E1, and E2, are located around 80 MPaHm, and the error bars were
well overlapping each other.
Figure 6 compares the reference temperature among the datasets. The error bar
shows the range of 6r of Eq 4. The scatter of 13 datasets with valid reference
FIG. 5 Comparison of KJc(med).

FIG. 6 Comparison of To.
temperatures To is less than 16 C and the error bars were well overlapping each
other. TQ by dataset D2 was located at the highest of TQ and To values, but the error
bars were still overlapping with those of valid Tos. This suggests that the loading
rate effect due to the difference in control mode is limited in the present conditions.
The comparison of To confirms that Mini-CT test technique is robust and different
machines and operators can obtain similar To by using the requirements as speci-
fied in ASTM E1921-10e1.
Statistics of Whole Data Sets

As shown in Fig. 1, all the specimens used in the round robin program were taken
from the half of a broken 4T-C(T) specimen and machined in the same factory.
Thus, it can be said that all the data in Table 2 were determined with the same mate-
rial and tested on comparable conditions. We summarized all KJc(1Teq) data from
the 7 organizations to a big dataset of 182 values, including 38 invalid data to check
the statistics.
Figure 7 shows the Weibull plot of the whole dataset. All the valid data aligned
linearly in the viewgraph. The Weibull exponent obtained from valid data is 3.7,
which is close to the assumption of the MC method, i.e., n ¼ 4.0. The closed sym-
bols are the Weibull plot obtained using 1T-C(T) specimens of the same material
[9]. Plots by Mini-CT and 1T-C(T) were overlapping each other very well. To eval-
uated with all data of Mini-CT specimens was 104 C, which is 8 K lower than
that obtained from 1T-C(T) specimens. This fact is consistent with the knowledge
that larger plastic strain due to local loss of constraint may give lower T0 in smaller
specimens [12]. However, the T0 difference, 8 K, is much smaller than the expected
scatter range come from the difference of data sets. Thus, it can be said that the
effect of local loss of constraint on T0 is not significant in the given conditions.
FIG. 7 Weibull distribution of whole dataset.
A set of 144 valid data points of Mini-CT and another set of 8 valid data points
of 1T-C(T) in Fig. 7 were subjected to Welch’s t-test [13] to clarify whether they
were extracted from the population with the same average value or not. Here,
Welch’s t-test was selected because it gives fairly robust results for the various
shapes of the population; however, it assumes that the population follows the nor-
mal distribution. In the t-test, we made a null hypothesis that no difference is in the
population mean value of two data sets. Here, we set the significance to 0.05. Then
test statistic tw can be calculated by the following equation:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(6) tw ¼ jX1 X
2 j= U1 =n1 þ U2 =n2
where:
n1 and n2 ¼ the size of the data set,
U1 and U2 ¼ the unbiased variance, and
1 and X
X 2 ¼ the average.
tw follows t distribution, whose df is :
ðU1 =n1 þ U2 =n2 Þ2

(7) ¼
ðU1 =n1 Þ =ðn1 1Þ þ ðU2 =n2 Þ2 =ðn2 1Þ
2
t (, 0.05) shows the t-distribution for given and significance level (0.05). If t (,
0.05) > tw, the hypothesis is adopted. Results are summarized in Table 4. The result
is tw < t (, 0.05), and P > 0.05. Therefore, the null hypothesis is adopted; it cannot
be said that there is a difference in the population mean value of KJc(1Teq) by Mini-
CT specimens and 1T-C(T) specimens.
TABLE 4 Summary of Welch’s t-test on Mini-CT and 1T-C(T) specimens.
Mini-CT 1T-C(T)
Number of data ni 144 8

Average 73.2 75.0
Unbiased variance Ui 206 214
Test statistic tw 0.3363
Degree of freedom 7.77
t(, 0.05) 2.306
Note: i ¼ 1 for Mini-CT, 2 for 1T-C(T).
Effect of Number of Specimens

on Evaluation Results
We randomly picked up KJc(1Teq) datasets from whole dataset. The size of dataset
was 7, 8, 10, 13, 17, or 20. We carried out 1000 trials for each size of the sampled
dataset. Mersenne twister algorithm [14] was used for the random sampling.
KJc(1Teq) sampling in a dataset was done with replacement. The Weibull exponent
and To were evaluated using the sampled datasets.
Table 5 summarized the simulated results. In the present study, the test temper-
ature T was about 20 C lower than To. Thus, 7 valid KJc(1Teq) data points were
required for the To evaluation in most cases. This led to the lower number of valid
TABLE 5 Summary of simulation.
N¼7 N¼8 N ¼ 10 N ¼ 13 N ¼ 17 N ¼ 20
Number of valid To in 1000 236 637 958 999 1000 1000

trials
Average of Weibull exponent 3.73 3.70 3.71 3.66 3.66 3.66
for all 1000 trials
Standard deviation of Weibull 1.73 1.50 1.19 0.94 0.78 0.69
exponent for all 1000 trials
Average of TQ, C, of all 1000 103.8 104.5 103.8 104.3 104.3 104.3
trials
Standard deviation of TQ by 8.0 7.5 6.6 5.6 4.6 4.5
analyses, C, of all 1000 trials
Average of To, C, of valid 97.5 101.2 103.2 104.2 104.3 104.3
trials
Standard deviation of To by 6.2 6.0 6.0 5.6 4.6 4.5
analyses, C of valid trials
Standard deviation by ASTM 7.9 7.5 7.0 6.4 5.9 5.7
E1921 Appendix X4.2
FIG. 8 Averaged Weibull exponent and its 95 % confidence interval.
To for the datasets with N ¼ 7. For the datasets with N ¼ 8, 66 % of the datasets
were eligible for the valid To evaluation.
Figure 8 shows the relationship between the number of specimens and the
Weibull exponent. The large circular symbols show the averages of n for all the
1000 trials for each size of dataset. They were around 3.7 and close to the assump-
tion of the MC method. The broken lines show the 2.5 and 97.5 % tolerance bounds
of scatter range in the simulation, while the solid lines show those of theoretically
obtained ones with the assumption of Kmin ¼ 20 MPaHm. The tolerance bounds of
the present study were located nearby and inside the theoretical one.
Figure 9(a) shows the average of TQ for all the 1000 trials for each size of data.
The obtained TQ is around 104 C and data size insensitive. The error bar shows
the standard deviation of analysis results. Circular symbols show the standard devi-
ation calculated by Eq 4, but using fixed b ¼ 18 and r ¼ N values for simplicity. For
the conditions of N ¼ 7 and N ¼ 8, the circular symbols and the error bars are very
close to each other; however, Eq 4 still gives reasonable margin. For the cases with
N ¼ 10 and grater, the error bars are consistently smaller than the margin showed
by circular symbols. Figure 9(b) compares the average of valid To for each size of the
dataset. Among the scatter in 1000 time trials, the lower region of TQ tended to be
determined as invalid because of a low number of valid KJc(1Teq). This was because
the dataset included invalid KJc(1Teq) due to exceeding KJc(limit). Thus, the averaged
To was increased after the removal of the invalid data. This is the reason why the To
for the datasets with N ¼ 7 and N ¼ 8 show slightly higher values than those of
N ¼ 10 and greater. The standard deviation of the analysis result was consistently
smaller than that as specified in ASTM E1921 X4.2. This result confirms that the
FIG. 9 Averaged and standard deviation of TQ and To.
statistical distribution of To determined with Mini-CT specimens is similar to those

of larger specimens, which were used as the technical basis of ASTM E1921.
The random sampling simulation results indicate that limited numbers of repli-
cated Mini-CT specimens (8 or 10 specimens) are eligible for the To determination
and the evaluation results are consistent to the assumption of the MC method.
Conclusion
Round robin tests of the MC fracture toughness evaluation using the miniature size
C(T) (Mini-CT) specimens were carried out in 7 organizations. The Mini-CT
specimens were manufactured by CRIEPI from the same piece of reactor pressure
vessel steel. Sixteen datasets, including 182 fracture toughness data, were obtained.
Comparison of the reference temperature among the organizations and the statisti-
cal analysis on the whole data were made. The results are summarized as follows:
1. The Mini-CT test technique is robust for the MC evaluation of RPV material;
many organizations could carry out the test and evaluation. Scatter of 13 valid
reference temperatures was less than 16 C.
2. Weibull exponent obtained by whole 182 data points is close to 4. Weibull
plots of KJc(1Teq) obtained by Mini-CT and 1T-C(T) overlap each other. These
results confirm the applicability of the MC method on Mini-CT testing.
3. The scatter of the Weibull exponent and reference temperature in Mini-CT
testing was about the same magnitude of those expected in the testing with
larger specimens.
References
[1] Wallin, K., “The Scatter in KIC-Results,” Eng. Fract. Mech., Vol. 19, No. 6, 1984, pp.
1085–1093.
[2] Wallin, K., Saario, T., and Torronen, K., “Statistical Model for Carbide Induced Brittle
Fracture in Steel,” Metal Sci., Vol. 18, 1984, pp. 13–16.
[3] ASTM E1921-10e1: Standard Test Method for Determination of Reference Temperature To,
for Ferritic Steels in the Transition Range, ASTM International, West Conshohocken, PA,
2010, www.astm.org.
[4] Scibetta, M., Lucon, E., Chaouadi, R., Walle, E., and Gerard, R., “Use of Broken Charpy
V-notch Specimens From a Surveillance Program for Fracture Toughness Determi-
nation,” J. ASTM Int., Vol. 4, No. 2, 2006, JAI 12450.
[5] Miura, N. and Soneda, N., “Evaluation of Fracture Toughness by Master Curve Approach
Using Miniature C(T) Specimens,” Trans. ASME J. Pressure Vessel Technol., Vol. 134,
2010, 021402.
[6] Yamamoto, M., Kimura, A., Onizawa, K., Yoshimoto, K., Ogawa, T., Chiba, A., Hirano, T.,
Sugihara, T., Sugiyama, M., Miura, N., and Soneda, N., “A Round Robin Program of Master
Curve Evaluation Using Miniature C(T) Specimens—First Round Robin Test on Uniform
Specimens of Reactor Pressure Vessel Material,” Proceedings of the ASME 2012 Pressure
Vessel and Piping Division Conference, Toronto, ON, July 15–19, 2012.
[7] Yamamoto, M., Onizawa, K., Yoshimoto, K., Ogawa, T., Mabuchi, Y., and Miura, N., “A
Round Robin Program of Master Curve Evaluation using Miniature C(T) Specimens—2nd
Report: Fracture Toughness Comparison in Specified Loading Rate Condition,” Proceed-
ings of the ASME 2013 Pressure Vessel and Piping Division Conference, Paris, France,
July 14–18, 2013.
[8] JIS G 3204:1988/Amendment 1:2008: Quenched and Tempered Alloy Steel Forging for
Pressure Vessels (Amendment 1), Japanese Industrial Standards Committee, Tokyo,
Japan, 2008.
[9] Miura, N., Soneda, N., Arai, T., and Dohi, K., “Applicability of Master Curve Method to
Japanese Reactor Pressure Vessel Steels,” Proceedings of the 2006 ASME Pressure
Vessels and Piping Division Conference, Vancouver, BC, July 23–27, 2006.
[10] ASTM E1820-09e1: Standard Test Method for Measurement of Fracture Toughness,
ASTM International, West Conshohocken, PA, 2009, www.astm.org.
[11] Wallin, K., “Statistical Modeling of Fracture in the Ductile-to-Brittle Transition Region,”
Defect Assessment in Components—Fundamentals and Applications, ESIS/EFG9, J. G.
Blauel and K.-H. Schwalbe, Eds., Elsevier, Amsterdam, the Netherlands, 1991, pp.
415–445.
[12] Ruggeri, C., Dodds, R. H., and Wallin, K.,“Constraint Effects on Reference Temperature,
T0, for Ferritic Steels in the Transition Region,” Eng. Fract. Mech., Vol. 60, No. 1, 1998,
pp. 19–36.
[13] Welch, B. L., “The Generalization of ‘Student’s’ Problem When Several Different Popula-
tion Variances are Involved,” Biometrika, Vol. 34, Nos. 1–2, 1947, pp. 28–35.
[14] Matsumoto, M. and Nishimura, T., “Mersenne Twister: A 623-dimensionally Equidistrib-

uted Uniform Pseudorandom Number Generator,” ACM Trans. Model. Comput. Simul.,
Vol. 8, No. 1, 1998, pp. 3–30.
Xiang Chen,1 Randy K. Nanstad,1 and Mikhail A. Sokolov1
J-R Curve Determination for

Disk-Shaped Compact Specimens
Based on the Normalization
Method and the Direct Current
Potential Drop Technique
Reference
Chen, Xiang, Nanstad, Randy K., and Sokolov, Mikhail A., “J-R Curve Determination for
Disk-Shaped Compact Specimens Based on the Normalization Method and the Direct Current
Potential Drop Technique,” Small Specimen Test Techniques: 6th Volume, STP 1576, Mikhail A.
ABSTRACT
Material ductile fracture toughness can be described by J-integral versus crack
extension relationship (J-R curve). As a conventional J-R curve measurement
method, unloading compliance (UC) becomes impractical in elevated
temperature testing because of relaxation of the material and a friction induced
back-up shape of the J-R curve. In addition, the UC method may underpredict
the crack extension for standard disk-shaped compact (DC(T)) specimens. To
address these issues, the normalization method and the dc potential drop
(DCPD) technique were applied for determining J-R curves at 24 C and 500 C
for 0.18T DC(T) specimens made from type 316L stainless steel. For comparison
purpose, the UC method was also applied in 24 C tests. The normalization
method was able to yield valid J-R curves in all tests. The J-R curves from the
DCPD technique need adjustment to account for the potential drop induced by
plastic deformation, crack blunting, etc. and after applying a newly developed DCPD
adjustment procedure, the post-adjusted DCPD J-R curves essentially matched J-R
curves from the normalization method. In contrast, the UC method underpredicted
Manuscript received February 6, 2014; accepted for publication April 1, 2014; published online July 14, 2014.
1
Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, United
States of America.
2
Houston, TX.
Copyright V
CHEN ET AL., DOI 10.1520/STP157620140012 71
the crack extension in all tests resulting in substantial deviation in the derived J-R
curves manifested by high Jq values than the normalization or DCPD method. Only
for the tests where the UC method underpredicted the crack extension by a very
small value, J-R curves determined by the UC method were similar to those
determined by the normalization or DCPD method.
Keywords
J-R curve testing, normalization method, DCPD, DC(T), high temperature
testing, fracture toughness
Introduction
To improve thermal efficiency, next-generation nuclear reactors aim at operating in
more severe environments, such as elevated temperatures and higher stress levels,
than current reactors. Therefore, characterization of mechanical properties of struc-
tural materials for next-generation nuclear reactors in extreme environments
becomes vitally important from both engineering design and safety management
point of view [1]. Among different mechanical properties of materials, J-integral
verse resistance curve (J-R curve) is a useful tool for evaluating material structural
integrity in the presence of pre-existing defects. Furthermore, developing material
J-R curves with relatively small specimens, such as disk-shaped compact (DC(T))
specimens, gains more popularity nowadays, especially in the case of post-
irradiation mechanical tests, because of the limitation of irradiation volume and
difficulties associated with handling large irradiated specimens and their disposal
[2]. To date, extensive efforts have been continuously devoted to develop simplified
and reliable methods for determining material J-R curve. A widely accepted practice
for conducting J-R curve testing is ASTM E1820-11 [3], in which the unloading
compliance (UC) method is recommended for online crack-length measurement.
However, the UC method becomes impractical in elevated temperature testing
because of stress relaxation of the material and enhanced friction interference
between the specimen, pin, and clevis, which results in a back-up shape of the J-R
curve. In addition, the UC method may underpredict the crack extension for
DC(T) specimens, which further limits its applications for the J-R curve determina-
tion in small specimens.
To address the issue associated with the J-R curve determination with the UC
method, two alternative methods for deriving J-R curves, i.e., the normalization
method and the dc potential drop (DCPD) technique, are investigated in this study.
The normalization method was initially developed by Herrera [4] and Landes et al.
[5] and later studied by Joyce [6] and Lee [7]. In contrast to the UC method, the
normalization method does not rely on the compliance measurement for on-line
crack-size measurement. Instead, the normalization method solely needs a load-
displacement record taken together with initial and final crack-size measurements
from the specimen fracture surface to derive the material J-R curve. Because of the
elimination of the compliance measurement, the load-displacement curve in the
FIG. 1 Comparison of load-displacement records for the unloading compliance method

and the normalization method in a J-R curve test.
normalization method does not need to have the unloading–reloading portion as in

the UC method (Fig. 1), which significantly simplifies the test and reduces the test
time.
As another alternative J-R curve test method [8–12], the DCPD technique com-
bines advantages of both the UC and the normalization methods. It does not
require the unloading compliance measurement, so the test is simplified and the
load-displacement curve is the same as that of the normalization method in Fig. 1.
In addition, the DCPD technique provides experimental real-time crack-size meas-
urements as in the UC method. The crack-size prediction in DCPD relies on the
passage of a constant dc through the specimen and the subsequent measurement of
the voltage generated across an area in the specimen (Fig. 2(a)). As the crack propa-
gates in the specimen, less area is available for the passage of the same constant cur-
rent, resulting in increase of the effective electrical resistance and the potential
measurement, i.e., potential drop in Fig. 2(b). Thus, a correlation exists between the
crack length and potential drop in DCPD.
In this work, efforts are focusing on applying the normalization method and
the DCPD technique for determining J-R curves for 0.18T DC(T) specimens. Both
conventional room temperature tests and elevated temperature tests are performed.
For comparison purpose, the UC method is also applied to room temperature J-R
curve tests.
CHEN ET AL., DOI 10.1520/STP157620140012 73
FIG. 2 Schematic for (a) dc potential drop measurement; and (b) crack-growth-
induced increase in potential drop.
Experimental
MATERIAL AND SPECIMENS
Standard 0.18T DC(T) specimens (Fig. 3) with the thickness B of 4.62 mm and the
width W of 9.24 mm were machined from type 316L stainless steel in both T-L and
L-T orientations [13]. Table 1 lists the material constants at two selected testing tem-
peratures (24 C and 500 C). Each specimen was fatigue pre-cracked to a0 (initial
crack length)/W ¼ 0.5 and then side-grooved to remove 10 % of specimen thickness
from each side of the specimen.
TEST CONDITIONS AND DCPD PROBE SETUP

All tests were performed with a quasi-static loading rate such that dK/dt during the
initial elastic portion is 2 MPaHm/s. Two test temperatures, 24 C and 500 C, were
selected, representing conventional room temperature and elevated temperature J-R
curve tests. Three different J-R curve analysis methods, namely, the UC method,
the normalization method, and the DCPD technique, were applied in room temper-
ature testing, whereas only the normalization method and the DCPD technique
were used in 500 C testing because of inherent difficulties for the UC method in
elevated temperature testing.
A servo-hydraulic test frame with infrared heating was employed for fracture
toughness testing. The specimen load-line displacement was measured by a clip-on
displacement gauge. For the DCPD measurement, current probes and potential
probes were spot welded to the specimen as shown in Fig. 4. The current probes
were located midway between the back end of the specimen and the center line of
FIG. 3 Standard 0.18T disk-shaped compact specimen design.
pin holes. Two potential probes with 2.54-mm-gauge span were spot welded diago-
nally across the starter notch to average measurements of non-uniform crack fronts
if there are any [14]. For each test, load-displacement data and DCPD signal were
acquired from the same specimen simultaneously so that comparison of J-R curve
results between three different analysis methods is made on the same specimen to
avoid any influence because of specimen-to-specimen differences.
Results and Discussion

J-R CURVE DETERMINATION BY THE NORMALIZATION METHOD
The procedures for applying the normalization method to the J-R curve determina-
tion can be found in Annex 15 of ASTM E1820-11 [3]. Specifically, procedures for
TABLE 1 Material constants at testing temperatures.
Material Temperature ( C) Yield Strength (MPa) Tensile Strength (MPa) Young’s Modulus (GPa)
SS316L 24 423.5 810.1 193.1

500 131.0 413.7 165.0
CHEN ET AL., DOI 10.1520/STP157620140012 75
FIG. 4 Locations of current and potential probes for dc potential drop measurement.
applying the normalization method to DC(T) type specimens are elaborated in this
section. Because the J-integral calculation in the normalization method is same as
in the UC method, only the derivation of crack size in the normalization method is
presented in detail.
The starting point for the normalization analysis is the specimen load verse
load-line displacement record. From initial load up to, but not including the maxi-
mum load, each ith load Pi is normalized using:
P
(1) PNi ¼ i
W abi gpl
WB
W
where:
gpl ¼ a dimensionless parameter that relates plastic work done on a
specimen to crack-growth resistance defined in terms of deformation theory
J-integral [15] and ¼ 2 þ 0.522b0/W (here b0 is the initial uncracked ligament
and ¼ W a0), and
abi ¼ the blunting corrected crack size at the ith data point given by:
Ji
(2) abi ¼ a0 þ
2rY
where:
rY ¼ the effective yield strength and ¼ the average of the material yield strength
and the ultimate tensile strength at the testing temperature, and
the provisional J-integral at the ith data point, Ji, is calculated from:
Ki2 ð1 2 Þ
(3) Ji ¼ þ Jpli
E
where:
Ki ¼ the stress intensity at the ith data point, is Poisson’s ratio, E is Young’s
modulus of the material, and Jpli is the plastic component of Ji. The equation for
calculating Ki is:
Pi
Ki ¼
ðBBN WÞ1=2
a 4
ai a
i
a 2
i
a 3
i i
(4) 2þ 0:76 þ 4:8 11:58 þ11:43 4:08
W W W W W
ai 3=2
1þ
W
where:
BN ¼ the net specimen thickness, and
ai ¼ the crack size for the ith data point.
The equation for calculating Jpli is given by:

g Apli Apli1 ai ai1
(5) Jpli ¼ Jpli1 þ i1 1 ci1
bi1 BN bi1
where:
gi1 ¼ 2.0 þ 0.522bi1/W,
ci1 ¼ 1.0 þ 0.76bi1/W, bi1 ¼ the uncracked ligament for the (i 1)th data
point and ¼ W ai1, and
Jpli1 ¼ the plastic part of J-integral for the (i 1)th data point and assuming
Jpl0 ¼ 0 (initial plastic component of J-integral ¼ 0).
The quantity Apli Apli1 is the increment of plastic area under the load verse
load-line displacement record between lines of constant plastic displacement vpli
and vpli1 and can be calculated from the following equation:
ðPi þ Pi1 Þðvpli vpli1 Þ

(6) Apli Apli1 ¼
2
where vpli, the plastic part of the ith load-line displacement data point, is given by:
(7) vpli ¼ vi Pi CLLi
where:
vi ¼ the load-line displacement at the ith data point, and
CLLi ¼ the equivalent compliance corresponding to ai and is calculated by:
CHEN ET AL., DOI 10.1520/STP157620140012 77
a a 2 a 3
1 W þ ai 2 i i i
CLLi ¼ 2:0462 þ 9:6496 13:7346 þ6:1748
EBe W ai W W W
(8)
where the effective thickness Be ¼ B(BBN)2/B.

Similar to each normalized load PNi, the load-line displacement is normalized
0
to yield a normalized plastic displacement vpli using:
0 vpli vi Pi CLLi
(9) vpli ¼ ¼
W W
where CLLi is given in Eq 8.

From the first load-displacement pair up to, but not including, the load-
displacement pair at the maximum load, the normalized load PNi and normalized
0
plastic displacement vpli given in Eqs 1 and 9 are calculated using the initial crack
size a0. Namely, all the ai and ai1 values in Eqs 4, 5, and 8 should be substituted
with a0. Afterward, the last load-displacement pair is normalized using the same
form in Eqs 1 and 9, but with the final crack size af:
Pf
(10) PNf ¼
W af gpl
WB
W
0 vplf vf Pf CLLf
(11) vplf ¼ ¼
W W
where CLLf is calculated by replacing ai with af in Eq 8.

All of the normalized load and plastic displacement data points calculated in
previous procedures are plot as in Fig. 5(a). Then a tangent line is drawn from the
final data point in that plot to the remaining data as shown in Fig. 5(b). Data to the
right of the tangent point will be excluded from the normalization function fit. In
0
addition, data with normalized plastic displacement vpli 0.001 will also be
excluded from the normalization function fit. After applying the aforementioned
two eligibility criteria, qualified data are least square fitted with the following nor-
malization function (Fig. 5(c)):
0 02
a þ bvpl þ cvpl
(12) PN ¼ 0
d þ vpl
where:
a, b, c, and d ¼ fitting constants, and
0
PN and vpl ¼ the normalized load and plastic displacement given by the normal-
ization function.
If the normalization function can fit all the normalized load (PNi) and plastic
0
displacement (vpli ) pairs with a maximum deviation less than 1 % for the final data
FIG. 5 Normalization function fit procedures: (a) calculation of normalized load and

plastic displacement; (b) determination of eligible data for the normalization
function fit based on the tangent line drawn from the final load displacement pair
to the remaining data; and (c) starting from the normalized plastic displacement
of 0.001, application of the normalization function fit on eligible data from (b).
CHEN ET AL., DOI 10.1520/STP157620140012 79
point, the normalization method can be applied for deriving the J-R curve of the
test.
0
Next, an iterative procedure is adopted to make PNi and vpli match the normal-
ization function in Eq 12. This involves adjusting the crack size ai used to calculate
0 0
PNi and vpli so that the updated PNi and vpli would fall on the function in Eq 12. In
0
detail, for the first load-displacement pair with the original vpli 0.002 and assum-
ing ai ¼ a0, the normalized load is recalculated as:
Pi
(13) PNi ¼
W ai gpl
WB
W
0 0 0
and vpli is recalculated as in Eq. 9. The recalculated vpli is input as vpl in Eq 12 to
calculate PN. Then PN is compared with PNi in Eq 13. If the difference between PN
0
and PNi is larger than 60.1 %, the crack size ai is adjusted from a0 and PNi and vpli
0
is recalculated with the adjusted ai. Afterwards, the new vpli value is used to calcu-
late PN again using Eq 12, and the comparison between PN and PNi is performed for
a second time. This whole process is repeated until the difference between PN and
PNi is less than 60.1 % and at that condition the crack size ai is in its final form.
For each subsequent load-displacement pair,3 same treatment is applied with
the aiþ1 value inheriting the ai value from the previous step. Once the crack-size
adjustment process is complete, the finalized ai and load-displacement pairs with
0
the original vpli 0.002 are used to evaluate the J-integral using Eq 3. Combining
the J-integral value with ai yields the J-R curve for the normalization method.
J-R CURVE DETERMINATION BY THE DCPD TECHNIQUE

The DCPD technique derives experimental real-time crack sizes based on the
potential drop measurement across an area in the specimen. Similar to the normal-
ization method, the J-integral calculation in the DCPD technique is same as in the
UC method. Combining the J-integral calculation with crack sizes, the J-R curve
can be determined from the DCPD technique.
A number of constitutive equations exist for converting the potential drop mea-
surement into the crack size in the DCPD technique. Among those constitutive
equations, Johnson’s equation [16,17] has been widely used and is given by:
2W coshðpy=2WÞ
(14) a¼ cos1
p coshfðU=U0 Þ cosh1 ½coshðpy=2WÞ= cosðpa0 =2WÞg
where:
a ¼ the crack length corresponding to potential drop U,
y ¼ one half of the potential gauge span (i.e., 0.127 mm per Fig. 4(b)), and
a0 and U0 ¼ initial crack length and potential drop, respectively.
3
The subsequent load-displacement pairs include all the load-displacement data after the first load-
0
displacement pair with the original vpli 0.002.
FIG. 6 Original J-R curve determined with the DCPD technique.
Based on the crack-size calculation in Eq 14 and the corresponding J-integral

calculation in Eq 3, the original DCPD J-R curve is determined with one example
shown in Fig. 6.4
The initial portion of the J-R curve in Fig. 6 indicates fast crack growth and
does not follow the construction line closely, resulting in a relatively low Jq value.
Indeed, as noted in the work of Bakker [18], during the J-R curve test material, a
potential drop can also result from deformation, crack blunting, and void growth in
the process zone ahead of the crack. If the influences of these factors are not
accounted in the crack-length prediction, the DCPD technique would not predict
the crack size accurately and can result in much lower Jq values. Therefore adjust-
ment on original DCPD J-R curve is needed. In early DCPD adjustment methods
[19], a slope change point, counted as the critical point distinguishing crack blunt-
ing from the onset of slow stable crack growth, is identified by visual inspection of
the displacement potential drop curve. However, the slope change point in displace-
ment potential drop curve of a test may not be clearly identified on occasion and
the selection of the critical point tends to be arbitrary, so the repeatability of the
analysis is poor. In a more recent work by Chen et al. [20], new DCPD adjustment
methods were developed with improved repeatability and excellent match with the
UC and normalization methods in terms of Jq values. The major drawback in this
method is that the post-adjustment DCPD tearing modulus results show an average
difference of 17 % from the UC and normalization results. A possible explanation
for that result is that because the potential gauge span in that work was relatively
large, the potential drop signal was measured from a large volume of material,
which enhanced the influence of plastic deformation on the DCPD measurement.
4
Construction lines in Fig. 6 and subsequent figures containing J-R curves are given by J ¼ 2rYDa.
CHEN ET AL., DOI 10.1520/STP157620140012 81
FIG. 7 (a) Critical point selection based on the peak point of the first-order derivative
of the original DCPD J-R curve in Fig. 6 with Savitzky-Golay smoothing; and (b)
adjustment of the original DCPD J-R curve in Fig. 6.
To address these issues, shorter potential gauge span is adopted in this study.
In addition, a newly developed semi-empirical DCPD adjustment procedure is
proposed. The DCPD adjustment procedure is mainly composed of two steps. The
first step aims for identifying the critical point distinguishing crack blunting from
the onset of slow stable crack growth. To achieve this, a first-order derivative of the
original DCPD J-R curve coupled with Savitzky-Golay [21] second-order polyno-
mial smoothing with 19 points of window is applied as shown in Fig. 7(a). The peak
point in Fig. 7(a) represents the data point from which the slope of the original
DCPD J-R curve starts to decrease and is indicative of the onset of slow stable crack
growth. Therefore, the peak point is selected as the critical point. The combination
of first-order derivative of the original DCPD J-R curve and Savitzky-Golay
smoothing eliminates the ambiguity in the selection of the critical point and greatly
suppress the noise, if there is any, in the original DCPD J-R curve data for the
critical point selection. Once the critical point is identified, all data points prior to
and including the critical point itself from the original DCPD J-R curve are shifted
left onto the construction line because these data points correspond to the plastic
deformation and crack blunting stage of the specimen (Fig. 7(b)).
After crack initiation, stable crack growth dominates the increase in the poten-
tial drop although material deformation and specimen shape change may still influ-
ence the potential drop measurement. To address the influence of those factors on
DCPD-predicted crack length at the stable crack-growth stage, the second step of
the DCPD adjustment procedure covers data points between the crack initiation
point and last point in a J-R curve. The concept of the second step adjustment is to
force the final crack-length prediction from DCPD to match the optically measured
final crack extension so that the final crack-length prediction from DCPD is correct.
Assuming the crack size at crack initiation is correct after the first step adjustment
in DCPD, data points between the crack initiation point and final crack-length
measurement in a post-adjustment J-R curve should be reasonably valid. To realize
this concept, crack sizes for data points subsequent to the critical point in the origi-
nal DCPD J-R curve are adjusted such that the final crack extension prediction
from the DCPD technique matches the measured final crack extension (Fig. 7(b)).
To do so, the crack size of the ith data point subsequent to the critical point in the
original DCPD J-R curve is adjusted according to the following equation [18]:
apdi0 ¼ Dapdcritical0 þ Dapdi Dapdcritical
(15) updi updcritical
ðDapdfinal Dapdcritical Da0 Þ
updfinal updcritical
where:
Dapdcritical0 ¼ the crack extension of the critical point after the first step
adjustment,
Dapdi ¼ the original DCPD crack extension for the ith data point,
Dapdcritical is the original crack extension of the critical point,
updi is the displacement record for the ith data point,
updcritical is the displacement record for the critical point,
updfinal is the displacement record for the last data point in the J-R curve,
Dapdfinal is the original DCPD crack extension for the last data point in the J-R
curve, and
Da0 is defined by:
CHEN ET AL., DOI 10.1520/STP157620140012 83
TABLE 2 Comparison of J-R curve results among the UC method, the normalization method, and
the DCPD technique after adjustment.
Jq (kJ/m2) J-R Curve Slopeb
Specimen Temperature Adjusted Adjusted

IDa ( C) UC Normalization DCPD UC Normalization DCPD
1-TL 24 304.3 224.4 234.4 163.9 189.1 169.8

2-TL 24 330.7 256.2 262.5 97.0 118.0 110.5
3-TL 24 303.2 253.3 262.1 124.1 144.3 122.1
4-LT 24 430.9 419.0 415.1 232.8 241.9 244.2
5-LT 24 440.2 445.9 421.1 187.0 188.0 204.7
6-TL 500 NA 189.6 197.9 NA 108.3 98.1
7-TL 500 NA 182.0 190.8 NA 116.0 103.7
8-LT 500 NA 285.5 289.9 NA 179.7 161.9
9-LT 500 NA 279.0 286.0 NA 184.1 158.1
a
TL and LT designate the orientation of the specimen.
b
J-R curve slope is determined by linear fitting of the J-R curve portion between the first and second
exclusion lines.
(16) Da0 ¼ Dameasured Dapdcritical0
where:
Dameasured ¼ the optically measured final crack extension.
Once the updated crack sizes from the DCPD adjustment procedure become
available, they are used to recalculate J-integral values.
COMPARISON OF J-R CURVES DETERMINED BY THE UC METHOD, THE

NORMALIZATION METHOD, AND THE DCPD TECHNIQUE AFTER ADJUSTMENT
Major J-R curve test results, namely, Jq value and the J-R curve slope for tearing
modulus determination, are summarized in Table 2. Excellent agreement is observed
between J-R curves from the normalization method and adjusted DCPD J-R curves
in all tests. The average differences for Jq and J-R curve slopes between the normal-
ization method and the DCPD technique after adjustment are only about 3.3 % and
9.5 %, respectively. Indeed, J-R curves from the normalization method and the
DCPD technique after adjustment essentially match each other as shown in Fig. 8(a)
and 8(b). For J-R curve results based on the UC method, Table 2 reveals substantial
deviation in Jq values between the UC method and the normalization or DCPD
method, except for specimen 4-LT and 5-LT. Figure 8(a) further confirms the appa-
rent difference in J-R curves between the UC method and the normalization or
DCPD method. A closer look of J-R curve results from the UC method indicates
underprediction of the crack extension by the UC method as shown in Fig. 9.
Because the specimen design used in this study incorporated an “outboard” clip
gauge, the load line displacement was measured far from the crack plane. This may
FIG. 8 Comparison of J-R curves from the unloading compliance method, the
normalization method, and the DCPD technique after adjustment. (a) and (b)
represent two cases showing that the unloading compliance method
underpredicted the final crack extension by a large amount and a small amount,
respectively.
affect the accuracy and sensitivity of the compliance measurement in the UC

method and likely is the cause of the underestimation of the final crack length [22].
This underprediction of the crack extension would result in high J-integral value in
CHEN ET AL., DOI 10.1520/STP157620140012 85
FIG. 9 Comparison of measured final crack extension with the unloading compliance
method predicted final crack extension.
the J-integral calculation if load and displacement data are same. In addition, this
also shifts the J-R curve to the left and further increases the Jq value. Only if the UC
method underpredicts the crack extension by a very small value, which is the case
for specimen 4-LT and 5-LT as shown in Fig. 9, the J-R curve from the UC method
can be similar to those from the normalization or DCPD method as in Fig. 8(b).
Conclusions
J-R curve testing was performed at 24 C and 500 C on standard 0.18T DC(T)
specimens made from type 316L stainless steel in both T-L and L-T orientations.
To address the issue associated with the J-R curve determination with the UC
method, especially at elevated temperatures and small specimen testing, alternative
J-R curve analysis methods, namely, the normalization method and the DCPD
technique, were investigated in this study. The normalization method was able to
yield valid J-R curves in all tests. The J-R curves from the DCPD technique need
adjustment to account for the potential drop induced by plastic deformation, crack
blunting, etc. and after applying a newly developed DCPD adjustment procedure,
the post-adjusted DCPD J-R curves essentially matched J-R curves from the nor-
malization method. In contrast, the UC method underpredicted the crack extension
in all tests resulting in substantial deviation in the derived J-R curves manifested by
higher Jq values than the normalization or DCPD method. Only for tests where the
UC method underpredicted the crack extension by a very small value, J-R curves
determined by the UC method were similar to those determined by the normaliza-
tion or DCPD method.
ACKNOWLEDGMENTS
The writers extend their appreciation to Eric Manneschmidt for performing part of
mechanical testing, and to Dr. Lizhen Tan for technical review. This manuscript was
authored by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the
U.S. Dept. of Energy. The publisher, by accepting the article for publication, acknowl-
edges that the United States Government retains a non-exclusive, paid-up, irrevocable,
world-wide license to publish or reproduce the published form of this manuscript, or
allow others to do so, for U.S. Government purposes.
References
[1] Zhu, X. K., Lam, P. S., and Chao, Y. J., “Application of Normalization Method to Fracture
Resistance Testing for Storage Tank A285 Carbon Steel,” Int. J. Pres. Ves. Pip., Vol. 86,
No. 10, 2009, pp. 669–676.
[2] Kim, S. W., Tanigawa, H., Hirose, T., and Kohyama, A., “Effects of Surface
Morphology and Distributed Inclusions on the Low Cycle Fatigue Behavior of
Miniaturized Specimens of F82H Steel,” Small Specimen Test Techniques: 5th Volume,
ASTM STP 1502, M. Sokolov, Ed., ASTM International, West Conshohocken, PA, 2009,
p. 159.
[3] ASTM E1820-11: Standard Test Method for Measurement of Fracture Toughness, ASTM
International, West Conshohocken, PA, 2011, www.astm.org.
[4] Herrera, R. and Landes, J. D., “Direct J-R Curve Analysis: A Guide to the
Methodology,” Fracture Mechanics: Twenty-First Symposium, ASTM STP 1074, J.
Gudas, J. Joyce, and E. Hackett, Eds., ASTM International, West Conshohocken, PA,
1990, p. 24.
[5] Landes, J., Zhou, Z., Lee, K., and Herrera, R., “Normalization Method for Developing J-R
Curves with the Lmn Function,” J. Test. Eval., Vol. 19, No. 4, 1991, pp. 305–311.
[6] Joyce, J., “Analysis of a High Rate Round Robin Based on Proposed Annexes to ASTM
E1820,” J. Test. Eval., Vol. 29, No. 4, 2001, pp. 329–351.
[7] Lee, K., 1995, “Elastic-Plastic Fracture Toughness Determination Under Some Difficult
Conditions,” Ph.D. dissertation, University of Tennessee, Knoxville.
[8] Hicks, M. A. and Pickard, A. C., “A Comparison of Theoretical and Experimental Methods
of Calibrating the Electrical Potential Drop Technique for Crack Length Determination,”
Int. J. Fract., Vol. 20, No. 2, 1982, pp. 91–101.
[9] Lowes, J. M. and Featnehough, G. D., “The Detection of Slow Crack Growth in Crack
Opening Displacement Specimens Using an Electrical Potential Method,” Eng. Fract.
Mech., Vol. 3, No. 2, 1971, pp. 103–104.
[10] McGowan, J. J. and Nanstad, R. K., “Direct Comparison of Unloading Compliance and
Potential Drop Techniques in J-Integral Testing,” SEM Fall Conference, Computer-Aided
Testing and Modal Analysis, Milwaukee, WI, Nov 4–7, 1984.
CHEN ET AL., DOI 10.1520/STP157620140012 87
[11] Ritchie, R. O. and Bathe, K. J., “On the Calibration of the Electrical Potential Technique
for Monitoring Crack Growth Using Finite Element Methods,” Int. J. Fract., Vol. 15, No. 1,
1979, pp. 47–55.
[12] Tong, J., “Notes on Direct Current Potential Drop Calibration for Crack Growth in Com-
pact Tension Specimens,” J. Test. Eval., Vol. 29, No. 4, 2001, pp. 402–406.
[13] ASTM E1823-13: Standard Terminology Relating to Fatigue and Fracture Testing, ASTM
[14] ASTM E647-11: Standard Test Method for Measurement of Fatigue Crack Growth Rates,
[15] Turner, C. E., “The Eta Factor,” Post Yield Fracture Mechanics, 2nd ed., Elsevier Science,
New York, Amsterdam, 1984, p. 451.
[16] Johnson, H. H., “Calibrating the Electrical Potential Method for Studying Slow Crack
Growth,” Mater. Res. Stand., Vol. 5, 1965, pp. 442–445.
[17] Schwalbe, K. H. and Hellmann, D., “Application of the Electrical Potential Method to
Crack Length Measurements Using Johnson Formula,” J. Test. Eval., Vol. 9, No. 3, 1981,
pp. 218–220.
[18] Bakker, A. D., “A DC Potential Drop Procedure for Crack Initiation and R-Curve Measure-
ments During Ductile Fracture Tests,” Elastic-Plastic Fracture Test Methods: The User’s
Experience, ASTM STP 856, F. Loss, and E. Wessel, Eds., ASTM International, West Con-
shohocken, PA, 1985, p. 394.
[19] Dufresne, J., Henry, B., and Larsson, H., “Fracture Toughness of Irradiated AISI 304 and
316L Stainless Steels,” Effects of Radiation on Structural Materials, ASTM STP 683, J.
Sprague and D. Kramer, Eds., ASTM International, West Conshohocken, PA, 1979, p. 511.
[20] Chen, X., Nanstad, R. K., and Sokolov, M. A., “Application of Direct Current Potential
Drop for Fracture Toughness Measurement,” 22nd International Conference on Struc-
tural Mechanics in Reactor Technology, San Francisco, CA, Aug 18–23, 2013.
[21] Savitzky, A. and Golay, M., “Smoothing and Differentiation of Data by Simplified Least
Squares Procedures,” Anal. Chem., Vol. 36, No. 8, 1964, pp. 1627–1639.
[22] Marsh, K. J., Smith, R. A., and Ritchie, R. O., Fatigue Crack Measurement: Techniques and
Applications, Engineering Materials Advisory Services, Cradley Heath, U.K., 1991, p. 69.
J. May,1 J. Rouden,2 P. Efsing,2 M. Valo,3 and H. Hein1
Extended Mechanical Testing of

RPV Surveillance Materials Using
Reconstitution Technique for
Small Sized Specimen to Assist
Long Term Operation
Reference
May, J., Rouden, J., Efsing, P., Valo, M., and Hein, H., “Extended Mechanical Testing of RPV
Surveillance Materials Using Reconstitution Technique for Small Sized Specimen to Assist
Long Term Operation,” Small Specimen Test Techniques: 6th Volume, STP 1576, Mikhail A.
ABSTRACT
For the Ringhals 3 and 4 pressurized water reactors (PWR) reactor pressure
vessels (RPV), results from the irradiation surveillance program are available also
for neutron fluences, which cover long-term operation (LTO). These standard
surveillance results are based on the RTNDT concept. The belt-line welds of both
RPVs have an elevated nickel-content of 1.6 wt. % and, as a consequence,
irradiation response is higher than predicted by model equations. Therefore, the
mechanical testing program has been expanded, exceeding the requirements of
the standard testing program and covering both base and weld materials. To
improve the understanding of the material behavior, extended Master Curve
testing was performed on PCCV and subsize SE(B) specimens from irradiation
surveillance capsules with the help of specimen reconstitution technique. Special
care has been taken on the limited amount of weld material within the available
broken Charpy halves before specimen reconstitution. Results have been
Manuscript received December 16, 2013; accepted for publication March 18, 2014; published online July 14,
2014.
1
AREVA GmbH, D-91052 Erlangen, Paul-Gossen-Str. 100, Germany.
2
Vattenfall Ringhals AB, SE-432 85 Väröbacka, Sweden.
3
VTT Technical Research Centre of Finland, P.O. Box 1000, FI-02044 VTT, Finland.
4
ASTM Sixth International Symposium on Small Specimen Test Techniques on January 29–31, 2014, in
Houston, TX.
Copyright V
MAY ET AL., DOI 10.1520/STP157620130189 89
compared to existing data on similar base and weld materials from the German
research programs CARISMA and CARINA. Late-blooming effects or sudden
saturation effects are not observed for base or weld materials under LTO
conditions. The data for the four different weld materials of similar chemical
composition indicate that not only the chemical composition, but also other
influencing factors like, e.g., the welding heat treatment, may be important for
the reference temperature of the unirradiated state as well as for the irradiation
behavior. To investigate this effect in more detail, a future investigation program
will be discussed including manufacturing of a surrogate weld material with the
same chemical composition as in Ringhals 3 and 4 RPV. The influence of heat
treatment condition can be investigated by applying different heat treatments
and subsequently performing test reactor irradiation and mechanical testing.
Specimen reconstitution will be required due to limited space inside the test
reactor irradiation capsules.
Keywords
reactor pressure vessel, irradiation embrittlement, fracture toughness, master
curve, material testing, small sized specimens, reconstitution technique
Introduction
The standard irradiation surveillance programs for the reactor pressure vessels
(RPV) of western type pressurized water reactors (PWR) which are currently in
operation are usually based on the RTNDT concept, where the neutron irradiation
induced shift in transition temperature of Charpy tests is used to adjust the refer-
ence temperature. Therefore, the surveillance capsules which are usually located
within the RPV contain, among others, Charpy specimens from the RPV base
material (BM), weld material (WM) and sometimes also from the heat affected
zone (HAZ). As most of the existing PWR have been commissioned in the 1970s or
1980s, most of the surveillance capsules have already been evaluated. In many cases,
the standard irradiation surveillance program is able to cover also the neutron flu-
ence levels beyond the original design lifetime of 40 years because neutron flux
reduction measures like low-leakage core loading have been applied after some
years of operation. Most of the PWRs have at least one surveillance capsule that
covers a long-term operation (LTO) of 60 years, and some of the surveillance pro-
grams can even cover 80 or more years.
In the 1990s, when a large part of the surveillance capsules had already been
evaluated, a new reference temperature concept was established. The T0 concept
[1], also known as the Master Curve concept, has the advantage of a direct measure-
ment of fracture toughness instead of the indirect correlation between the Pellini/
Charpy tests and fracture toughness, as it is done in the RTNDT concept. With the
reference temperature RTT0 ¼ T0 þ 19.4 K [2] an alternate reference temperature
was defined, which can be used alternatively to RTNDT. As most of the RPV surveil-
lance programs were designed before this alternate concept was established, the
surveillance capsules of most PWRs did not contain specimens that were specifi-
cally dedicated to Master Curve testing. However, after standard Charpy testing, the
broken Charpy specimens can be used to manufacture new precracked Charpy-V
(PCCV) or subsize SE(B) specimens with the help of the specimen reconstitution
technique. In this way, Master Curve testing can be performed to obtain the refer-
ence temperature RTT0 in the irradiated material condition.
In the present investigations, the specimen reconstitution technique was used
to perform extended Master Curve testing on BM and WM of several irradiation
sets of the RPV surveillance programs of PWR Ringhals 3 (R3) and Ringhals 4
(R4), located in Sweden. Data from test reactor irradiation and from the German
research programs CARISMA [3] and CARINA [4] have been used for
comparison. Special attention has been paid for the high nickel weld materials,
which are known to have an irradiation response that is higher than predicted by
model equations [5].
Ringhals 3 and 4 RPV Materials

The Ringhals Units 3 and 4 nuclear reactors were designed as standard Westing-
house 900 MWe 3-loop pressurized water reactors. Unit 3 was commissioned
in 1981, Unit 4 in 1983. The reactor pressure vessels were manufactured by the
Swedish welding company Uddcomb, using SA 508 Class 2 ring forgings supplied
by Kloeckner company in accordance to technical specifications from Westing-
house. The chemical analysis of the forgings and the weld metal used in the core
region of the RPV of Unit 3 and 4 is shown in Table 1. As seen in the table, the nickel
content of the circumferential weld is high: 1.58 wt. % in Unit 3 and 1.66 wt. % in
Unit 4.
The details of the irradiation surveillance programs of Ringhals Unit 3 and 4
are described in [5]. The main results are summarized in Fig. 1 for BM and Fig. 2 for
WM. In these diagrams, in comparison to earlier results from Ref. [5], the neutron
fluence values have been recalculated and high fluence data points have been added.
TABLE 1 Chemical composition of the RPV materials in the core region for Ringhals Unit 3 and 4.
Chemical Composition (wt. %)
Material C Si Mn P S Cr Mo Ni Cu
R3 BM lower shell 0.19 0.19 0.74 0.006 0.006 0.42 0.60 0.90 0.10
R3 BM inter. shell 0.18 0.19 0.76 0.006 0.006 0.43 0.54 0.88 0.08
R3 WM 0.052 0.21 1.46 0.009 0.006 0.07 0.54 1.58 0.08
R4 BM lower shell 0.21 0.18 0.75 0.006 0.008 0.43 0.63 0.88 0.10
R4 BM inter. shell 0.21 0.20 0.71 0.006 0.007 0.45 0.63 0.88 0.09
R4 WM 0.068 0.14 1.35 0.015 0.004 0.04 0.50 1.66 0.05
MAY ET AL., DOI 10.1520/STP157620130189 91
FIG. 1 Surveillance data for R3 and R4 base materials and corresponding predictions
from Reg. Guide 1.99 Rev. 2 Pos. 1.
For all four base materials, at elevated neutron fluence values, the shift in
Charpy energy transition temperature DT41 exceeds the prediction of the Regula-
tory Guide 1.99 Rev. 2 model equation. In contrast to the evaluation in Ref. [5], the
chemistry factor of the R.G. model has not been fitted to the data points; it has
instead been calculated with the Cu and Ni contents from Table 1. One of the rea-
sons for the underprediction is that the Reg. Guide model equation is based on a
data base where only a few data points are in a regime of relatively high neutron
FIG. 2 Surveillance data for R3 and R4 weld materials and corresponding predictions
from Reg. Guide 1.99 Rev. 2 Pos. 1 and from JEAC4201-2007 with recalibrated
model parameters [8].
fluence [6], so that the high fluence data of R3 and R4 BM are not well represented
by the model. Advanced prediction equations which are based on larger data bases,
like, e.g., the wide range WR-C (5) model [7], have a better prediction capability for
the R3 and R4 BM [5].
The results from the surveillance testing of the R3 and R4 RPV belt-line welds in
Fig. 2 show that irradiation induced transition temperature shifts of more than 200 K
are observed at neutron fluences above 5 1019 n/cm2 (E > 1 MeV). These shifts are
significantly higher than the prediction of the R.G. 1.99 and can be attributed to the rel-
atively high Ni-content of the weld materials, which leads to the formation of Ni-Mn-
Si rich phases [9]. This high Ni-content is out of the applicability range of the R.G. 1.99
prediction equation; however, from the table of chemistry factors of R.G. 1.99, it can be
concluded that at the time of model development, it was assumed that the irradiation
sensitivity of low-Cu RPV steels with respect to the Ni-content would saturate at Ni-
contents above 1.0 wt. %. Even advanced prediction equations are known to underesti-
mate the irradiation response of high nickel RPV materials [5,10]. An evaluation of cur-
rently available models has shown that the Japanese JEAC4201-2007 with recalibrated
model parameters from PVP2013 [8] has the best predictive capability for these high
nickel weld materials among all currently available models that are representative for
western RPV steels. This prediction is also shown in Fig. 2 for comparison. Especially at
high fluences, the observed shift in transition temperature of the high Ni WM is higher
than predicted. It should be mentioned, however, that a very low reference temperature
in the unirradiated state of these high Ni materials to some extend compensates the ele-
vated irradiation response [10]. As an example, for R3 and R4 WM, the initial reference
temperature is RTNDT ¼ –75 C [5].
Master Curve Testing Program

The above mentioned results from the surveillance program are based on the
RTNDT-concept and therefore represent an indirect determination of fracture
toughness. Meanwhile, the Master Curve concept [1], which allows direct determi-
nation of fracture toughness data, was well established. In order to improve the
understanding of the behavior of R3 and R4 RPV materials, Master Curve testing
was performed. Because the surveillance specimens were already tested in the frame
of the standard surveillance program, specimen reconstitution technique was
applied in Master Curve testing by preparing new specimens from broken halves of
Charpy-V specimens. Two different laboratories performed the testing, using differ-
ent specimen types and reconstitution techniques. In all cases, the original specimen
orientation (T–L) was maintained in reconstitution.
The testing program comprised the lower and intermediate shells (L.S., I.S.) as
well as the belt-line welds of R3 and R4 RPV. The unirradiated and irradiated material
states from several different surveillance sets were used. However, not all materials
were available in every irradiation condition. Additionally, a similar weld material like
the one used for the R3 and R4 beltline weld was obtained from a replaced RPV head
MAY ET AL., DOI 10.1520/STP157620130189 93
and it was subsequently irradiated in the Halden high flux test reactor to several neu-
tron fluences. This weld material has also been manufactured by Uddcomb company
and has a Cu-content of 0.12 wt. % and a Ni-content of 1.56 wt. %. The weld speci-
mens prepared from the replaced RPV head weld were irradiated as full Charpy-size
bars, they were thereafter split into half Charpy size bars by electric discharge machin-
ing (EDM). The length of weld in the bars (35 mm) allowed performing three frac-
ture toughness tests with each bar. The first test of each split bar was a full specimen
test the crack being located in the center of the weld and, thereafter, two additional
tests were performed with reconstituted specimens prepared from the broken halves
of the first fracture toughness specimen. In the reconstituted specimens, the distance
between the reconstitution weld and specimen crack was 8 mm, i.e., equivalent full
insert length was 16 mm.
Details about the test matrix (neutron fluences, irradiation facilities, specimen
type, and testing laboratory for the individual specimen sets) can be obtained from
the summary of experimental results in Table 2.
The VTT laboratory performed testing on side-grooved (2*10 %) single-edge
notched bend specimens (SE(B)) of nominal dimensions B ¼ 5 mm (thickness),
W ¼ 10 mm (width), and L ¼ 55 mm (length). Due to application of the EDM-
technique, actual specimen thickness was B ¼ 4.85 mm. The applicability of this rather
small specimen size has been validated in Ref. [11]. The number of new reconstituted
specimens for surveillance material prepared from tested specimens varied from two to
four depending on the amount of weld in the broken specimens. Notice that splitting
of the specimen doubles the number of new specimens as compared to full Charpy size
specimen. Material from tested half inch CT specimens was also utilized.
The AREVA laboratory performed testing on side-grooved single-edge notched
bend specimens (SE(B)) of dimensions B ¼ 10 mm, W ¼ 10 mm, and L ¼ 55 mm,
which were reconstituted by electron beam welding. This specimen geometry is well
established and nearly identical to the PCCV geometry (pre-cracked Charpy-V
notch, also known as PCVN), except that the V-notch is replaced by a spark erosion
wire cut. Typically, two reconstituted specimens were manufactured from one bro-
ken Charpy specimen. The insert length was in most cases 15 mm for the RPV weld
materials and 18 mm for the RPV base materials. For the unirradiated state of the
Ringhals 4 base materials, additionally some bulk (non-reconstituted) specimens
were available for testing. Results from testing of these specimens were in accord-
ance with the results from the experiments on the reconstituted specimens, con-
firming the applicability of the reconstitution technique.
Validation of Specimen Reconstitution

Technique
A critical issue for specimen reconstitution from broken specimens is the heat input
of the applied welding technique. Especially for irradiated materials, it is important
that during welding the temperature in the area of the subsequent crack
propagation and the surrounding plastic deformation zone does not exceed the irra-
diation temperature. Otherwise, microstructure and mechanical properties could be
influenced. It is also important to consider the influence of the plastic zone size as
well as the residual stresses introduced by the reconstitution welding.
The electron beam welding reconstitution process of the AREVA laboratory was
validated in the frame of the RESQUE project [12]. Meanwhile, a new electron beam
welding equipment was established in the laboratory, allowing even more narrow weld-
ing gap and heat affected zones. For this new equipment, part of the validation work
has been repeated. As an example, Fig. 3 shows the measurement of temperatures for
the new welding process at distinct positions within the insert (insert size 15 mm). The
measurements represent the second welding pass, which shows higher temperatures in
comparison to the first welding pass due to remaining heat from the first pass.
A maximum temperature of 191 C has been measured at 5 mm distance from
the welding seam. After a few seconds, temperature significantly decreased. In the
middle of the insert, which is the most relevant position for subsequent material
testing, the maximum temperature was 140 C. These temperatures are well below
the irradiation temperature of the surveillance capsules inside a PWR RPV, which
is typically around 285 C – 290 C.
VTT was used for the specimen reconstitution stud arc welding technique
before 2010 and thereafter electron beam welding. Ringhals surveillance data was
created partly with stud arc technique and partly with electron beam technique. All
RPV head data have been created with electron beam welding technique.
The main validation of the reconstitution welding technique relies on heat
transient temperature measurements. The measured maximum temperature as a
FIG. 3 Time dependent temperature profiles at four different positions within an 15 mm

insert during welding (second welding seam) for the reconstitution of a SE(B)
10 mm by 10 mm by 55 mm specimen; the vertical dashed line at 50 s represents
the beginning of welding.
MAY ET AL., DOI 10.1520/STP157620130189 95
FIG. 4 Maximum temperature at specimen center line as a function of distance from

weld surface in stud arc welding.
function of distance from the weld surface for stud arc welding is shown in Fig. 4.
For the measurements, 0.5 mm outer diameter nickel tube sheeted, type K thermo-
couples were used. The thermocouple joint was isolated from the sheet and the
exact location of the joint was determined with X-ray photographs for each thermo-
couple. Typically, the joint was located 0.3–0.5 mm from the tip of the tube. The
thermocouple was silver soldered in a tight hole in a mock up specimen. The depth
of the hole was exactly known and correct location of the thermocouple at the bot-
tom of the hole was observed by the downwards movement of the thermocouple at
the moment, when a small piece of silver located in the hole melted. The nominal
response time of the thermocouple given by the manufacturer is 0.1 s. The nearest
distance of the crack location from weld surface was 6 mm, where the measured
maximum temperature is 150 C.
In electron beam welding, low heat input was aimed on purpose. First low
energy tag welds were applied. Thereafter, both side faces of the 10 mm by 5 mm
cross-sectional fracture toughness specimens were welded up to weld depth of
2.5 mm. One side weld seam was in addition welded in two runs, i.e., first from one
corner to middle of the face, whereafter the beam current was reduced gradually to
zero. After first weld, the system was let to cool to room temperature and the sec-
ond run was realized from the other end of the seam. Because the electron beam
weld has conical shape, it is not self-evident, where the maximum temperature
occurs. Temperatures were measured on the center line (2.5 mm depth) and near
the surface (1.25 mm depth). The measured data is shown in Fig. 5.
The RPV head testing created a large database for the validation of the applied
reconstitution technique because both full specimens (1/3) and reconstituted
FIG. 5 The peak temperature versus distance measured from the surface to be welded
for 5.00 mm by 10.00 mm cross-sectional specimens. The specimen protrudes
1.1 mm out from the cooling clamps, which point is shown as a vertical line in the
figure. Temperature rise is measured relative to room temperature. The heat
input 275 J is the max heat input applied to the specimen at one time. Welding is
performed in four steps.
specimens (2/3) were included in each transition curve. Comparison of T0-values of

the two sub-sets, i.e., full and reconstituted specimens, does not show any difference
in the measured T0-values. The original weld portion was approximately 35 mm
long and, hence, the weld length in the reconstituted specimens was approximately
16 mm long. Twelve specimens were tested for each Master Curve, which does
not allow comparing the effect of reconstitution on the measured T0-values. Hence
the non-reconstituted specimen data was joined together as well as the reconsti-
tuted data. Because yield stress varies between the data sets, censoring was per-
formed for each data set separately, i.e., the end of test cleavage and ductile (few)
test values were lowered to the m ¼ 30 line, if this line was exceeded, and this data
was treated as a ductile end of test data as described in the Master Curve standard.
The relative temperature coordinate T–T0 is used as temperature in the joined anal-
ysis. Figure 6 shows the Master Curves determined for all data, all reconstituted data
and all non-reconstituted data.
Experimental Results
Before Master Curve testing of the RPV weld materials with the SE(B) 10 mm by
10 mm by 55 specimen size, it was confirmed that a sufficient amount of RPV weld
material is available in the broken Charpy-V halves for preparing a 15 mm insert
MAY ET AL., DOI 10.1520/STP157620130189 97
FIG. 6 Joined pressure RPV head weld data for the comparison of the effect of
reconstitution on the measured KJC- and T0-values. The indicated data is size
corrected to B ¼ 25 mm and the bold data points indicate censored data, which
was lowered to the validity limit m ¼ 30 of each individual data set. Most of the
censored data is cleavage initiation data. The 2 x 10 % size grooved specimen of
dimensions W ¼ 10 mm, B ¼ 4.85 mm were used and 1/3 of the data is measured
with non-reconstituted specimens and 2/3 with reconstituted specimens. T0 of
all data is zero due to normalization of the temperature scale. The deviation of
T0 from zero is 1 C in case of reconstituted specimens and 2 C in case of non-
reconstituted specimens, which deviations are negligible compared to the max
measured T0 shifts of approximately 200 C.
for reconstitution. This has been performed by polishing and etching—see Fig. 7 as
an example. Only for one of the tested specimens, a reduced insert size had to be
used because of insufficient amount of available RPV weld material.
Figure 8 summarizes the Master Curve test results of all experiments of the
SE(B) 10 mm by 10 mm by 55 specimen size. Testing was performed according to
ASTM E1921-11 a [13]. In total, 99 valid fracture toughness values and 23 censored
values have been generated to obtain the 14 T0 values. In the diagram, where the
valid fracture toughness data points are plotted versus T–RTT0, it can be seen that
the general scattering of data points is reasonably low. However, a few data points,
mainly in the lower shelf regime, are not covered by the ASME KIC curve if the
FIG. 7 Determination of available RPV weld material inside broken Charpy-V specimen
halves by polishing and etching.
ASME KIC curve is indexed by RTT0. One of these data points is from weld mate-
rial, the other ones are from base material.
In addition, for the VTT data, the data scattering was as expected, as it is shown
in Fig. 6. The number of data points below the 2 % confidence limit corresponds
well with the Master Curve description for the RPV head weld shown.
A summary of results is given in Table 2. As the main focus of the investigations
was the understanding of the irradiation behavior of the high Ni weld materials,
in this table additional data are also plotted from the high Ni weld P16 from the
German research programs CARISMA [3]/CARINA [4]. This weld material, manu-
factured also by Uddcomb company, has a Cu-content of 0.08 wt. % and a Ni con-
tent of 1.69 wt. %. In the 1980 s, it was irradiated in the former VAK test reactor at
high flux to different neutron fluences.
FIG. 8 Fracture toughness of the R3 and R4 RPV materials obtained from the
reconstituted SE(B) 10 mm by 10 mm by 55 specimens and normalized by
T-RTT0.
MAY ET AL., DOI 10.1520/STP157620130189 99
TABLE 2 Experimental results from Master Curve testing on the R3 and R4 materials from the RPV
surveillance programs, from the high Ni Uddcomb weld material obtained from the
replaced Ringhals RPV head and, for comparison, data from German research programs
CARISMA [3]/CARINA [4] on the high Ni Uddcomb weld P16.
Neutron Fluence Specimen Testing Irradiation

Material (cm–2) (E > 1 MeV) T0 ( C) Size (mm) Laboratory Facility
Ringhals 3 BM lower shell 4.34 1019 –78.5 10 10 55 AREVA Ringhals 3

6.39 1019 –44.5 10 10 55 AREVA Ringhals 3
Ringhals 3 BM inter. shell 4.34 1019 –81.5 10 10 55 AREVA Ringhals 3
6.39 1019 –31.5 10 10 55 AREVA Ringhals 3
Ringhals 3 WM 3.72 1019 57 5 10 55 VTT Ringhals 3
5.47 1019 85 5 10 55 VTT Ringhals 3
5.47 1019 102 5 10 55 VTT Ringhals 3
7.43 1019 137 10 10 55 AREVA Ringhals 3
Ringhals 4 BM lower shell unirr. –135 10 10 55 AREVA None
3.30 1019 –26 10 10 55 AREVA Ringhals 4
6.03 1019 10 10 10 55 AREVA Ringhals 4
Ringhals 4 BM inter. shell unirr. –123 10 10 55 AREVA None
3.30 1019 –73 10 10 55 AREVA Ringhals 4
6.03 1019 –52.5 10 10 55 AREVA Ringhals 4
Ringhals 4 WM unirr. –125 5 10 55 VTT None
2.94 1019 4 5 10 55 VTT Ringhals 4
5.23 1019 34 10 10 55 AREVA Ringhals 4
5.39 1019 47 5 10 55 VTT Ringhals 4
5.39 1019 43 5 10 55 VTT Ringhals 4
7.27 1019 57.5 10 10 55 AREVA Ringhals 4
8.36 1019 76 10 10 55 AREVA Ringhals 4
PV Head WM unirr. –97 5 10 55 VTT None
2.52 1019 0 5 10 55 VTT Halden (high flux)
P16 WM unirr. –86 CT-25 HZDR None
8.36 1018 –36 10 10 55 AREVA VAK (high flux)
5.43 1019 120 10 10 55 AREVA VAK (high flux)
Some of the data sets, especially from the base materials, had data points
outside the 96 % confidence interval. An additional dataset analysis on the base
material data concerning bimodal or multimodal behavior according to Ref. [14]
showed that for only 3 out of 10 base material data sets, inhomogeneous
behavior could not be excluded. However, T0 values of these data sets are in good
agreement with the T0 values of data sets where no indication of inhomogeneity
was observed.
Discussion of the Results on the High Nickel

Weld Materials
The results on the high Ni weld materials from Table 2 are plotted in Fig. 9. The data
of the materials have been fitted by a simple equation
(1) RTT0 ¼ RTT0;unirr þ A /n
where:
RTT0,unirr. ¼ reference temperature RTT0 in the unirradiated state
(RTT0 ¼ T0 þ 19.4 K),
/ ¼ neutron fluence,
A ¼ fluence coefficient, and
n ¼ fluence exponent.
For the data fitting of the R3 weld materials, where unfortunately no unirradi-
ated state was available for material testing, it has been assumed that the RTT0 of
the unirradiated state is identical to that of R4 WM, even though it might be
expected that the RTT0 of R3 WM could be slightly higher.
From the R3 and R4 WM data shown in Fig. 9, it can be concluded that the
influence of the testing laboratory is negligible. The different specimen sizes and the
different applied reconstitution techniques, i.e., electron beam welding and arc stud
welding, do not have a significant influence on the test results. The applicability of
the smaller specimens used in the VTT investigations is confirmed. The stable trend
FIG. 9 Reference temperatures RTT0 for several high nickel weld materials as a function
of neutron fluence.
MAY ET AL., DOI 10.1520/STP157620130189 101
of increasing RTT0 with increasing fluence indicates that the influence of data scat-
tering is rather low.
All materials that are shown in Fig. 9 have high toughness in the unirradiated
state, but they show a pronounced irradiation response, confirming the observa-
tions from the standard surveillance testing. The fluence exponent n, calculated
from Eq 1, varies between 0.41 n 0.76. R4 WM has the lowest RTT0 in the uni-
rradiated as well as in the irradiated conditions, P16 has the highest RTT0 in the
unirradiated condition and, due to higher fluence exponent, also the highest refer-
ence temperatures in the irradiated condition. The data of the exchanged PV head
and of R3 WM lie in between the data of R4 WM and P16. From the comparison to
the behavior of the other materials, it could be expected that the RTT0 of the unirra-
diated R3 WM, which has not been determined, might be slightly higher than that
of R4 WM.
From Fig. 9, it can be clearly seen that the irradiation response follows a power
law function even in the higher fluence range. MnNi rich late-blooming phases [15]
or unexpected saturation effects are not observed in the current investigations. The
fluence exponents n of up to 0.76 are slightly lower than the fluence exponent of
0.8 used in the Russian safety evaluation guide for VVER-1000 high Ni
(1.00–1.90 wt. %) base and weld materials [16,17], a prediction equation that is
explicitly designed for elevated contents of nickel.
As the chemical composition of the compared materials is similar, the differen-
ces in RTT0 of the unirradiated condition might be attributed to manufacturing
effects like, e.g., the heat treatment conditions. However, even if scatter of Master
Curve testing is taken into account (1r 8 C according to ASTM E1921 [13], if 7
valid tests have been used to determine T0), with increasing neutron fluence the dis-
tance between the individual curves increases because the fluence exponent of the
curves is different. This might lead to the conclusion that not only the differences
in RTT0 of the unirradiated conditions but also the differences in irradiation
response might be influenced by manufacturing conditions, as the content of irradi-
ation relevant elements (mainly Ni, Cu, P) for these materials is very similar. By
examination of the manufacturing documentations, an attempt has been made to
find manufacturing differences between these materials, which could explain the
differences in material behavior. One particular difference was the heat treatment
during manufacturing. The tempering temperature from the manufacturing process
might be relevant for the toughness in the unirradiated state as well as for the
irradiation response of these high Ni welds: P16 had the lowest tempering tempera-
tures, R4 WM the highest (see Table 3). It might lead to the conclusion that for these
high Ni weld materials, lower tempering temperatures, especially when combined
with short holding times, lead to a higher irradiation response and maybe also to a
higher reference temperature in the unirradiated state.
Further investigations will be necessary to investigate this effect in more detail.
Therefore, an investigation program will be established, including manufacturing of
a surrogate weld material with the same chemical composition as in Ringhals 3 and
TABLE 3 Tempering parameters during manufacturing of high nickel weld materials.
Material Temperature ( C) Time (h)
Ringhals 3 WM (W4) 575 6 25 10

620 6 15 2
575 6 25 18
620 6 15 26.5
Ringhals 4 WM (W4) 620 6 15 6
(W4-0) 620 6 50 8.5
(W4-0) 620 6 15 15
P16 WM (BW3) 535-565 2
" 2.1
" 2.2
" 2.2
" 3
585-610 8
PV Head WM 560 1.5
620 5.7
4 RPV, according to original welding specifications. If the material properties of the

original RPV surveillance materials can be successfully reproduced, the influence of
heat treatment condition will be investigated by applying different heat treatments
and subsequently performing test reactor irradiation and mechanical testing. As in
the present work, specimen reconstitution will be required due to limited space
inside the test reactor irradiation capsules.
Discussion of the Results on the Base Materials

The results of the base materials presented in Table 2 show that even in the high flu-
ence regime, all base materials of Ringhals 3 and 4 have relatively low reference
temperatures, especially in comparison to the weld materials. All experiments on
the base materials have been performed in one laboratory with one specimen geom-
etry; therefore, a comparison of laboratory or specimen type influence cannot be
performed from these results.
For an interpretation of the test results, available results on similar materials
from the German research programs CARISMA [3] and CARINA [4] have been
used for comparison. These materials are 22NiMoCr3-7 forgings named P147 BM,
P150 BM and P151 BM. Their detailed chemical composition is shown in Table 4.
For comparison, the chemical composition from the Ringhals base materials from
Table 1 is also shown in Table 4.
From Table 4 it can be seen that these CARISMA / CARINA 22NiMoCr3-7
forgings are very similar to the SA 508 Class 2 ring forgings of R3 and R4. P151 BM
MAY ET AL., DOI 10.1520/STP157620130189 103
TABLE 4 Chemical composition of the CARISMA [3] / CARINA [4] forgings P147 BM, P150 BM and
P151 BM as well as from the Ringhals Unit 3 and 4 base materials.
Chemical Composition (wt.)
Material C Si Mn P S Cr Mo Ni Cu
P147 BM 0.21 0.22 0.85 0.006 0.006 0.39 0.55 0.84 0.05
P150 BM 0.19 0.23 0.84 0.008 0.009 0.35 0.52 0.83 0.05
P151 BM 0.18 0.15 0.81 0.006 0.008 0.40 0.53 0.96 0.09
R3 BM lower shell 0.19 0.19 0.74 0.006 0.006 0.42 0.60 0.90 0.10
R3 BM inter. shell 0.18 0.19 0.76 0.006 0.006 0.43 0.54 0.88 0.08
R4 BM lower shell 0.21 0.18 0.75 0.006 0.008 0.43 0.63 0.88 0.10
R4 BM inter. shell 0.21 0.20 0.71 0.006 0.007 0.45 0.63 0.88 0.09
was manufactured by Kloeckner, the same manufacturer as for the Ringhals forg-
ings. P147 BM and P150 BM were produced by JSW.
In Fig. 10, the RTT0 data of the RPV base materials (core region) of Ringhals 3
and Ringhals 4 are plotted as a function of neutron fluence, based on the data from
Table 2. For comparison, the CARISMA [3] / CARINA [4] base materials P147,
P150, and P151 are also shown. Data points from each material have been con-
nected by dashed lines. These lines, however, do not represent a curve fit, since for
some of the materials the number of data points is rather low. As in all of the
Master Curve experiments on the Ringhals base materials, only one data point
slightly violated the stable crack growth criterion. An influence of extensive ductile
FIG. 10 Comparison of RTT0 data from the Ringhals 3 and 4 base materials and from
the CARISMA [3] / CARINA [4] materials P147, P150 and P151.
tearing can be excluded. The stable trend of increasing RTT0 with increasing fluence
indicates that the influence of data scattering should be rather low.
The base materials of Ringhals 4 show a very high toughness in the unirradi-
ated state. Data for Ringhals 3 are not available in the unirradiated state, but from
the data in the irradiated state it can be concluded by extrapolation to zero fluence
that the toughness in the unirradiated state is also very high and perhaps the RTT0
values for both R3 BM could be even lower than that of Ringhals 4.
In addition, in the irradiated state, all base materials have a good toughness up
to the neutron fluence range investigated. The lower shell of Ringhals 4, however,
shows higher reference temperatures than the other materials. From the chemical
composition, the reason for this increased irradiation response is not obvious; the
content of chemical elements relevant for irradiation hardening (mainly Cu, Ni, P)
does not significantly differ from that of the other Ringhals base materials.
In case of the CARISMA [3] / CARINA [4] base materials P147, P150, and
P151, the RTT0 data of the unirradiated as well as in the irradiated states are within
a very narrow temperature range. This is somehow surprising, as for example P151
has a slightly higher Ni content and a significantly higher Cu content than P147
and P150, therefore, also a higher irradiation response would be expected.
The base materials of Ringhals 4 show a higher toughness in the unirradiated
state, compared to the CARISMA/CARINA materials. However, it has to be men-
tioned that the scatter of data points was relatively high for the Master Curve of R4
lower shell (unirr.). The data points of R3 surveillance capsule W (neutron fluence
4.34 1019 n/cm2, E > 1 MeV) are in very good agreement to the RTT0 data deter-
mined for the CARISMA/CARINA materials. At larger neutron fluences, the RTT0
values of the Ringhals base materials are higher than that of P147, P150, and P151;
nevertheless, the toughness of all materials is still very good.
It has been confirmed that the irradiation induced shift in T0 is similar to the
irradiation induced shift in the Charpy transition temperature T41 [4]. Therefore,
prediction equations for transition temperature shift can also be applied to the
shifts of T0 data. In Fig. 11, the measured T0 shifts of the Ringhals 4 base materials
are compared to the predictions from the Reg. Guide 1.99 Rev. 2 Pos. 1 model.
All of the materials which are plotted in Fig. 11 are relatively similar in terms of
overall chemical composition; however, the behavior in comparison to the predic-
tion equation shows relatively strong differences. Especially for the data from the
lower shell of Ringhals 4, there is a strong deviation from the Reg. Guide prediction.
The unirradiated state of R4 lower shell showed an unusually low reference temper-
ature of T0 ¼ –135 C, accompanied by slightly enhanced scattering of data points.
This data set has also been one of the three base material datasets, where an evalua-
tion according to Ref. [13] showed indications of inhomogeneity. It might lead to
the conclusion that a more representative T0 of this material state could be slightly
higher, e.g., if more data points would be measured, leading to a lower shift in DT0
for the irradiated states. In that case, the data points from the lower shell of Ring-
hals 4 in Fig. 11 would come closer to the 45 line.
MAY ET AL., DOI 10.1520/STP157620130189 105
FIG. 11 Comparison of measured shift in T0 to the predicted shift according to Reg.

Guide 1.99 Rev. 2 Pos. 1.
Furthermore, the comparison of measured Charpy shift data and Master Curve
shift data in Fig. 12 indicates that the DT0 for the lower shell of Ringhals 4 is higher
than usually observed. Usually, shift in Charpy transition temperature and shift in
T0 are close to a 1:1 correlation [4]. For the lower shell of the Ringhals 4 RPV, how-
ever, shift in T0 is significantly higher for both data sets, showing that the very low
measured T0 value of the unirradiated state could be higher, if more fracture tough-
ness specimens would be measured for this material state.
Nevertheless, since Fig. 10 showed that in the irradiated condition the reference
temperatures also deviate from that of materials with similar chemical composition,
and from the application of the Reg. Guide model even under consideration of a
possible bias from the unirradiated lower shell data from Ringhals 4, it seems that
the irradiation response of R4 lower shell cannot be explained only on the basis of
chemical composition. It could be possible that the manufacturing process, heat
treatment, etc., also play an important role. Moreover, the irradiation response of
the intermediate shell is underpredicted by the Reg. Guide 1.99 Rev. 2 model equa-
tion. This confirms the observations from the standard irradiation surveillance pro-
gram shown in Fig. 1, where at higher neutron fluences all Charpy shift data of the
Ringhals 3 and 4 base materials are above the Reg.Guide 1.99 Rev. 2 prediction.
FIG. 12 Comparison of measured shift in T0 to the measured shift in T41 for the Ringhals
4 base materials.
From Fig. 11, it can also be concluded that the CARISMA/CARINA base materi-
als P147, P150, and P151 are better predicted by the Reg. Guide model than the
Ringhals base materials. P147 and P150 are in particularly good agreement with the
Reg. Guide 1.99 Rev. 2 prediction. For P151, the Reg. Guide prediction is higher
than the observed DT0 shift because the increased Cu and Ni content of this mate-
rial surprisingly do not have an enhanced irradiation response as consequence. The
results from P151 and from the Ringhals 4 base materials indicate that the chemical
composition alone is not able to explain the irradiation behavior of the RPV materi-
als. As for the weld materials described above, also for the base materials, the manu-
facturing history might have a significant influence.
Summary and Conclusions

For the Ringhals 3 and 4 PWR RPV, the mechanical testing program has been
expanded, exceeding the requirements of the standard testing program and cover-
ing both base and weld materials. To increase the understanding of the material
behavior, extended Master Curve testing was performed on SE(B) 10 mm by 10 mm
by 55 mm as well as on smaller SE(B) 5 mm by 10 mm by 55 mm specimens from
MAY ET AL., DOI 10.1520/STP157620130189 107
irradiation surveillance capsules with the help of specimen reconstitution

technique.
It has been shown that the influence of the testing laboratory is negligible. The
different specimen sizes and the different reconstitution techniques electron beam
welding and arc stud welding, which have been applied, do not have a significant
influence on the test results. The applicability of the smaller SE(B) 5 mm by 10 mm
by 55 mm specimens has been confirmed.
Results have been compared to existing data on similar base and weld materials
from the German research programs CARISMA and CARINA. Late-blooming
effects or sudden saturation effects have not been observed for the base and high
nickel weld materials investigated under LTO conditions.
The strong irradiation response of the high Ni weld materials observed from
standard irradiation surveillance testing has been confirmed with the Master curve
test results. The data for the four different high Ni weld materials of similar chemi-
cal composition indicate that not only the chemical composition but also other
influencing factors like, e.g., the welding heat treatment may be important for the
reference temperature of the unirradiated state as well as the irradiation behavior. A
program for future investigations on this heat treatment effect has been outlined.
The results of the Ringhals SA 508 Class 2 base materials have shown that
even in the high fluence regime all base materials of Ringhals 3 and 4 have rela-
tively low reference temperatures, especially when compared to the weld materi-
als. Nevertheless, especially the irradiation response of the lower RPV shell of
Ringhals 4 is underpredicted by the Reg. Guide 1.99 Rev.2 model. It seems that
also for these base materials the irradiation behavior cannot be explained only on
the basis of chemical composition. The comparison to similar base materials
from the German research programs CARISMA and CARINA shows that in base
metal there might also be additional factors from the manufacturing history,
which might have to be taken into account for a full understanding of material
behavior under irradiation.
ACKNOWLEDGMENTS
The technical input of Inga Terharen concerning the commissioning of the electron
beam welding equipment within her diploma thesis is gratefully acknowledged.
References
[1] 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, Paper No. NUREG/
CP-0131 (ORNL/TM-12413), Oct 1993.
[2] ASME Boiler and Pressure Vessel Code: An American National Standard, Code
Case N-629, “Use of Fracture Toughness Data to Establish Reference
Temperature for Pressure Retaining Materials,” Section XI, Division 1, ASME, New
York, 1999.
[3] Hein, H., Keim, E., Schnabel, H., Seibert, T., and Gundermann, A., “Final Results from the
Crack Initiation and Arrest of Irradiated Steel Materials Project on Fracture Mechanical
Assessments of Pre-Irradiated RPV Steels Used in German PWR,” J. ASTM Int., Vol. 6,
No. 7, 2009, JAI101962.
[4] Hein, H., Keim, E., Schnabel, H., Barthelmes, J., Eiselt, C., Obermeier, F., Ganswind, J., and
Widera, M., “Final Results from the CARINA Project on Crack Initiation and Arrest of Irra-
diated German RPV Steels for Neutron Fluences in the Upper Bound,” Proceedings of
the 26th Symposium on Effects of Radiation on Nuclear Materials, Indianapolis, IN, June
12–13, 2013.
[5] Efsing, P., Rouden, J., and Lundgren, M., “Long Term Irradiation Effects on the Mechani-
cal Properties of Reactor Pressure Vessel Steels from Two Commercial PWR Plants,”
Effects of Radiation on Nuclear Materials: 25th Volume, ASTM STP 1547, ASTM Interna-
tional, West Conshohocken, PA, 2012.
[6] Randall, P. N., “Basis for Revision 2 of the U. S. Nuclear Regulatory Commission’s Regula-
tory Guide 1.99,” Radiation Embrittlement of Nuclear Reactor Pressure Vessel Steels: An
International Review (Second Volume), ASTM STP 909, ASTM International, Philadelphia,
PA, 1986.
[7] Kirk, M., “A Wide-Range Embrittlement Trend Curve for Western Reactor Pressure Ves-
sel Steels,” Effects of Radiation on Nuclear Materials: 25th Volume, ASTM STP 1547,
ASTM International, West Conshohocken, PA, 2012.
[8] Soneda, N., Nishida, K., Nakashima, K., and Dohi, K., “High Fluence Surveillance Data and
Recalibration of RPV Embrittlement Correlation Method in Japan,” Proceedings of the
ASME 2013 Pressure Vessels and Piping Conference, Paper No. PVP2013-98076, Paris,
France, July 14–18, 2013.
[9] Miller, M. K., Powers, K. A., Nanstad, R. K., and Efsing, P., “Atom Probe Tomography
Characterizations of High Nickel, Low Copper Surveillance RPV Welds Irradiated to High
Fluences,” J. Nucl. Mater., Vol. 437, 2013, pp. 107–115.
[10] May, J., Hein, H., Ganswind, J., and Widera, M., “Irradiation Behavior of German PWR
RPV Steels Under Operating Conditions,” Proceedings of Fontevraud 7, Avignon, France,
Sept. 26–30, 2010.
[11] Wallin, K., Planman, T., Valo, M., and Rintamaa, R., “Applicability of Miniature Size Bend
Specimens to Determine the Master Curve Reference Temperature T0,” Eng. Frac. Mech.,
Vol. 68, 2001, pp. 1265–1296.
[12] Keim, E., Langer, R., van Walle, E., Scibetta, M., Valo, M. J., Viehrig, H.-W., Richter, H.,
Atkins, T., Wootton, M. R., Debarberis, L., and Horsten, M., “RESQUE: Determination of
the Lowest Acceptable Insert Length for Reconstituted CV-Impact and Precracked CV
Specimen,” Small Specimen Test Technique: Fourth Volume, ASTM STP 1418, ASTM Inter-
national, West Conshohocken, PA, 2002.
[13] ASTM E1921-11a: Standard Test Method for Determination of Reference Temperature, T0,
for Ferritic Steels in the Transition Range, ASTM International, West Conshohocken, PA,
2011, www.astm.org.
MAY ET AL., DOI 10.1520/STP157620130189 109
[14] Wallin, K., Nevasmaa, P., Laukkanen, A., and Planman, T., “Master Curve Analysis of Inho-
mogeneous Ferritic Steels,” Eng. Frac. Mech., Vol. 71, 2004, pp. 2329–2346.
[15] IAEA, “Integrity of Reactor Pressure Vessels in Nuclear Power Plants: Assessment of
Irradiation Embrittlement Effects in Reactor Pressure Vessel Steels,” IAEA Nuclear
Energy Series, No. NP-T-3.11, 2009, IAEA, Vienna, Austria.
[16] Margolin, B. Z., Nikolaev, V. A., Yurchenko, E. V., Nikolaev, Y. A., Erak, D. Y., and Nikolaeva,
A.V., “Analysis of Embrittlement of WWER-1000 RPV Materials,” IAEA Technical Meeting
on Irradiation Embrittlement and Life Management of RPVs, Znojmo, Czech Republic,
Oct. 18–22, 2010.
[17] Ahlstrand, R., Margolin, B., Akbashev, I., Chyrko, L., Kostylev, V., Yurchenko, E., Piminov,
V., Nikolaev, Y., Koshkin, V., Kharshenko, V., Bukhanov, V., and Comsa, O., “TAREG 2.01/
00 Project, ‘Validation of Neutron Embrittlement for VVER 1000 and 440/213 RPVs,
With Emphasis on Integrity Assessment,’ ” Proceedings of the 3rd International Confer-
ence on Nuclear Power Plant Life Management (PLiM), Salt Lake City, UT, May 14–18,
2012.
Matti J. Valo,1 Tapio K. Planman,1 and Kim R. Wallin1
Rotation Point and KJC

Estimations for Miniature
CT-Specimens Based on
Off-Load Line Displacement
Reference
Valo, Matti J., Planman, Tapio K., and Wallin, Kim R., “Rotation Point and KJC Estimations for
Miniature CT-Specimens Based on Off-Load Line Displacement,” Small Specimen Test
ABSTRACT
In fracture toughness tests, the energy absorbed by the specimen is calculated
from load versus load point displacement data. If tests are made with small
CT-specimens, direct measurement of load point displacement is often not
possible due to small specimen dimensions; subsequently, specimen
displacement is measured on the front face of the specimen. The measured
displacement is transferred into load point displacement with the help of a
geometric rotation point concept. The measurements were performed with
miniature CT-specimens of dimensions B ¼ 4 mm, W ¼ 8 mm having the integral
clip seated on the specimen front face. The location of the rotation point was
concluded from the measured initial compliance by applying the rotation point
values of r ¼ 0.5 and r ¼ 0.13 and by comparing the calculated crack lengths with
the measured crack lengths. The value of r ¼ 0.5 was observed to be too high
and the value r ¼ 0.13 produced the measured crack lengths accurately. A linear
combination of elastic and plastic deformations is proposed to be used for
calculating the location of the rotation point outside the elastic loading
range. The load line load–displacement curves were derived and the measured
KJC-values were calculated with the assumptions r ¼ 0.5 and with the linear
Manuscript received January 27, 2014; accepted for publication January 8, 2015; published online February
11, 2015.
1
VTT Technical Research Centre of Finland, P.O. Box 1000, 02044 VTT, Finland.
2
Houston, TX.
Copyright V
VALO ET AL., DOI 10.1520/STP157620140010 111
combination described in the paper. The linear combination believed to be more

relevant than the value r ¼ 0.5 resulted in lower KJC-values than r ¼ 0.5. However,
the difference was only about 1.6 % in KJC-values, which changes the T0 values
approximately by 1 C. Hence, the measured KJC-values are not sensitive to the
exact location of the rotation point used in the analysis.
Keywords
fracture toughness test, miniature CT specimen, rotation point, off-load line
deflection
Introduction
In fracture toughness analysis, the basic data is load and load line displacement.
When small CT-specimens are used, it is not always possible to measure specimen
displacement on the load line due to space limitations. Conversion of the off-load
line displacement to load line values is conventionally performed with the help of a
rotation point concept. The exact position of the rotation point is material specific
because it is defined by the stress–strain properties of the material and by specimen
plasticity, and hence the rotation point approach is approximate. In this paper,
the value of the rotation point is derived from the elastic loading of miniature CT
specimens used in the CRIEPI co-ordinated International Research Programme
“Miniature CT Master Curve Round-Robin Test” targeting to validate this speci-
men size for surveillance testing [1]. The effect of two rotation point approxima-
tions and their combination on the measured KJC-values is estimated. The CRIEPI
miniature CT specimen is shown in Fig. 1. In the CRIEPI programme, master curve
T0-values were measured; hence the end of test displacement values were relatively
small and geometric rotation correction was not necessary.
ROTATION POINT CONCEPT

When CT specimens are loaded, tension loads are characteristic in the near crack
locations and compressive loads in the back end of specimen ligament. Neutral axis
is located some distance from the centre of the ligament towards the crack tip and
the specimen rotates momentarily around this point. The relevant parameters for
consideration are defined in Fig. 2.
Location of the rotation point is defined by the rotation factor
Ra
(1) r¼
Wa
where:
R ¼ radius of rotation,
W ¼ specimen width, and
a ¼ crack length.
FIG. 1 Miniature CT specimen used in the international CRIEPI round robin exercise.
FIG. 2 Relevant dimensional parameters and definition of the rotation point.

VALO ET AL., DOI 10.1520/STP157620140010 113
FIG. 3 Calculated rotation point values for elastic and plastic specimen loading [2–5].
The geometric relation between the load line and off-load line displacements is
a þ r ðW aÞ
(2) DLL ¼ kða; r; W; XÞ DXLL ¼ DXLL
a þ r ðW aÞ þ X
with parameters defined in Fig. 2. The relation is simply the ratio of the distances of
the measuring locations from the rotation point. The position of the rotation point
is determined by the stress–strain curve of the material and the degree of plasticity
of the specimen ligament. Calculated rotation point values are given in Ref. [2]; the
data is based on Refs. [3–5] and is shown in Fig. 3. The values are approximately
r 0.13 for the fully elastic condition and r 0.43 for the fully plastic condition for
a wide range of deep cracks.
Linear interpolation between the extreme values has been proposed [2] in the
form
0:13 ðDx Dpl Þ þ 0:43 Dpl

(3) r
Dx
where:
Dx ¼ Dpl þ Del ¼ total displacement,
Dpl ¼ plastic displacement, and
Del ¼ linear–elastic displacement.
This expression is referred in the text as r ¼ f(l,pl). It is also valid outside the
load line measuring location because it is written as a ratio of displacements and a
common factor in displacement values cancels out. This expression is used for
FIG. 4 Definition of the elastic and the plastic displacement components from the off
load line load–displacement curves.
analysing the measured KJC-values. The rotation point moves towards the centre of
the ligament, when plasticity of the specimen increases during the test. In conven-
tional tests, crack length is near a/W ¼ 0.5, where the value of r is relatively constant
as a function of a/W. In the current work, the elastic behaviour is defined by fitting
a straight line between the load values of 0.4Pm and 1.2Pm, where Pm is the prefati-
guing force defined in ASTM E1820-13, A2.3 [6]. The elastic and plastic deflection
components are determined as described in Fig. 4.
ROTATION POINT DERIVED FROM SPECIMEN COMPLIANCE

The position of the rotation point can be derived from the measured specimen
compliance and the elastic modulus of the material by comparing the calculated
crack length values with the measured crack length values. The applicability of the
rotation point values of r ¼ 0.5, which is suggested to be used in data analysis in the
round robin programme and also used in ASTM E1820-13 [6], and r calculated
from Eq 3 are compared. The off-load line displacement values, i.e., the values
measured from specimen front face, are transformed to the load line values with
the help of the formula from Eq 2. The analysis is performed in the elastic part of
the loading curve. Two values for the elastic modulus E are used, i.e., E ¼ 219 GPa
suggested for the round robin exercise [1] and E ¼ 212 GPa given in Ref. [7]. The
later data is shown in Fig. 5.
The rotation point values of r ¼ 0.13 and r ¼ 0.5 are used for transforming the
off load line linear behaviour to load line linear behaviour. The corresponding crack
VALO ET AL., DOI 10.1520/STP157620140010 115
FIG. 5 Temperature dependence of elastic modulus according to Ref. [7].
lengths are calculated from the ASTM E1820-13 (A2.12) [6] formula (Eq 4). The
crack length formula applies to load line compliance only, and it is said to have bet-
ter accuracy than 0.5 %. The two elastic modulus values are combined with the
rotation point values and the calculated crack lengths are compared with the meas-
ured values. The difference of the measured and calculated values is the criterion
for consistency.
(4) a=W ¼ 1:000196 4:06319u þ 11:242u2 106:043u3 þ 464:335u4 650:677u5
where
u ¼ 1=ð½BEC 1=2 þ 1Þ,
C ¼ specimen load-line compliance in the linear loading range,
E ¼ elastic modules, and
B ¼ specimen thickness.
The difference of the estimated and measured crack lengths for each tested
specimen is shown in Fig. 6. The data shows that r ¼ 0.5 results in too long crack
lengths, the use of the value r ¼ 0.13 improves the consistence clearly, and change
of the modulus from 219 to 212 GPa practically produces the measured crack
lengths. The value r ¼ 0.5 leads to higher transfer coefficient k than the value
r ¼ 0.13 and hence to excessive displacement values, which means higher crack
lengths (see also Fig. 9). Typical specimen fracture surface is shown in Fig. 7, where
the magnitude of the difference found in the analysis, i.e., 0.2 mm, is shown on the
scale of the fracture surface.
FIG. 6 Crack length values estimated from the measured specimen compliance (aestim)
are compared with the measured crack lengths (ameas) for each tested
specimen in the linear range of loading. Two rotation factor descriptions, i.e.,
r ¼ 0.5 and r ¼ 0.13, and two values for elastic modulus, i.e., E ¼ 219 and 212 GPa,
were used in the estimations. The combination r ¼ 0.13 and E ¼ 212 GPa
reproduces the measured crack lengths with good accuracy, the overestimation
being only 0.03 mm on average.
THE ESTIMATED KJC-VALUES AND THE ROTATION POINT

Plasticity of the specimen increases during the test and hence the rotation point
moves during the test as described by the formula in Eq 3. The off-load line dis-
placement values are transferred to load line values by the formula in Eq 2, where r
is one parameter. Hence KJC-values depend on the applied r and this dependence is
estimated for two values of the r-parameter.
If r ¼ 0.5 the measured off-load line displacement values will be multiplied by a
factor close to 0.75 in order to get the load line values. If the formula in Eq 3 is
applied, a slightly more complicated process is needed. The elastic and plastic dis-
placement components are derived from the off-load line data as shown in Fig. 4
and the r-values are calculated from the formula in Eq 3. At low displacement val-
ues, the formula in Eq 3 gives noisy data due to a small plastic component; however,
it is known that at low displacement, r ¼ 0.13, which is used. Because the rotation
point moves during the test, the load line displacement is calculated as a sum of dis-
placement increments with the momentary r-value. The end of test displacement is
X
end
(5) DLL ¼ kða; ri ; W; XÞ ðDXLl;i DXLl;i1 Þ
i¼1
VALO ET AL., DOI 10.1520/STP157620140010 117
FIG. 7 Typical crack front in the specimens. The difference of the estimated and
measured crack lengths in the analysis is of the order of 0.2 mm, which measure
is indicated in the figure. The difference is 5 % of the crack length and it is well
measurable even if small.
where:
k(a,ri,W,X) ¼ coefficient in formula (Eq 2) at point i, and
DX-LL,i ¼ the measured off-load line displacement at point i.
The estimated load–displacement curves for specimens F0205-2 and F0232-17
are shown in Fig. 8. The derived corrected load–displacement curves are practically
identical except for the end of test displacement values, which cause some difference
to the plastic area and hence to the KJC-values.
FIG. 8 The off-load line, the load-line r ¼ 0.5 and the formula in Eq 2 based
load–displacement curves for specimens F0205-2 and F0232-17.
The evolution of r and k during loading of specimen F0205-2 is shown in Fig. 9.

Cleavage initiation KJC-values were calculated for all tested specimen. The data
is shown in Fig. 10. The KJC-values based on r ¼ 0.5 are approximately 1.6 % higher
than the values based on the more exact relation of r ¼ f(l,pl). The difference is
small and it corresponds to a difference of approximately 1 C in the T0-values.
Conclusion
Displacement measured from the front face of the CT-specimen leads to some
uncertainty in the measured KJC-values. Conventionally, the off-load line
FIG. 9 Rotation factor r and the coefficient k, which is used to transform off-load line
displacement to load line displacement based on the formulas in Eqs 2 and 3 are
shown as a function of off-load line displacement. For a fully plastic loading
condition r ¼ 0.43. For a constant rotation factor of r ¼ 0.5 the value of k ¼ 0.75.
VALO ET AL., DOI 10.1520/STP157620140010 119
FIG. 10 Difference of the KJC-values, i.e., the values based on r ¼ 0.5 and on r ¼ f(l,pl), as
a function of KJC(r ¼ f(l,p)l). The difference is approximately 1.6 %, which means
a difference of 1 C in T0 values.
displacement data is transformed into load line data with the help of a rotation
point concept, which is based on geometry. The location of the rotation point is a
material property and varies as a function of specimen plasticity; hence it cannot be
determined universally. In this work, the rotation point in the elastic loading range
is determined from the measured elastic compliance of the specimen. The off-load
line displacement is transferred into the load line displacement by using the value
r ¼ 0.5 as a rotation factor and a value defined as a linear combination of elastic
and plastic displacements. Compliance is also a function of elastic modulus E and
two values for this modulus are used. The rotation factor r ¼ 0.5 is clearly too high
and the value r ¼ 0.13 in the elastic loading part gives better consistency with meas-
ured specimen crack lengths. The elastic modulus of 212 GPa gives better consis-
tency with the measured crack length values than the value of 219 GPa.
The data considered is used for measuring KJC and the Master curve T0 with
miniature CT specimens. The load–displacement curves were estimated with rota-
tion factors r ¼ 0.5 and with a linear function of elastic and plastic deformations
described in the paper. In the elastic loading range, r ¼ 0.13, and this value grows
up to a value of r ¼ 0.365 in maximum in current tests. The k-coefficient used for
transferring the off-line displacement into load line displacement is 0.75, when
r ¼ 0.5. The k-values grow when r changes from 0.13 to 0.365, from k ¼ 0.69 to
0.73. Hence the displacement based on r ¼ f(l,pl) is always smaller than the
displacement based on r ¼ 0.5. Respectively, the measured KJC-values will be

smaller but only by a factor of 1.7 %. This difference in KJC-values corresponds to a
difference of 1 C in the T0-values, which is for all practical applications insignifi-
cant. Hence the measured KJC-values are not sensitive to the exact position of the
rotation point.
ACKNOWLEDGMENTS
The tests were performed as part of the CRIEPI miniature CT specimen round robin
exercise. CRIEPI provided the specimens and a clip gauge for the tests. The work was
performed within SAFIR2013 project, which is financed by the State Nuclear Waste
Research Fund (VYR) and VTT Technical Research Centre of Finland.
References
[1] Miura, N. and Soneda, N., “Evaluation of Fracture Toughness by Master Curve Approach
Using Miniature C(T) Specimens,” J. Pressure Vessel Technol., Vol. 134, No. 2, 2012.
[2] Wallin, K., Fracture Toughness of Engineering Materials, Estimations and Applications,
EMAS Publishing, Warrington, UK, 2011.
[3] Saxena, A. and Hudak, S. J., “Review and Extension of Compliance Information for Com-
mon Crack Growth Specimens,” Int. J. Fract., Vol. 14, No. 5, 1978, pp. 453–468.
[4] McMeeking, R. M., “Estimates of the J-Integral for Elastic–Plastic Specimens in Large
Scale Yielding,” J. Eng. Mater. Technol., Vol. 106, No. 3, 1984, pp. 278–284.
[5] Kumar, V., German, M. D., and Shih, C. F., “An Engineering Approach for Elastic–Plastic
Fracture Analysis,” EPRI Report NP-1931, Electric Research Institute, Palo Alto, CA, 1981.
[7] ASME B31.1-1995: Power Piping ASME Code for Pressure Piping, B31: An American
National Standard, ASME, New York, 1995, www.asme.org.
T. S. Byun,1 S. A. Maloy,2 and J. H. Yoon3
Small Specimen Reuse Technique

to Evaluate Fracture Toughness
of High Dose HT9 Steel
Reference
Byun, T. S., Maloy, S. A., and Yoon, J. H., “Small Specimen Reuse Technique to Evaluate
Fracture Toughness of High Dose HT9 Steel,” Small Specimen Test Techniques: 6th Volume,
STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 121–142, doi:10.1520/
ABSTRACT
A small specimen reuse technique has been developed to evaluate static fracture
toughness using the tested halves of subsize Charpy impact specimens. The new
testing and evaluation procedure consisted of introducing a diamond-saw notch
to the miniature bars (13 by 3 by 4 mm), precracking using embedded
displacement gage signal only, static fracture testing at high temperatures in
vacuum, and calculation of J–R curves using a modified normalization method.
The HT9 steel has been used as a primary core material for fast fission reactors
because of its low radiation-induced swelling and high resistance to irradiation
hardening and embrittlement. To maximize the use of existing high dose
materials in building mechanical property knowledgebase for the high dose HT9
steel, the fracture resistance (J–R) curves and fracture toughness values were
evaluated using the small single-edged bend (SEB) bars in nonirradiated, as-
irradiated, and thermally-annealed (at 550 or 650 C) conditions. The original
impact specimens were machined from the wall of the ACO-3 duct of fast flux
test facility (FFTF) and archive (nonirradiated) material plate. Regardless of
vastly different irradiation conditions, no fracture toughness
p data measured after
post-irradiation annealing was lower than 130 MPa m, indicating that the
Manuscript received February 26, 2014; accepted for publication May 28, 2014; published online August 29,
2014.
1
Oak Ridge National Laboratory, Oak Ridge, TN 37831 (Corresponding author), e-mail: byunts@ornl.gov
2
Los Alamos National Laboratory, Los Alamos, NM 87545.
3
Korea Atomic Energy Research Institute, Daejeon, Korea.
4
Houston, TX.
Copyright V
thermal annealing can be a feasible damage mitigation technique for future fast
reactor components.
Keywords
small specimen reuse technique, high dose HT9 steel, precracking, static frac-
ture testing, radiation effect, thermal annealing effect
Introduction
The materials for future power reactors such as sodium-cooled fast reactor (SFR) and
fusion reactors (FR) require excellent performance at high temperatures and high
doses as the reactor designs aim at increasingly higher thermal efficiency and longer
lifetime [1–3]. Therefore, the materials evaluation needs to handle very high dose
(100–400 dpa) specimens that can limit specimen size for reasonable cost and effi-
ciency of work in radiation area. Furthermore, the main core components in those
future reactors are generally not thick enough to apply the technologies for heavy sec-
tion components which have been fully established; for example, the fuel duct thick-
nesses in the fast flux test facility (FFTF) and traveling wave reactor (TWR) were
3–5 mm [4,5]. Compared to the evaluation of current pressurized water reactor
(PWR) materials; therefore, more extreme miniaturization techniques are needed for
the property evaluation of advanced reactor materials. A more restricted testing condi-
tion can further aggravate the difficulty in the evaluation procedure: the materials for
future reactors, which are mostly high temperature materials, need to be tested at high
temperatures (>300 C) where an environmental control is required. For the fracture
testing in a high temperature vacuum furnace, for instance, instrumentation for dis-
placement measurement from a miniature fracture specimen is extremely difficult and
thus needs to be minimized. This study aimed to establish a small specimen reuse
technique for fracture toughness evaluation that can resolve some experimental diffi-
culties and maximize the use of given high dose materials.
Since developing a new specimen reuse technique required various changes and
developments in experimental techniques as well as in analysis procedure, it was
necessary to explore new techniques and then combine with existing techniques
and procedures. Since the fracture specimens to be tested using this new technique
are prepared from an irradiated (i.e., used) component or from tested samples, the
first challenge that cannot be skipped is making a sharp fatigue crack from
machined notch tip. New techniques that can detect crack growth on fatigue load-
ing without attaching displacement gage to the specimen were explored and estab-
lished. In addition, handling of miniaturized specimens can impose more serious
practical difficulties if we have to follow the traditional standard testing procedures
for high temperature tests. In particular, although the standard fracture testing pro-
cedure needs a high precision recording on loading-unloading cycles, we often ex-
perience that the measurements of unloading compliance from miniature
specimens become highly inaccurate at elevated or high temperature because of sig-
nificant friction and plastic deformation at contact surfaces. Therefore, the major
BYUN ET AL., DOI 10.1520/STP157620140017 123
change made in the testing procedure should be removing the attachment of exter-
nal gage. In the data analysis procedure, some issues caused by this change were
resolved by adopting the curve normalization method for J–R curve construction
and by modifying other detailed calculation procedures accordingly.
The tested Charpy impact specimens (i.e., broken halves with dimensions of 3 by
4 by 13.5 mm) of high dose HT9 steel [6] were reused as the fracture specimens
tested in three-point bending (TPB) mode. Fracture toughness testing and evaluation
procedures have been newly developed or modified in accordance with the reuse of
miniature specimens. Results of fracture tests using the specimen reuse technique have
been published separately, focusing on the irradiation and thermal recovery effects in
HT9 steel [7,8]. This paper was aimed at describing the details of testing and evalua-
tion procedure including specimen preparation technique, simplification in data ac-
quisition, and J–R curve analysis procedure. Furthermore, selected results from the
application of the new techniques to HT9 steel are included in the later discussion in
order to provide examples for application and brief summaries on radiation and
annealing effects. It is noted that, because of the limitations in specimen size and ge-
ometry, the fracture toughness data presented in this article should be considered as
non-validated approximate values and be used for comparison purposes only.
Experimental
TEST MATERIALS
The test materials were the HT9 steels in pristine condition or used (irradiated) as
the ACO-3 duct in the FFTF [6,9,10]. The pristine HT9 steel had a tempered-
martensitic structure and was identified by the heat # 84 425 having nominal chem-
istry of Fe (bal.)–11.87Cr–1.02Mo–0.58Mn–0.55W–0.53Ni–0.3V–0.27Si–0.2 C (in
wt.%). Before irradiation the ACO-3 duct was processed by a final heat treatment
consisting of normalizing at 1065 C for 0.5 h/air cooling followed by tempering at
750 C for 1 h/air cooling [4,9].
To build an extended database for fracture toughness, the tested high dose HT9
steel Charpy specimens were reused for the three-point bend (TPB) fracture tests in
pristine, as-irradiated, and irradiated plus annealed conditions. The original
V-notched Charpy specimens (3 by 4 by 27 mm) in as-irradiated condition were
taken from the ACO-3 duct of FFTF and tested to measure the influence of irradia-
tion on impact properties, along with its nonirradiated archive material [6]. Irradia-
tion conditions for those as irradiated HT9 specimens have been reported in earlier
publications [6–11]: the irradiation doses and temperatures of the HT9 steel speci-
mens were in the range of 3–148 dpa and of 379 C–503 C, respectively.
PREPARATION OF TPB FRACTURE SPECIMENS

The small specimen reuse technique developed for high dose HT9 steel specimens
consisted of selection of test specimens, making diamond-saw notch in the middle
of specimens, fatigue precracking, static fracture (J–R) testing, and post-test fatigue
cracking and final separation [7,8]. A schematic flow of this specimen reuse tech-
nique is shown in Fig. 1. The specimens selected for this study were those tested at
relatively lower temperatures, i.e., < ductile-brittle transition temperature (DBTT),
which usually had minimal plastic deformation outside of fracture surface layer.
The selected sets of tested Charpy impact specimens had the dimensions of
3.05 mm (thickness ¼ B), 4 mm (width ¼ W), and 13.5 mm (length).
Prior to the static fracture (J–R) testing, the notched and precracked fracture
specimens were made from the broken halves of Charpy bars: first, the middle of
13 mm long bars was notched to a depth of 1–1.5 mm by a diamond wheel cutter
in a radiation-controlled hood. In making the notch, the diamond wheel was lubri-
cated and cooled by a water-soaked paper pad set at the bottom of bath. Later, the
wet paper pad with radioactive powders was dried by air flow in the hood and dis-
posed as solid waste. Second, creating a sharp crack in front of the diamond-saw
notch is required to conduct static fracture testing. Since devising an optical scope or
attaching a displacement gage to monitor fatigue crack growth was extremely difficult
for a small specimen being loaded in a controlled environment because of limited
space and accuracy, the most critical step for making fracture specimens was to de-
velop a precracking technique for the miniature specimens. For both precracking and
static J–R testing, the specimens were loaded in the TPB mode. After various attempts
to use machine generated signals only, it was found that the small change in the dis-
placement amplitude can reflect the crack growth during fatigue loading. It was also
FIG. 1 A schematic description for specimen reuse technique. Specimens with relatively
low plastic deformation in impact test were selected for J–R testing. Precracking
after introducing diamond-saw notch was performed at a cyclic load of
450 6 400 N to an a/W ratio of 0.5.
BYUN ET AL., DOI 10.1520/STP157620140017 125
figured out that an increase of 3 to 5 lm in the displacement amplitude was needed to

achieve about 0.5–1 mm crack growth. The amplitude change was carefully monitored
during fatigue loading so that the specimen were fatigue-precracked to roughly 2 mm
depth and the fatigue loading was stopped at a crack length-to-width ratio near 0.5.
All fatigue precracking operations were performed at room temperature in air. The
precracking condition varied with the initial notch depth and hardness of the mini-
bar specimens. A typical condition for starting crack growth was a cyclic load of
450 6 400 N at 30 Hz, and the numbers of cycles taken to 0.5–1 mm growth were in
the range of 30–100 k cycles for the majority of specimens. To avoid a coarse crack tip,
the final stage of the precracking imposed after the main precracking was to run at
least 10 k cycles after lowering the load amplitude to a half. At least six precracked
specimens were prepared per material condition [8]. Finally, it needs to be noted that
the precracking method and condition should vary with test setup because the load
train response to a cyclic loading is unique for each test setup.
SIMPLIFIED DATA ACQUISITION

In the post-irradiation fracture tests using remote manipulator in hotcells, any data
acquisition through direct connections to specimens can cause major drawbacks in
addition to general difficulties in handling miniature specimens. In the miniature tests
in a controlled environment such as vacuum, in particular, attaching a sensing device
to specimen is prohibitively difficult. It should be therefore considered if the test tech-
nique that uses gage lead or electrode directly attached to the specimen can be
avoided. Since the curve normalization method requires the simplest datasets to
obtain a J–R curve, i.e., the load–displacement curve and optical measurements of ini-
tial and final crack lengths, it was chosen for the present fracture testing of miniature
specimens in vacuum furnace. Furthermore, simplification in data acquisition has
been pursued based on the consideration that the measurement of highly accurate
displacement may not be needed for the normalization method as explained below.
In a typical fracture test system, a precision displacement gage, such as Wheat-
stone bridge type clip gage or linear variable differential transducer (LVDT), is usu-
ally attached to the fracture specimen to measure the specimen-mouth
displacement in compact tension (CT) specimens or the axial displacement (deflec-
tion) in bend-bar type specimens. In actual J–R curve calculation following the test
standard [12], the J-integral is divided into two components of elastic and plastic
works. Considering the descriptions and backgrounds of the test standard, it is eas-
ily recognized that the elastic J component is a well-defined function of load, crack
length, and elastic constants, none of which requires displacement measurement. A
precise measurement of elastic displacement or load/displacement slope is required
only if crack growth is monitored by unloading compliance method. The plastic J
component indeed requires displacement data, but a plastic component only.
Detailed practices in J-evaluation indicate that the elastic component of measured
displacement is not used in any methods of J–R curve calculation except for compli-
ance data in the unloading compliance method. In the normalized curve method, in
particular, the plastic displacement at each point can be easily separated from total
displacement using the initial slope of load–displacement curve and updated com-
pliance value, which is calculated by a function of crack length. To guarantee the ac-
curacy of plastic displacement data, however, it is important to take a correct
procedure that can effectively separate the plastic displacement component from
the total displacement measured. It is therefore emphasized that the step-1 in sec-
tion 3 below or similar step needs to be taken to the measured load–displacement
data prior to any next step calculation for constructing J–R curve.
In displacement control mode, most mechanical testing systems are controlled
by the displacement reading from built-in devices: an LVDT is typically embedded
in the actuator of a servohydraulic test system, and the rotation encoding signal is
converted to the crosshead movement or displacement in motor-driven systems.
These displacement data obtained from a built-in displacement measuring device
include the displacement components of all parts and contacts in the entire load
train, and therefore, are usually not accurate enough to be directly used in the
unloading compliance method, where an accurate measurement of elastic unload-
ing slopes is essential for converting those to crack length data. In the curve nor-
malization method, however, any non-plastic components of total displacement
can be easily removed in the J–R curve calculation procedure. Therefore, the dis-
placement reading from a built-in device can be used to extract the necessary plastic
component of displacement. In this work, therefore, a major simplification was
made in the testing setup: attachment of additional gage such as clip gage was elimi-
nated. The section entitled Procedure for J–R Curve Calculation describes the nor-
malization method modified for the simple data acquisition along with an
application example for the HT9 steel.
MINIATURE FRACTURE TEST

Based on the aforementioned considerations, the static fracture tests were
attempted without the external displacement gage attachment. Fracture resistance
(J–R) tests in quasi-static bending mode were carried out for the high-dose HT9
steel reuse specimens and non-irradiated archive HT9 specimens in an MTS 858
servo-hydraulic testing machine with a vacuum furnace. The static J–R tests were
conducted at selected temperatures ranging from 22 to 600 C in a vacuum of
104 Pa (106 torr), except for the room temperature (22 C) tests that were per-
formed in air. All of the J–R fracture tests were run in a displacement-controlled
TPB mode, and the static fracture testing and evaluation were performed following
the standard procedure described by the ASTM E1820-01 [12] and the modified
procedure used in Refs. [7,8] and described in detail here. All the J–R tests were car-
ried out at a crosshead speed of 0.005 mm/s with a temperature control within
61 C. Each specimen was soaked at the target temperature for 5–10 min prior to
the J–R testing.
To use the curve normalization method, two datasets need to be obtained: a
load–displacement curve, up to 50 % of the maximum load if passing the
BYUN ET AL., DOI 10.1520/STP157620140017 127
maximum load, and the initial and final crack lengths. The load versus load–line
displacement data were recorded at a data acquisition rate of 5 per second, and
used for the analysis to obtain interim fracture toughness (JQ). After each J–R test,
the unbroken specimen was fatigue-loaded in air to make a mark for the final crack
length before the final separation. The initial and final crack lengths were then
measured optically using the photographs as illustrated in Fig. 2.
Procedure for J–R Curve Calculation

Described below is a summarized procedure for the calculation of crack growth
(Da) and J-integral. The steps and equations described here are mostly following
the standard test method [12], expect for a few places where modifications were
made to accommodate the simplification in data acquisition (Eq 1, for example).
The main modification made in J–R data calculation is the process regarding the
use of the displacement data either from the built-in displacement device or from
cross-head movement.
STEP-1: LOAD–PLASTIC DISPLACEMENT CURVE

Since no external displacement gage is attached to the test specimen during testing,
the displacement data recorded contains undesirable components from the flexibil-
ity of whole load train including specimen-jig contacts. These erroneous compo-
nents need to be removed for a high accuracy in calculation. Using only the plastic
component is considered the best way to avoid incurring additional errors, and
therefore, the first step of the calculation procedure is to produce a load–plastic
displacement (P(i) – vp(i)) data from the recorded load–total displacement data
(P(i) – v(i)). For each data point (i), the plastic displacement component is extracted
by using load–line compliance (CLL(i)) and a constant CA:
FIG. 2 Example of fracture surface after ductile fracture test: the static and cyclic loads
produced discernable surfaces so that the initial and final crack lengths could be
measured.
e
(1) vpðiÞ ¼ vðiÞ ðCLLðiÞ þ CA ÞPðiÞ ¼ vðiÞ CLLðiÞ PðiÞ
The initial (composite) load–line compliance, CLL(0) þ CA, is measured from the
linear portion of experimental load–displacement curve and the constant CA, the
compliance from load train except specimen itself, is determined as CLL(0) can be
calculated from the measured initial crack length. The plastic displacement is set to
zero at the first point and calculated to be zero in the linear elastic loading region.
Afterwards, the plastic displacement can be calculated using the effective compli-
e
ance CLLðiÞ as the CLL(i) value is updated by using the following equation given as a
function of crack length (a(i)) [12]:
2
1 S
CLLðiÞ ¼
EBe W aðiÞ
a a 2 a 3 a 4
ðiÞ ðiÞ ðiÞ ðiÞ
(2) 1:193 1:98 þ 4:478 4:443 þ1:739
W W W W
where:
E ¼ the elastic modulus,
¼ the Poisson ratio (¼ 0.28),
Be ¼ the effective thickness (¼B without side grooves),
S ¼ the span of load supports, and
W ¼ the width of specimen.
The crack length and compliance are updated for each point as the curve normal-
ization calculation progressively provides the values of J-integral and crack length.
STEP-2. CALCULATION OF J-INTEGRAL

Although different techniques are used for evaluation of crack lengths, common
equations are used for calculation of J-integral and stress intensity factor K [12].
Those equations are summarized as noted below as many of them are intermingled
in the application of normalization method. The total J-integral at point (i) is
defined as the sum of elastic and plastic components (Je(i) and Jp(i)):
2
KðiÞ ð1 2 Þ
(3) JðiÞ ¼ JeðiÞ þ JpðiÞ ¼ þ JpðiÞ
E
where K(i) the stress intensity factor or linear-elastic component of fracture tough-
ness. When all load–displacement and geometrical parameters are known, the plas-
tic component Jp(i) can be calculated by
2ApðiÞ
(4) JpðiÞ ¼
BN ðW aðiÞ Þ
where:
Ap(i) ¼ the plastic energy applied to the specimens or area below load–plastic
displacement curve, and
BYUN ET AL., DOI 10.1520/STP157620140017 129
BN ¼ the net specimen thickness ( ¼B without side grooves).

In practical calculation, a discretized equation is used:

2 ApðiÞ Apði1Þ aðiÞ aði1Þ
(5) JpðiÞ ¼ Jpði1Þ þ 1
bi1 BN bði1Þ
where a(i) and b(i) are the crack length and uncracked ligament at point (i), respec-
tively, and

PðiÞ þ Pði1Þ vpðiÞ vpði1Þ
(6) ApðiÞ Apði1Þ ¼
2
The stress intensity factor K(i) is given as a function of load P and geometrical pa-
rameters only:
PðiÞ S
KðiÞ ¼ 1 3
ðBBN Þ2 ðW Þ2
1
n h io
2
3 aðiÞ =W 2 1:99 ðaðiÞ =WÞð1 ðaðiÞ =WÞ 2:15 3:93 aðiÞ =W þ 2:7 aðiÞ =W
3
2ð1 þ aðiÞ =WÞ 1 aðiÞ =W 2
(7)
STEP-3. CALCULATION OF NORMALIZED LOAD AND DISPLACEMENT

UP TO THE MAXIMUM LOAD POINT
This section of a load–displacement curve corresponds to the crack blunting regime
or the initial section of a J–R curve with a steep theoretical slope of 2rY. Here the
flow stress rY is defined as the average of yield stress and ultimate tensile stress
measured at the fracture test temperature. As described in the ASTM standard, the
normalized load and displacement (PN(i), v’p(i)) are defined as generalized parame-
ters that are insensitive to, if not independent of, the geometry of test specimen and
given in stress and strain units, respectively. The normalized load and displacement
up to, but not including the maximum load is calculated by [12]:
PðiÞ
(8) PNðiÞ ¼
W aðiÞ 2
WB
W
e
0 vpðiÞ vðiÞ CLLðiÞ PðiÞ
(9) vpðiÞ ¼ ¼
W W
Here, the crack length a(i) is calculated by:
JðiÞ
(10) aðiÞ ¼ a0 þ
2rY
As can be quickly realized in practical calculation, evaluation of these equations

requires solving the circular references among parameters: in the above four
equations, for example, the a(i) needs to be evaluated first to calculate others such as
e
CLLðiÞ (or CLL(i)), and then PN(i). Therefore, we need to adopt an iteration technique
or a simpler technique of using previously calculated values for the (i–1)th point. In
this work, the latter technique is used as the data acquisition interval can be easily
adjusted to be very small (for example, 0.01–0.02 mm interval in total displacement
in this study) so that the error from using the values of the (i–1)th point instead of
those of the (i)th point becomes negligible. It is recommended that the data interval
is smaller than or at least the same as the unloading-reloading cycle interval in the
standard unloading compliance method [12].
STEP-4. CALCULATION OF NORMALIZED LOAD AND DISPLACEMENT

FROM THE MAX-LOAD POINT TO FINAL POINT
Since the crack blunting line (Eq 10) cannot be used, an approximation using a
proper function is needed in this step which represents the major crack growth to
final failure or test stop. The efficiency of the approximation depends on how we
connect the maximum load point (known from Step-3) with the final data
point (known from the measurement of final crack length) in the normalized
0
load–displacement curve (PN(i) – vp(i) curve). A line should be drawn from the final
normalized load–displacement pair tangent to the curve around the maximum load
point calculated in the previous step. The connection with a simple linear line can
provide highly accurate outcome for the J–R curve in most cases, although in the
standard method a normalization function with four fitting coefficients is recom-
mended to be used. A similar non-linear function described by four constants was
suggested by the author and has been used for high strength steels:
h i
PNðmÞ þ x vp0 ðiÞ vpðmÞ
0
(11) PNðiÞ ¼ h i2
1 þ y vp0 ðiÞ vpðmÞ
0
where the point denoted by the subscript m is the maximum load point, and the
variables x and y are changed until the best fit curve is reached. Two criteria used
for searching the best fitting function are (a) if the functional continuity between
the fitting function and the calculated data using the blunting line (J ¼ 2rYDa) is
achieved around the maximum load point, and (b) if the normalized load at final
data point given by the fitting function matches the value calculated from the
measured final crack length. An example displayed for a normalized load versus
displacement curve for a specimen is provided in Fig. 4, which corresponds to the
non-normalized curve given in Fig. 3.
STEP-5. CALCULATION OF CRACK LENGTHS FROM THE MAX-LOAD

POINT TO FINAL POINT
As the full normalized load versus displacement is established, the crack extension
beyond the crack blunting regime can be easily calculated by an inverse function
of Eq 8:
BYUN ET AL., DOI 10.1520/STP157620140017 131
( 1=2 )
PðiÞ 1
(12) aðiÞ ¼ W 1
PN ðiÞ WB
STEP-6. CONSTRUCTION OF J–R CURVE

In the process of calculating crack lengths using normalization, the J–R curve, i.e.,
J–Da curve is simultaneously obtained at each data point since the crack extension
amount Da is simply given by a(i) – a0. The J–R curve obtained for the present
example case is displayed in Fig. 5. The comparison of this J–R curve with that by
unloading compliance method has not been successful because the displacement
measurement using an attached LVDT was not accurate enough to apply the
unloading compliance method to the present reuse specimen. It has been shown,
however, that the two methods produce very close J–R curves for a DCT specimen
[13].
Application Results and Discussion

EXAMPLES OF J–R CURVES AND THEIR CHARACTERISTICS
Some examples of the results of J–R calculation using the procedure described in
the section above are provided in Figs. 6–8, respectively, for room temperature,
300 C, and 500 C tests. Here, a few general characteristics of the small specimen
FIG. 3 A load versus load–line displacement curve for a reused HT9 specimen in
irradiated and annealed condition. This load–displacement data was recorded
using the built-in LVDT within servohydraulic actuator, which contains all
compliance components from load train.
FIG. 4 Normalized load versus plastic displacement curve (PN versus vp/W curve)
constructed from the load–displacement curve in Fig. 3 and crack length data.
The normalized curve from the max load to final data point was described by a
four-parameter Eq 11.
test and analysis techniques are noticed and mentioned first to help correct inter-
pretation of the results. First, the small maximum crack growths of about 1 mm or
less could be obtained from the reused specimens because the nominal initial
uncracked ligament was about 2 mm and, for the ductile cases, the J–R test was
stopped when the applied load passed its maximum and decreased to about one
half of the maximum load. Such small crack growths and resulting short J–R curves
can impose some limitations in application and interpretation as will be discussed
in the last section. Second, as always observed in an indirect crack measurement
method, the normalization method uses an artificial linear line (J ¼ 2rYDa) for
crack blunting regime and thus the transition to the J–R curve portion in the crack
growth regime is often not continuous, which may not be natural. In general, the
discontinuity becomes more evident in lower ductility cases. Third, the slope of the
J–R curve from small specimen is relatively steep (or larger tearing modulus) com-
pared to those of larger specimens. This is because smaller uncracked ligament and
thickness can lead to a relatively larger portion of plasticity zone in the specimen
volume. The tearing modulus should increase with increasing ductility, i.e., at high
temperature or after a complete anneal softening.
Regardless of these complex characteristics originated from the use of minia-
ture specimens, the J–R curves are believed to correctly reflect the effects of irradia-
tion and thermal annealing in HT9 steel. Figures 6(a) and 6(b) display the room
temperature J–R curves for the HT9 steel specimens in different conditions. The
BYUN ET AL., DOI 10.1520/STP157620140017 133
FIG. 5 J-Integral versus Da curve (J-Resistance curve) obtained for the reused HT9
specimen in Fig. 3 using the normalized load–displacement curve and crack
length data. Note that the crack blunting line is given by a theoretical line
(2rYDa).
J–R curve of the nonirradiated HT9 steel shows high J-integral value around crack
initiation and high tearing modulus (slope) in the main crack growth regime. In the
room temperature test after relatively low temperature (380 C) irradiation, a
very short blunting line was obtained from the as-irradiated HT9 steel, as noted in
Fig. 6(a).
The recovery of fracture resistance by both of the thermal anneals, at 550 and
650 C for 2 h, was significant. A full J–R curve could be obtained after 550 C
annealing and only long blunting line obtained after 650 C annealing. In this case,
an over-recovery resulted in large crack blunting followed by general yielding in the
specimen ligament without major crack growth. After the higher temperature
(465 C) irradiation, milder annealing recovery of J–R curve was observed as the
reduction of J-Resistance was less significant when compared to the case of lower
temperature irradiation.
In the 300 C test, the HT9 steel irradiated at 380 C still showed brittle fracture
in linear loading region as found in Fig. 7(a). The annealed specimens did show sig-
nificant recovery and stable crack growth; however, the recovery was incomplete for
both annealing treatments. Figure 7(b) shows that no brittle fracture has occurred in
the steel after higher temperature irradiation, where only full J–R curves are com-
pared and no significant recovery was observed in the annealing treated specimens.
In the 500 C test results displayed in Figs. 8(a) and 8(b), both HT9 specimens in as-
irradiated condition have stable crack growth and thus full J–R curves. The low
FIG. 6 J–R curves at room temperature for the FFTF HT9 steel in nonirradiated, as-
irradiated, and irradiated plus annealed conditions: after (a) relatively low
temperature irradiation and (b) high temperature irradiation. See the inset in (a)
for the curve of as-irradiated material after 380 C irradiation.
temperature irradiation followed by 650 C annealing resulted in more than 100 %

recovery in the J–R curve. As in the 500 C tests, the 550 C annealing did not show
evidence of fracture toughness recovery.
SUMMARY ON THE TEMPERATURE DEPENDENCE OF FRACTURE TOUGHNESS

As the final step of fracture testing and evaluation [8], fracture toughness values
were determined on the J–R curves following ASTM E1820 [12]. The interim frac-
ture toughness values (JQ) were determined at the intersection of the constructed
BYUN ET AL., DOI 10.1520/STP157620140017 135
FIG. 7 J–R curves at 300 C for the FFTF HT9 steel in nonirradiated, as-irradiated, and
irradiated plus annealed conditions: after (a) relatively low temperature
irradiation and (b) high temperature irradiation.
J–R curves and the 0.2 mm offset line of the blunting line (¼2rYDa). The final frac-
ture toughness data were given in the form of stress intensity factor, KJQ, which can
be converted from the JQ data using the following relationship:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(13) KJQ ¼ ðJQ EÞ=ð1 2 Þ
Since it is easily predicted that using the present small specimen technique cannot
avoid multiple violations of the requirements for valid fracture toughness in the
ASTM standard method [12], the fracture toughness data obtained in this work are
expected to stay as interim values. Thus, the dataset might be useful for comparison
FIG. 8 J–R curves at 500 C for the FFTF HT9 steel in nonirradiated, as-irradiated, and
irradiated plus annealed conditions: after (a) relatively low temperature
irradiation and (b) high temperature irradiation.
purposes only or for thin reactor parts such as fuel ducts. In addition, since this ar-
ticle is aimed at a full description of testing and analysis procedure, the detailed dis-
cussion on the material properties before and after irradiation and annealing
treatment is left to a separate publication [7,8]. Therefore, only a small part of the
entire dataset produced through the aforementioned procedure are shown here as
examples: only two extreme cases of seven total datasets for the test temperature
dependence of fracture toughness [6] are discussed below.
The fracture toughness (KJQ) data for the HT9 steel in as-irradiated conditions,
irradiated plus annealed conditions, and nonirradiated condition are plotted in
BYUN ET AL., DOI 10.1520/STP157620140017 137
Figs. 9 and 10 for displaying 550 C annealing effect after high temperature irradia-
tion and for 650 C annealing effect after low temperature irradiation, respectively.
In the complete datasets [8] these two sets represent the least and most recovery
cases, respectively. Forp the pristine HT9 steel, the KJQ value at room temperature
was around 270 MPa m and the fracture toughness showedpa weak tendency for
decreasing with temperature and it remained at >200 MPa m up to 600 C [7].
The data for the as-irradiated conditions and archive material can be found in ear-
lier publications, which include the data from both DCT and TPB specimens [7,11].
Figure 9 illustrates the test temperature dependence of fracture toughness and
annealing recovery behavior after high dose, high temperature irradiation (100.3
dpa, 463.6 C). Since the irradiation temperature was above a critical temperature of
430 C [6,11], below which the loss of toughness due to irradiation becomes sig-
nificant for HT9 steels, and the anneal temperature, 550 C, was within 90 C from
the irradiation temperature, no meaningful annealing recovery of fracture tough-
ness was observed. In the as-irradiated condition, the fracture p toughness of HT9
steel decreased gradually
p with test temperature: from 210 MPa m at room temper-
ature to 134 MPa m at 600 C. Little recovery observed in this case indicated that
relatively stable microstructure features such as precipitation and chemical segrega-
tion have been formed and accumulated during the long term, high temperature
irradiation. Also observed in the temperature dependence is that the trending down
FIG. 9 Effect of 550 C thermal annealing on fracture toughness in HT9 steel irradiated
at about 464 C [8]. The recovery of fracture toughness is not significant over
the entire test temperature.
FIG. 10 Effect of 650 C annealing on fracture toughness in HT9 steel irradiated at

about 380 C [8]. The recovery of fracture toughness is significant at test
temperatures except 600 C, indicating that the large difference between the
irradiation temperature and the annealing temperature resulted in a large
thermal annealing recovery.
slope of fracture toughness became slightly steeper after thermal annealing.

Although no evident thermal recovery was p observed over the test temperature
range, the fracture toughness values >140 MPa m were measured up to the highest
test temperature and no brittle failure was observed among the cases in Fig. 9,
regardless of different material conditions.
Presented in Fig. 10 is the fracture toughness data obtained after more rigorous
annealing at 650 C for 2 h. This case was examined to check if a full recovery in
fracture toughness can be achieved by such an extreme annealing that can induce
excessive softening. Therefore, the case in Fig. 10 represents a complete or excessive
recovery behavior, which can be found only after low temperature irradiation: 18.8
dpa, 379 C. After the 650 C annealing, indeed, a complete or excessive recovery
was observed at all test temperatures except 600 C. At two test temperatures of 379
and 550 C the irradiated HT9 steel actually showed more than 100 % recovery.
This indicates that the 650 C annealing has induced additional microstructural
changes in addition to the thermal annihilation of small radiation-induced defects.
It is suggested that the additional changes include the coarsening of precipitations
and/or annihilation of dislocations, probably enhanced by the changes in radiation
induced defects [8]. It was also noted that a significant drop in the fracture
BYUN ET AL., DOI 10.1520/STP157620140017 139
toughness occurredp between 500 and 600 C, but still a high fracture
toughness >170 MPa m was retained at 600 C after the 650 C annealing.
CONSIDERATIONS FOR FURTHER APPLICATIONS

The main purpose for the development of the small specimen reuse technique was
to apply it to circumstances when the preparation and testing of facture specimens
needs to be performed in a high radiation area, where a simple testing setup and
less instrumentation are always desirable for work efficiency and cost saving. Here,
some positive and negative aspects of the newly developed technique and analysis
procedure are discussed to inform future users of the benefits and limitations of the
aforementioned techniques and procedures.
In specimen preparation, assuming that we have the capacity for machining
miniature specimens from either tested specimens or a piece of irradiated compo-
nents, precracking process is the most challenging step. The displacement signal
recorded and used in this study contains all components from load train including
the crack opening displacement of specimen. In the present procedure, the pre-
cracking technique using variations in the displacement signal from the embedded
gage turned out to be effective in recognizing the initiation of crack growth and to
monitor the crack growth and stop cyclic loading after proper crack growth.
Although the fatigue crack growth amount was small ( <1 mm) and corresponding
changes in displacement readings were also small ( <0.1 mm in average displace-
ment and <10 % in amplitude), producing a fatigue crack with approximate target
length was not difficult and became a routine practice.
Furthermore, fatigue cracking both before and after a J–R fracture testing
turned out to be a convenient method of revealing different tones of crack surfaces
formed by different loading modes. This allowed us to eliminate the traditional heat
tinting to discern crack surface colors. It was also found that any uneven cuts, made
by diamond saw, were automatically corrected to a more balanced crack front dur-
ing fatigue precracking because of the natural force towards evenly distributed
stress concentration over the crack front line. Unbalanced crack lengths were often
corrected as fatigue cracking proceeds, which is because the side with shorter crack
has to sustain higher load, and therefore, crack growth rate is higher at that side.
This mechanism, however, could not help the slower crack growth at surfaces due
to lower constraint.
To conduct J–R facture tests in a controlled environment or at high tempera-
ture in vacuum furnace, simplification in the testing procedure was strongly
required. In the test procedure, instrumentation was therefore simplified by elimi-
nating any externally-attached displacement gage. In the fracture test using
loading-unloading compliance method, accurate reading of displacement on
unloading is critical to calculate crack lengths [12]. In the small specimen testing at
high temperatures, however, the compliance measurements become inaccurate as
the pin-specimen contact surfaces become soft and higher friction at contacts
contribute to the displacement reading. Without an accurate reading of unloading

compliance, the curve normalization method is the only practical method to con-
struct J–R curves from available data. As given by Eq 1, the procedure for calcula-
tion of crack length was modified to accommodate the experimental simplification:
the plastic displacement component, which is needed for calculation of plastic
energy and plastic J component at each step, was obtained by eliminating the whole
elastic component that includes load train flexibilities and by updating compliance
with new crack length. It is an encouraging aspect that these modifications made in
the procedures of fracture testing and J–R curve construction have not incurred any
collateral problems in application.
Since machining side grooves in a miniature specimen is extremely difficult in
the radiation area, even with a modern computerized machine, and the miniature
specimen usually has a low constraint near the specimen side surface, controlling
the shape of crack tip might be a very challenging task. A major drawback of this
small specimen reuse technology is the curvature of the fatigue crack front, which
often results in the violation of an ASTM requirement. The ASTM standard [12]
specifies that the crack length measurements need to be within 10 % deviation from
a straight line for the validity of fracture toughness data. The curved crack line is
mostly due to the difference of stress constraint between near-surface and interior
portion of the specimen. A secondary precracking at low load amplitude after a
major precracking for target crack growth is recommended to minimize the curva-
ture at crack front line. Furthermore, such a low constraint oriented drawback per-
sists in the static J–R testing. It is noted that majority of the toughness data
obtained in this study cannot satisfy the constraint requirements of ASTM E1820
[12]. Finally, the loading span to width ratio (S/W) was only 2.5 (10 mm/4 mm) for
the reused specimens in this study. The ASTM standard method [12] specifies that
the span S needs to be 4 W for the validity of equations, and thus the use of the
short specimens has caused error in the calculated fracture resistance (J) and tough-
ness (KJQ) values. The tearing modulus from the reused specimens, for example,
should be higher than that from longer specimens and may not be used for other
than comparison purposes. This violation due to size limitation may be easily
avoided in other circumstances.
Another aspect worth being noted is that the fracture toughness data deter-
mined following the ASTM procedure [12] provide different meanings for different
specimen sizes. In determining JQ, the 0.2 mm offset of blunting line is used to find
intercept point with the main J–R curve. This may not be appropriate for small
specimens with small crack growth: the typical crack growth amount in the ductile
cases of this study was only about 1 mm and the 0.2 mm offset is 20 % of the whole
crack growth, while the 0.2 mm will be less than 2 % in the standard 1 T CT speci-
men. Furthermore, it can incur too much difference between JQ with 0.2 mm offset
and JQ without offset due to higher tearing modulus in small specimens. The
J-integral value difference between the initiation and the intersection with 0.2 mm
offset line could be as large as 60 % for the 12.5 mm diameter DCT specimens [11].
BYUN ET AL., DOI 10.1520/STP157620140017 141
A detailed guideline needs to be developed for the practice of determining fracture

toughness values using miniature specimens.
Summary
(1) A specimen reuse technique for the tested Charpy specimens has been devel-
oped and applied to the evaluation of FFTF ACO-3 duct HT9 steel. Fracture
properties, J–R curve, and fracture toughness were successfully obtained from
the tested Charpy specimen halves (3.05 by 4 by 13 mm).
(2) Modified procedures for fracture testing and J–R curve construction have
been developed to test miniature specimens in high radiation areas: fracture
testing procedure was simplified by eliminating any externally-attached
displacement gage. In the J-integral evaluation procedure, the load–
displacement curve normalization method for calculation of crack lengths
was modified to accommodate the experimental simplification.
(3) Effects of high dose irradiation and thermal annealing treatment on fracture
behavior have been studied using the new technique. After 550 C annealing,
the recovery of fracture toughness was only partial or negligible, while after
650 C annealing, nearly complete or more than 100 % recovery of fracture
toughness was observed except for the 600 C test case.
No brittle fracture mode waspobserved in the fractography after thermal annealing
at 550 or 650 C ( < 130 MPa m).
ACKNOWLEDGMENTS
This research was part of Fuel Cycle R&D Program/Core Materials sponsored by U.S.
Department of Energy, Office of Nuclear Energy under Contract DE-AC05-
00OR22725 with UT-Battelle, LLC. The writers would like to express special thanks to
Dr. Frank Chen for his technical review and thoughtful comments.
References
[1] Zinkle, S. J. and Ghoniem, N. M., “Prospects for Accelerated Development of High Per-
formance Structural Materials,” J. Nucl. Mater., Vol. 417, 2011, pp. 2–8.
[2] Anderoglu, O., Byun, T. S., Toloczko, M. B., and Maloy, S. A., “Mechanical Performance of
Ferritic Martensitic Steels for High Dose Applications in Advanced Nuclear Reactors,”
Metall. Mater. Trans. A, Vol. 44A, 2013, pp. S70–S83.
[3] Allen, T. R., Busby, J. T., Klueh, R. L., Maloy, S. A., and Toloczko, M. B., “Cladding and Duct
Materials for Advanced Recycle Reactors,” J. Metals, Vol. 60, 2008, pp. 15–23.
[4] Maloy, S. A., Toloczko, M. B., Cole, J., and Byun, T. S., “Core Materials Development for
the Fuel Cycle R&D Program,” J. Nucl. Mater., Vol. 415, 2011, pp. 302–305.
[5] Petroski, R., Cheatham, J., Hejzlar, P., Povirk, G., Schloss, P., and Whitmer, C., “Traveling
Wave Reactor Core Design Using Massively Parallel Precomputation,” Trans. Am. Nucl.
Soc., Vol. 106, 2012, pp. 24–28.
[6] Byun, T. S., Lewis, W. D., Toloczko, M. B., and Maloy, S. A., “Impact Properties of Irradi-
ated HT9 from the Fuel Duct of FFTF,” J. Nucl. Mater., Vol. 421, 2012, pp. 104–111.
[7] Baek, J. H., Byun, T. S., Maloy, S. A., and Toloczko, M. B., “Investigation of Temperature
Dependence of Fracture Toughness in High-Dose HT9 Steel Using Small-Specimen
Reuse Technique,” J. Nucl. Mater., Vol. 444, 2014, pp. 206–213.
[8] Byun, T. S., Baek, J. H., Anderoglu, O., Maloy, S. A., and Toloczko, M. B., “Thermal
Annealing Recovery of Fracture Toughness in HT9 Steel after Irradiation to High Doses,”
J. Nucl. Mater., Vol. 449, Nos. 1–3, 2014, pp. 263–272.
[9] Sencer, B. H., Kennedy, J. R., Cole, J. I., Maloy, S. A., and Garner, F. A., “Microstructural
Analysis of an HT9 Fuel Assembly Duct Irradiated in FFTF to 155 dpa at 443 C,” J. Nucl.
Mater., Vol. 393, 2009, pp. 235–241.
[10] Toloczko, M. B., “Inventory Overview and Irradiation Conditions for the Fast Flux Test Fa-
cility (FFTF) Liquid Metal Reactor (LMR) Ferritic/Martensitic Specimens and the FFTF
ACO-3 HT-9 Duct,” PNNL-16666, Pacific Northwest National Laboratory, Richland, WA,
2007.
[11] Byun, T. S., Toloczko, M. B., Saleh, T. A., and Maloy, S. A., “Irradiation Dose and Tempera-
ture Dependence of Fracture Toughness in High Dose HT9 Steel from the Fuel Duct of
FFTF,” J. Nucl. Mater., Vol. 432, 2013, pp. 1–8.
[13] Byun, T. S., “Testing and Evaluation Techniques for Miniature Fracture Specimens in High
Radiation Area,” ORNL/TM-2013/199, Oak Ridge National Laboratory, Oak Ridge, TN,
2013.
SMALL PUNCH TESTING
Karel Matocha1
Small-Punch Testing for Tensile

and Fracture Behavior:
Experiences and Way Forward
Reference
Matocha, Karel, “Small-Punch Testing for Tensile and Fracture Behavior: Experiences and Way
Forward,” Small Specimen Test Techniques: 6th Volume, STP 1576, Mikhail A. Sokolov and
Enrico Lucon, Eds., pp. 145–159, doi:10.1520/STP157620140005, ASTM International, West
Conshohocken, PA 2015.2
ABSTRACT
The present paper summarizes the procedures described in CWA 15627 [CEN
Workshop Agreement, “Small Punch Test Method for Metallic Materials,” CWA
15627:2007 D/E/F, European Committee for Standardization, Brussels, Belgium,
2007] for determination of tensile and fracture characteristics of metallic
materials from the results of small-punch tests together with the corrections of
this document proposed in the frame of its prospective conversion into a
European standard. The corrections were proposed on the basis of the
experiences obtained in the period 2007–2012 at Material and Metallurgical
Research, Ltd. (Ostrava, Czech Republic).
Keywords
small-punch test, FATT, load-deflection curve, small-punch fracture energy ESP
Introduction
The effort to extend the design lifetime of industrial plants operating for a long
time at elevated temperatures requires the knowledge of the residual lifetime of
the critical components. Residual lifetime assessment is unthinkable without the
knowledge of:
5, 2015.
1
Material and Metallurgical Research, Ltd., Ostrava, Czech Republic.
2
Houston, TX.
Copyright V
1. mechanical properties of materials prior to operation, respecting all technolog-

ical operations realized throughout the manufacturing of the component, and
2. mechanical properties after actual time of operation (actual mechanical prop-
erties), because the material properties can be reduced throughout the service
life by aging, service loading and temper, and hydrogen and/or radiation
embrittlement.
The conventional mechanical tests require relatively large volumes of testing
material, and extracting it from an operating component can impair its integrity. In
such situations, mechanical test techniques based on small specimen techniques are
promising for characterizing the mechanical properties of components. The small
specimen test techniques can be generally divided into two categories:
1. reduction of test specimen size proportionate to its conventional equivalent
while retaining the specimen geometry and loading configuration, and
2. development of novel specimen designs and loading configurations that has
no conventional equivalent.
Among these innovative techniques, a technique called the small-punch (SP)
test has emerged as a promising candidate [1,2]. It can be used to obtain tensile,
fracture, and creep data from very small quantities of experimental material. Major
benefits of this method are:
1. It can be applied as a virtually non-destructive tool to monitor in-service com-
ponents in industrial plants.
2. It enables determination of the tensile, fracture, and creep characteristics of
materials at critical locations of the components.
3. The test itself is rather simple to perform and inexpensive.
4. It gives the possibility to study locations such as interfaces, coatings, welded
joints [base metal (BM), heat-affected zone (HAZ), and weld metal (WM)]
and exotic materials, e.g., anisotropic.
The use of this method was accelerated by the development of scoop cutter
sampling technology and/or electro-discharge machining technology for obtaining
and retrieving a sample of material for analysis in a virtually non-destructive man-
ner not requiring post-sampling repair. Material and Metallurgical Research, Ltd.
carries out sampling by two SSam-2 scoop sampling machines [3] (see Fig. 1).
It removes a small button-shaped piece of material that is typically 25 mm in
diameter and 4 mm in thickness.
Code of Practice for Small-Punch Testing

The industrial use of the small-punch test technique has been limited in Europe
because of lack of standard codes for SP testing and sampling. The European Com-
mittee for Standardisation (CEN) provides two standardization products—a formal
standard and a workshop agreement [4]. Given the state of knowledge of the SP test
and the need to quickly develop a set of guidelines to encourage uniformity in test-
ing, the workshop agreement path was chosen. In the September 2004 CEN Work-
shop 21, a small-punch test method for metallic materials was established by CEN.
MATOCHA, DOI 10.1520/STP157620140005 147
FIG. 1 SSam-2 scoop sampling machine.
On the basis of the technical consensus of 32 participating organizations (without

geographic restriction), CWA 15627, Small-Punch Test Method for Metallic Mate-
rials, was issued in December 2007. It is comprised of two parts: Part A: A Code
of Practice for Small-Punch Creep Testing, and Part B: A Code of Practice for
Small-Punch Testing for Tensile and Fracture Behaviour. ANNEX B1, “Derivation
of Tensile and Fracture Material Properties,” describes the methods for estimation
of yield and tensile strength, ductile-brittle transition temperature (DBTT), and
fracture toughness of the metallic materials from SP test records. ANNEX B2,
“Guidance on Relevant Technological Issues: Specimens Sampling from
Components,” describes potential applications of the SP test.
The Code of Practice for Small-Punch Testing

for Tensile and Fracture Behavior
The main elements of the code cover the apparatus, the test specimen preparation,
the test temperature considerations, the test procedure, the post-test examination,
and the approaches to the derivation of yield strength (YS), ultimate tensile strength
(UTS), fracture appearance transition temperature (FATT), and fracture toughness
from SP tests results.
The code takes into consideration both the small-punch bulge (disc clamped
peripherally between two dies) and the small-punch drawing test (disc simply
supported). Disc-shaped test specimens with d ¼ 8 mm in diameter and initial
thickness of h0 ¼ 0.5 mm are recommended for both the SP bulge and SP drawing
FIG. 2 Load-punch–displacement curve recorded during a small-punch test of a ductile

material.
tests. Specimen thickness of 0.5 mm assures that the number of grains through the
thickness is adequate to permit bulk properties to be obtained.
The objective of the test is to produce a load-punch–displacement and/or
load-specimen-deflection record (see Fig. 2) under crosshead control, which
contains information about the elastic–plastic deformation and strength properties
of the material. A displacement rate of the punch in the range between 0.2 to
2 mm/min is recommended [5].
The following SP-related parameters are used for determination of YS, UTS,
FATT, and fracture toughness JIC from such a load/punch displacement curve:
Fm (N) ¼ maximum load recorded during SP test,
Fe (N) ¼ load characterizing the transition from linearity to the stage associated
with the spread of the yield zone through the specimen thickness,
um (mm) ¼ displacement corresponding to the maximum load Fm,
uf (mm) ¼ displacement corresponding to load Ff ¼ 0.8 Fm, and
ESP (J) ¼ SP fracture energy obtained from the area under the load-
punch–displacement curve up to fracture (up to uf).
Figure 3 shows the comparison of load-punch–displacement and
load-specimen-deflection curves produced simultaneously during the SP test. Load
Fe and displacement um obtained from these curves are significantly different. The
difference between these two curves depends on the stiffness of the testing machine
and the testing rig.
The load-punch–displacement curves obtained can be analyzed either in terms
of elastic–plastic finite element method or they can be utilized to derive empirical
correlations between SP and standardized test results. Both of the above-mentioned
approaches are used at the present time for determination of YS, UTS, and JIC.
However, DBTT, measured by FATT and/or 41 J absorbed energy transition
temperature (TT) is determined only using empirical correlations.
MATOCHA, DOI 10.1520/STP157620140005 149
FIG. 3 Load-punch–displacement; load-specimen-deflection record.
Determination of Actual Yield Strength,

Tensile Strength and FATT on the Basis
of Empirical Correlations
Figure 4 shows the effect of disc specimen thickness on the dependence of load Fe
on yield strength, determined according to the code. The effect of the disc specimen
thickness can be eliminated by normalizing Fe as Fe =h20 , where h0 is the initial thick-
ness of the disc (see Fig. 5). Figure 6 shows the effect of disc specimen thickness on
FIG. 4 The effect of disc specimen thickness on empirical correlation of Fe versus yield
strength.
FIG. 5 Dependence of Fe/h02 versus yield strength independent on disc thickness.
dependence of load Fm on tensile strength. The effect of disc specimen thickness

can be eliminated in this case by normalizing Fm as Fm/(um, h0) (see Fig. 7).
However, the empirical correlations for determination of tensile strength
mostly show significantly lower scatter band in comparison with correlations for
yield strength determination (see Figs. 5 and 7) [6]. The use of the offset method
(offset displacement of 0.1 mm) for determination of Fe, instead of the procedure
described in the code, resulted in significant lowering of scatter band of the
empirical correlation (see Figs. 8 and 9). In this case, Fe represents the point in the
load-punch–displacement curve where fully plastic behavior takes over from mixed
FIG. 6 The effect of disc specimen thickness on empirical correlation of Fm versus

tensile strength.
MATOCHA, DOI 10.1520/STP157620140005 151
FIG. 7 Dependence of Fm/(um h0) versus tensile strength independent on disc

thickness.
elastic–plastic behavior (see Fig. 10). However, further study is needed to verify
the sensitivity of empirical correlations in detecting and evaluating degradation
corresponding to various degradation modes.
Empirical correlations for FATT estimation are based on the fact that steels
exhibiting standard Charpy impact ductile to brittle transition behavior also show
the transition behavior in the temperature dependence of fracture energy of pene-
tration test, but shifted to a lower temperature. Fracture energy of penetration test
is calculated from the area under the force-punch displacement until the specimen
failure (punch displacement of uf) [5]. According to CWA 15627, transition
temperature TSP is defined as the temperature corresponding to half of the sum of
FIG. 8 Empirical correlation YS versus Fe =h20 in accordance with CWA 15627.

FIG. 9 Empirical correlation YS versus Fe =h20 for Fe ¼ Fu ¼ 0.1 mm.
the maximum and minimum fracture energy calculated by the method of least
squares from the temperature dependence of fracture energy calculated from the
experimentally measured data in transition areas.
Transition temperature TSP determined according to the code is correlated with
the FATT determined from a series of Charpy V impact tests in the form
FATT ¼ a TSP or FATT ¼ a TSP þ b
where:
a, b ¼ constants (see Fig. 11).
FIG. 10 Procedure for determination of Fe by offset method (offset displacement of

0.1 mm).
MATOCHA, DOI 10.1520/STP157620140005 153
FIG. 11 Empirical correlation between FATT and TSP for 14MoV6-3 steel.
Factors Affecting the Determination of TSP

According to Ref 5 (Section B1.3, Embrittlement-Transition Temperature), the SP
test orientation is such that the normal to the punch disc plane is parallel to Charpy
specimen crack-propagation directions (see Fig. 12).
Figure 13 shows the effect of orientation of the disc on transition temperature
TSP determined from the temperature dependence of fracture energy ESP for discs
FIG. 12 Orientation of disc test specimen in accordance with CWA 15627.

FIG. 13 Effect of orientation of test disc on transition temperature TSP determined for
material of the pipe ø219 16 mm made of 14MoV6-3 steel after exposition of
256|000 h at 510 C. Crosshead speed 1.5 mm/min.
taken in the direction T-L (in accordance with CWA 15627) and discs taken away
from a pipe ø219 16 mm made of 14MoV6-3 steel by SSam-2 (R-L orientation)
[6]. The results of penetration tests indicate significant effect of test specimen
orientation on the temperature dependence of fracture energy mainly in the transi-
tion area.
Figure 14 shows the effect of punch diameter (2.0 mm, 2.5 mm) on the tempera-
ture dependence of fracture energy of the pipe made of 14MoV6-3 steel in the
as-received state. The results show that in the transition region both the tempera-
ture dependence of fracture energy and transition temperature TSP is not affected
significantly by punch diameter.
Figure 15 shows the temperature dependence of fracture energy obtained for the
cast plate made of steel P91. It is evident that temperature dependence of fracture
energy in the transition region (red and green dots) shows considerable scatter,
which may be attributable to the local inhomogeneity of the structure of the casting.
The results have been described with two temperature dependencies with signifi-
cantly different transition temperatures TSP.
The principal difference between the SP testing technique and standardized
impact testing lies in the fact that the penetration tests carried out in accordance
with CWA 15627 use disc-shaped test specimens without a notch [7–9]. Especially
in tough materials, the temperature dependence of fracture energy in the transition
MATOCHA, DOI 10.1520/STP157620140005 155
FIG. 14 Effect of punch diameter on temperature dependence of fracture energy ESP.

Crosshead speed 1.5 mm/min.
area is very steep and lies close to the temperature of liquid nitrogen. The procedure
recommended in the CWA for the determination of TSP can, in this case, lead to
significant errors in its determination. Efforts to move the transition area of pene-
tration testing closer to the transition area of standardized impact tests led to the
proposal of the disc specimen with a “U”-shaped notch in the axis plane of the disc
FIG. 15 Temperature dependence of fracture energy for casting made of P91 steel.
Crosshead speed 1.5 mm/min.
FIG. 16 Disc test specimen with “U” notch in the axis of disc plane.
(see Fig. 16). Initial results indicate that the use of the specimen with a notch shifts
significantly the transition temperature TSP to higher temperatures, and the
temperature dependence of fracture energy is also less steep (see Fig. 17). The empir-
ical correlations between TSP and FATT can be then affected by alloy type, punch
velocity, disc specimen orientation, and use of notched disc specimens.
When the penetration tests are carried out at low temperatures, sudden drops
in force are occasionally recorded on the force-punch displacement record (see
Fig. 18). By scanning the sample surface during the penetration test using a camera,
it has been shown that the occurrence of the first force drop is a result of the
FIG. 17 The effect of the notch on temperature dependence of fracture energy. Pipe
made of 14MoV6-3 steel. Crosshead speed 1.5 mm/min.
MATOCHA, DOI 10.1520/STP157620140005 157
FIG. 18 Load drop associated with the initiation of the first circumferential crack. SP
test at 140 C manufactured of 14MoV6-3 steel.
FIG. 19 The comparison of small-punch transition temperature TSP determined in

accordance with CWA 15627 and from the energy for the initiation of the first
crack.
initiation of the first circuit cracks. In such cases, the fracture energy should be
calculated as the area under the force-punch displacement until the first crack ini-
tiation. Figure 19 shows the comparison of small-punch transition temperature TSP
determined in accordance with CWA 15627 and from the energy for the initiation
of the first crack.
Conclusions
• The empirical correlation for determination of material yield strength from
the results of SP tests has been expressed as a dependence of yield strength on
Fe/(h0) [2].
• The empirical correlation for estimation of material tensile strength from the
results of SP tests has been expressed as a dependence of tensile strength on
Fm/(um h0).
• The use of the offset method (offset displacement of 0.1 mm) for determina-
tion of Fe, instead of the procedure described in the code, resulted in signifi-
cant lowering of scatter band of the empirical correlation.
• The orientation of the SP test specimen can affect significantly the empirical
correlation between FATT, measured on standard Charpy V-notch specimens,
and transition temperature TSP determined on SP disc test specimens.
• The orientation of SP test specimen described in CWA 15627 is not suitable for
the estimation of actual values of FATT of in-service components materials.
• The scatter of temperature dependence of fracture energy in the transition region
can be affected significantly by local inhomogeneity of the tested material.
• The significant shift in transition temperature TSP was observed as a result of
the use of a U-shaped 0.2-mm-deep notch in the axis of the disc plane.
• If multiple load drops are observed on a load-punch–displacement curve,
the fracture energy should be calculated as the area under the
load-punch–displacement curve up to first load drop.
• Further study is needed to verify the sensitivity of empirical correlations in
detecting and evaluating degradation corresponding to various material
degradation modes.
ACKNOWLEDGMENTS
This paper was created as part of project No. CZ.1.05/2.1.00/01.0040, Regional
Materials Science and Technology Centre, within the framework of the operation pro-
gramme, Research and Development for Innovations, financed by structural funds
and from the state budget of the Czech Republic.
References
[1] Lucas, G. E., “Review of Small Specimen Test Techniques for Irradiation Testing,” Metall.
Trans. A, Vol. 21, No. 5, 1990, pp. 1105–1119.
MATOCHA, DOI 10.1520/STP157620140005 159
[2] Bicego, V. and Lohr, R. D., “Mechanical Testing on Miniature Specimens by the Small
Punch Method,” Proceedings of the International Symposium on Materials Ageing and
Life Management, Kalpakkam, India, Oct 3–6, 2000, pp. 1–10.
[3] Hurst, R. and Matocha, K., “The European Code of Practice for Small Punch Testing—
Where Do We Go from Here?” Proceedings of the 1st International Conference SSTT on
Determination of Mechanical Properties of Materials by Small Punch and Other Miniature
Testing Techniques, Ostrava, Czech Republic, Aug 31–Sept 2, 2010, pp. 5–11.
[4] Hurst, R. and Matocha, K., “Where Are We Now With the European Code of Practice for
Small Punch Testing?” Proceedings of the 2nd International Conference on Determina-
tion of Mechanical Properties of Materials by Small Punch and Other Miniature Testing
Techniques, Ostrava, Czech Republic, Oct 2–4, 2012, pp. 4–18.
[5] Matocha, K., “The Evaluation of Mechanical Properties of Structural Steels by Penetra-
tion Tests,” Technical University Ostrava and Material and Metallurgical Research, Ltd.,
Ostrava, Czech Republic (in Czech), 2010.
[6] Matocha, K., “Determination of Actual Tensile and Fracture Characteristics of Critical
Components of Industrial Plants Under Long Term Operation by SPT,” Proceedings of
the ASME 2012 Pressure and Piping Division Conference PVP 2012, Toronto, Ontario,
Canada, July 15–19, 2012.
[7] CEN Workshop Agreement, “Small-Punch Test Method for Metallic Materials,” CWA
15627:2007 D/E/F, European Committee for Standardization, Brussels, Belgium, 2007.
[8] Matocha, K., “Determination of Actual Tensile and Fracture Characteristics of Critical
Components of Industrial Plants Under Long Term Operations—Experiences and Way
Forward,” Two Day Pre-SMIRT Seminar on Miniature Specimens for Evaluation of
Mechanical Properties of Structural Materials (MEMP 2011), Mumbai, India, Nov 4–5, 2011.
[9] Turba, K., Hurst, R. C., and Hähner, P., “Small Punch Fracture Testing Applied to the
NESC-I Spinning Cylinder Material,” Proceedings of the 2nd International Conference on
Determination of Mechanical Properties of Materials by Small Punch and Other Miniature
Testing Techniques, Ostrava, Czech Republic, Oct 2–4, 2012, pp. 196–202.
R. Kopriva,1 M. Brumovsky,1 M. Kytka,1 M. Lasan,2 J. Siegl,3

and K. Matocha4
Application of Miniature
Small Punch Test Specimen
in Determination of
Tensile Properties
Reference
Kopriva, R., Brumovsky, M., Kytka, M., Lasan, M., Siegl, J., and Matocha, K., “Application of
Miniature Small Punch Test Specimen in Determination of Tensile Properties,” Small Specimen
Test Techniques: 6th Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 160–
167, doi:10.1520/STP157620140015, ASTM International, West Conshohocken, PA 2015.5
ABSTRACT
Nuclear reactors and related components must comply with high standards of
reliability and safety in order to operate. The contemporary practice of operating
life extensions and reactor power upgrades constitutes new challenges in the
process of ensuring safe long-term operation. The primary input data for safety
analysis were obtained by means of in-service inspections and surveillance
programs. The limitations of surveillance programs, necessitate the need to
develop more progressive mechanical testing methods. This work focused on the
use of small punch tests for the determination of tensile properties of VVER
reactor type materials.
Keywords
small punch test, mechanical testing, component integrity, hot cell facility,
nuclear reactors, plant life extension
Manuscript received February 10, 2014; accepted for publication April 8, 2015; published online May 5, 2015.
1
Integrity and Technical Engineering Division, UJV Rez, a.s., Hlavni 130, Rez, 25068 Husinec, Czech Republic.
2
Research Centre Rez, Hlavni 130, Rez, 25068 Husinec, Czech Republic.
3
Faculty of Nuclear Sciences and Physical Engineering, Department of Materials, Czech Technical Univ. in
Prague, Trojanova 13, 12000, Prague 2, Czech Republic.
4
Material & Metallurgical Research Ltd., Pohraničnı́ 693/31, Vı́tkovice, 703 00 Ostrava, Czech Republic.
5
Houston, TX.
Copyright V
KOPRIVA ET AL., DOI 10.1520/STP157620140015 161
Introduction
The structural integrity of the reactor pressure vessel (RPV) and related piping
of the nuclear power plant (NPP) is under close scrutiny. The RPV is considered
irreplaceable and serves as one of the main barriers of radioactivity in the nuclear
reactor. To ensure their safe operation, the RPV and the piping is subjected to
in-service inspections and failure analysis. The input data for the analysis originates
from RPV surveillance programs, consisting of specimens for mechanical testing
irradiated for a specific time and subsequently tested in a hot cell facility. The oper-
ating life extensions and power upgrades result in the need for further material vali-
dation beyond the scope of the original surveillance programs. The lack of the
original test material, combined with the necessity of testing materials not included
in the original surveillance programs, calls for the use of progressive mechanical
testing methods [1,2]. The small punch test (SPT), requires only a small amount of
test material and compensates for the limited amount of archive material available,
providing the option of direct material sampling from the component without com-
promising its structural integrity or requiring extensive repairs.
Experimental Method
The principle of the small punch test is the penetration of the small clamped planar
specimen by a hemispherical surface. After the initial development phase, recent
years were marked with efforts to broaden the range of the SPT applicability. The
method has the potential of providing information about materials available in lim-
ited quantities. These are the cases of weld materials, heat affected zones, thin layer
of RPV cladding, fuel cladding, turbine blades, and generally materials from the
past, where the amount of available test material is not sufficient for standard size
specimens. In the case of radioactive specimens testing, small size specimens have
the benefit of lower radioactivity and dose rate.
For the following studies, test specimens with dimensions Ø8 by 0.5 mm were
used. The studies were performed at three separate laboratories at UJV Rez, the
Czech Technical University, and Material & Metallurgical Research Ltd. The frame
for the methodology is described in Ref. [3].
Materials
To obtain a wide range of test values, multiple materials were used in combination
to investigate the dependence of the tensile properties correlations on the specific
materials. The following selection of materials was used:
• 15Kh2MFA: a carbon steel used as the base metal of reactor pressure vessels of
VVER-440 nuclear reactors
• 08Kh18N10T: a stainless steel used for the internals of VVER-440 nuclear
reactors
• 10GN2MFA: a carbon steel used for the primary collector of steam generators
of VVER-1000 nuclear reactors
162
TABLE 1 Chemical composition of the tested materials.
C Mn Si P S Ni Cr Mo V Ti Cu
15Kh2MFA 0.13 0.30 0.17 – – – 2.50 0.60 0.25 – –

0.18 0.60 0.37 0.025 0.025 0.40 3.00 0.80 0.35 – –
08Kh18N10T – – – – – 9.50 17.0 – – – –
0.10 2.00 1.00 0.045 0.030 12.00 19.0 – – – –
10GN2MFA 0.08 0.80 0.17 – – 1.80 – 0.40 0.03 – –
0.12 1.10 0.37 0.008 0.005 2.30 0.30 0.70 0.07 0.015 0.30
A533B (JRQ) 0.16 1.33 0.17 0.019 0.001 0.65 0.10 0.47 – – 0.13
0.18 1.35 0.25 0.021 0.003 0.75 0.12 0.51 – – 0.17
• A533B (JRQ): an International Atomic Energy Agency (IAEA) reference mate-

rial and a carbon steel
The chemical composition of the tested materials is provided in Table 1. The
15Kh2MFA steel and A533B steel were tested at a range of temperatures from
196 to 300 C. The steels 08Kh18N10T and 10GN2MFA were tested at laboratory
temperature. The 08Kh18N10T steel was subjected to 5, 10, and 15 % levels of plas-
tic deformation before the test specimens were manufactured, the 10GN2MFA steel
was subjected to 3 different levels of heat treatment.
The test specimens were cut using electrical discharge machining to give disc
sizes of Ø8 by 0.6 mm. To achieve the desired thickness of 0.5 6 0.005 mm, the
specimens were grinded and polished.
TESTING EQUIPMENT
The schematics of the test setup, which are shown in Fig. 1, correspond to the test
setup in all three of the participating laboratories. A puncher with r ¼ 1.25 mm
radius and receiving hole with d ¼ 4 mm diameter were used in all test setups. The
tests were performed on electromechanical universal testing systems, which were
FIG. 1 Small punch test setup: (1) test specimen, (2) puncher, (3) receiving die, (4)
clamping die, and (5) extensometer.
fitted with environmental chambers (UJV Rez and Material & Metallurgical
Research Ltd) for high and low temperature testing. Linear variable differential
transformer extensometers were used to measure the deflection of the test speci-
mens. A total number of 48 data points were obtained, consisting of averaged
results from the three separate laboratories.
TEST PROCEDURE
The tests were undertaken in air with the specimens being fully clamped (bulge test
conditions). The displacement rate of the puncher during the test was 0.3 mm/min.
The loading of the specimen proceeded from the start of the test up until the point
of maximum load and then continued to a 20 % drop of the maximum load. This
drop occurred suddenly in the case of brittle fracture; although it was observed
continuously in the case of ductile fracture.
During the test, the load and the displacement of the specimen tip, measured
by an extensometer, were continuously recorded to obtain a load displacement dia-
gram as seen below in Fig. 2. For the tensile properties determination, the following
values were recorded:
• Fe (N): elastic–plastic transition load
• Fm (N): maximum load recorded
• um (mm): displacement of the specimen tip at the maximum load
• ue (mm): displacement of the specimen tip at the elastic–plastic transition load
ue was determined by a bilinear fit function of an exponential curve. The bilin-

ear function is shown in Eq 1, where uB ¼ h0 ; fB ¼ Fðh0 Þ, and the parameter
FIG. 2 Test evaluation.

FIG. 3 Fe determination by a bilinear fit function.
uA ue was subjected to the least square fit according to Eq 2. The process of

determining Fe is shown in Figs. 2 and 3: correlation of the ultimate tensile strength.
8
> fA
>
< u for 0 u uA
uA
(1) f ðu Þ ¼
>
> f fA
: B ðu uA Þ þ fA for uA u uB
uB uA
ð uB
(2) ½F ðuÞ f ðuÞ2 du ! Min
0
TEST RESULTS
The obtained results from SPT were normalized using the original thickness of the
test specimen h0 and displacement of the specimen tip at the maximum load um.
The normalized results were correlated with the results of tensile tests performed
according to Ref. [4]. The results for the yield strength are displayed in Fig. 4, and
the results for the ultimate tensile strength are displayed in Fig. 5. The correlation
was performed using linear regression without offset with the least squares crite-
rion. The correlation formula for determining the yield strength from the SPTs is
presented in Eq 3 and the correlation formula for determining the ultimate tensile
strength from SPT is presented in Eq 4. Subsequently, the standard deviation for
both values was calculated. The results are given in Table 2.
FIG. 4 Correlation of the yield strength.
Fe
(3) Rp0:2 ¼ b
h20
Fm
(4) Rm ¼ b
um h0
FIG. 5 Correlation of the ultimate tensile strength.

TABLE 2 Results.
Rp0.2 Rm
Correlation coefficient b 0.5084 0.2655

Standard deviation 651.8 MPa 666.4 MPa
Conclusions
Using four different materials in various states and at a range of temperatures, the
correlations for the tensile properties determination of unirradiated RPV and pip-
ing materials were developed. Contrary to the practice of developing separate corre-
lations for each material, the correlations represent a reasonably good fit. Although
the value of the standard deviation is higher than encountered in the standard ten-
sile test, in cases where tensile tests cannot be performed for any reason, small
punch tests can provide valuable information about the mechanical properties. This
study demonstrated the potential for developing material-independent SPT tensile
correlations. Further investigation, requiring a broad range of test data, is necessary
to validate this theory.
ACKNOWLEDGMENTS
This work is supported by the Technology Agency of the Czech Republic within the
frame of project TA02020811.
References
[1] International Atomic Energy Agency, “Reference Manual on the IAEA JRQ Correlation
Monitor Steel for Irradiation Damage Studies,” IAEA-TECDOC-1230, International Atomic
Energy Agency, Vienna, Austria, 2001.
[2] International Atomic Energy Agency, “Assessment and Management of Ageing of Major
Nuclear Power Plant Components Important to Safety: PWR Pressure Vessels,” Technical
Report IAEA-TECDOC-1120, International Atomic Energy Agency, Vienna, Austria, 1999.
[3] CWA 15627:2006 E: Small Punch Test Method for Metallic Materials, European Commit-
tee for Standardization, Brussels, Belgium, 2006.
[4] ASTM E8/E8M-13a: Standard Test Methods for Tension Testing of Metallic Materials,
Raghu V. Prakash1 and S. Arunkumar2
Evaluation of Damage in
Materials Due to Fatigue Cycling
Through Static and Cyclic Small
Punch Testing
Reference
Prakash, Raghu V. and Arunkumar, S., “Evaluation of Damage in Materials Due to Fatigue
Cycling Through Static and Cyclic Small Punch Testing,” Small Specimen Test Techniques: 6th
Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 168–186, doi:10.1520/
ABSTRACT
Results of an experimental investigation of damage due to prior fatigue of metallic
materials through static and cyclic Small Punch Test (SPT) is presented in this
paper. Monotonic small punch tests were performed on virgin specimens of copper,
brass, and stainless steel (SS-304) as well as on fatigue damaged copper samples
to understand the effect of prior damage on tensile properties. It is observed that,
the slope of initial linear portion of load–displacement response when normalized
by initial thickness of the SPT specimen is proportional to the elastic modulus.
Damage fraction estimated based on conventional damage theories was found to
correlate reasonably with the data derived from static small punch testing. Fatigue
response of materials was evaluated through a novel cyclic small punch testing
under global compression–compression loading. Variation of punch displacement
and hysteresis energy was monitored as a function of the number of cycles.
Distinct change in punch displacement, cumulative hysteresis energy, was
observed prior to complete failure of specimen. An empirical equation relating to
the level of damage and hysteresis energy during cyclic SPT is proposed.
Manuscript received February 5, 2014; accepted for publication August 6, 2014; published online September
26, 2014.
1
Machine Design Section, Dept. of Mechanical Engineering, Indian Institute of Technology Madras, Chennai-
600 036, India.
2
Machine Design Section, Dept. of Mechanical Engineering, Indian Institute of Technology Madras, Chennai-
600 036, India (Corresponding author), e-mail: raghuprakash@iitm.ac.in
3
Houston, TX.
Copyright V
PRAKASH AND ARUNKUMAR, DOI 10.1520/STP157620140011 169
Keywords
cyclic small punch test, fatigue damage, hysteresis energy, copper, fracture
Introduction
The small specimen test technique is a promising method for assessing the mechan-
ical properties locally at critical locations of in-service components. These small
specimen test methods are useful in life prediction and component integrity assess-
ment [1]. As the specimen size become smaller, the mechanical response of materi-
als to loading becomes difficult to evaluate and the results show considerable scatter
due to variation in defect densities in the limited volume chosen for investigation.
Thus, a pre-requisite for using small specimen test technique is to establish a corre-
lation between small specimen test results and traditional full scale tests; several
researchers globally are focusing on this area.
Of the handful of testing techniques available for testing at small volume level,
automated ball indentation (ABI) and small punch test (SPT) are used extensively in
the recent years in view of their capability to predict the post-yield hardening
response of materials. ABI has been successfully used to characterize the mechanical
properties (static and dynamic) of bulk, thin films, functionally graded, and prior
damaged materials through static and cyclic indentation [2–13]. This technique is
primarily an in situ method suitable for thick specimens, but not viable for thin sam-
ples (less than 1 mm thick which have been extracted from components using boat
sampling methods), as the deformation beneath the indenter affects the test results.
Manahan developed a Miniature Disk Bend Test (MDBT) using disk of the size
comparable to TEM grid specimens (3 mm in diameter). Numerical techniques
were used to understand the strain hardening response, creep response, and frac-
ture ductility from load–displacement response during bend testing of materials.
Analytical expressions have been derived for determining the stress–strain response
of materials from bending of a thin specimen; however, its applicability for materi-
als after-damage is not clearly established.
In case of thin specimens scooped from components, small punch test (SPT)
seems to be a promising technique as specimens as small as 3 mm (similar to Trans-
mission Electron Microscopy (TEM) grids) is sufficient for estimating the (local)
mechanical properties [14]. The primary difference between MDBT specimens and
SPT specimens is the level of constraint during bending of specimens (free bending
versus constrained bending). Based on several studies on SPT in Europe and Japan,
empirical correlations between yield and ultimate tensile strength of a material
derived from SPT and conventional large specimen test method have been proposed
[15,16].
Several other researchers arrived at improved analytical expressions based on
small deflection conditions, classical plate bending theory, and sheet stretching
theory to determine the tensile properties [17–19]. Isselin and Shoji [20] developed
a new method based on the elastic deformation energy to estimate yield strength
from SPT, with an objective of reducing the scatter in the result and for the better
evaluation of the yield strength. It was also claimed that this energy method is supe-
rior to the two-tangent and offset method of estimating the yield strength.
Bayoumi and Bassim [21] proposed a theoretical model relating fracture tough-
ness and ductility parameter. The fracture toughness was proportional to the square
of the ductility parameter in the linear-elastic region, while it varied linearly with
the ductility parameter in the elastic-plastic region. However, these models use con-
stants which have to be derived experimentally. Mao and Takahashi [15,22]
observed a nearly linear relationship between equivalent fracture strain obtained
under biaxial stress state and ductile fracture toughness (JIC). In order to estimate
brittle fracture toughness (KIC), KIC was related to fracture stress (rf(SP)) empiri-
cally, as it controls the failure of the brittle materials. Baik et al. [23] demonstrated
the capability of SPT to measure DBTT of tempered and embrittled steels. This pro-
vided a method for several researchers to use SPT for extracting DBTT and they
found similar linear correlation between SPT results and DBTT [16,24].
A number of researchers have obtained information about creep properties
[25–30] and stress corrosion cracking susceptibility [31] through SPT. Thus, mono-
tonic SPT load-deflection response has been successfully deployed to extract various
mechanical properties of materials. However, the test technique has few limitations:
(i) the correlations developed are dependent on material, environment, and applica-
ble only to a class of alloy for which they have been developed; (ii) empiricism and
uncertainty in the correlations developed results in large scatter in the properties
estimated; and (iii) it is assumed that fracture occurs at the peak load and thus
ignoring the initiation of crack prior to maximum load. To investigate this,
researchers employed several experimental methods, such as endoscopes and acous-
tic emission techniques to find out the onset of crack initiation [32].
One of the primary objectives of life extension is determination of current
fatigue properties of materials, especially after service exposure. The cyclic ABI test
technique is found to correlate with the large specimen fatigue properties to a rea-
sonable level, but the sample size required for cyclic ABI is much large (especially
the thickness), which makes it difficult to estimate the fatigue properties of scooped
samples. This poses restrictions on fatigue data generation for in-service compo-
nents. Studies on the fatigue behavior of materials through cyclic small punch test
are limited [33]. This paper presents a novel test method to estimate the fatigue
response of thin miniature specimens—referred to as cyclic small punch test.
Results of static and cyclic small punch test on a limited set of metallic materials are
presented in the as-received and post-fatigue damage condition.
Experimental Procedures
SPTs have been performed on three different materials: commercially pure copper,
brass, and stainless Steel-304, all sourced in sheet form. The SPT specimens had a
nominal dimension of 10 by 10 by 0.3 mm, which were extracted from sheet
samples. Tensile specimens were prepared from the same sheet specimens as per
the ASTM-E8 [34] standard to estimate the tensile response of materials.
Figure 1 presents the schematic and photograph of the SPT fixture used in this
study. The SPT fixture consists of a pair of dies, namely, the upper and lower die,
one each for holding the punch and specimen respectively. Disc specimen after
mounting in the lower die is held firmly in position by clamping screws. The pres-
sure on the specimen was controlled using a torque wrench. Tungsten carbide ball
of 2.5 mm diameter is placed over the specimen. The punch is then inserted into
the lower die through the guide hole. The whole fixture set up is then placed
between the grips of a 100 kN MTS-810 servo-hydraulic test system. In order to
measure the local deflection of the specimen, a crack opening displacement gage
was mounted between the knife edges of the dies [35]. The force and displacement
transducers of the test system were scaled down to 1/10 of its full range during SPT
test. Monotonic SPT tests were performed on all specimens at a displacement rate
of 0.5 mm/min at room temperature in laboratory air environment. The load–
displacement data was captured at a data acquisition rate of 20 Hz after averaging
FIG. 1 Schematic and photograph of SPT fixture and tensile test. (a) schematic of small
punch test fixture, (b) photograph of small punch test fixture, and (c)
photograph of SPT fixture on test system.
and was analyzed for elastic modulus, peak displacement of punch to estimate the
tensile properties of the material.
The specimen undergoes different modes of deformation under biaxial stress
state before final fracture as shown in Fig. 2. These are: (a) linear elastic bending,
Zone I, (b) plastic bending, Zone II, (c) membrane stretching, Zone III, (d) plas-
tic instability regime, Zone IV, and (e) fracture regime, Zone V. To ensure
repeatability, monotonic tests were carried out at least five times for each speci-
men under similar testing condition. The error in the response was found to be
less than 2 %.
The clamping torque has a significant influence on the load-displacement
response during SPT testing. Clamping torque changes the levels of constraint of
small sheet specimens from that of free bending to fully constrained bending.
Figure 3 presents the load–displacement response for different clamping torque val-
ues for identically prepared specimens. In case of excessive clamping torque to hold
the sheet specimen, thinning of specimen close to edges of small punch occurs and
it distorts the stress–strain scenario for SPT. Deformation during small punch test-
ing is influenced by certain parameters such as friction at the interfaces of clamping.
To ensure repeatability of test results, all specimens were prepared under identical
conditions and they were polished and degreased prior to testing. In order to char-
acterize the optimum levels of clamping torque, tests were repeated to find the
range of clamping torque within which the load–displacement response was within
a scatter band. The ranges thus determined for commercially pure copper, brass,
and SS-304 were 12–15, 14–17, and 25–30 N-m, respectively. The variation in
FIG. 2 Typical load–deflection curve during monotonic small punch testing for a
commercially pure copper material.
FIG. 3 Influence of clamping torque (CT) on load–displacement response for the tested
specimens: (a) commercially pure copper, (b) brass, and (c) SS 304.
TABLE 1 Details of the axial fatigue test parameters for commercially pure copper.
Stress Range, Number of Cycles 60 % of Life 80 % of Life

(Dr), MPa Stress Ratio, R Frequency, Hz to Failure (cycles) (cycles)
32 0.88 1 14965 8979 11972
load–displacement response in all these cases was less than 1 % when the tests
were carried out in the aforementioned clamping torque ranges.
Dog-bone specimens similar to those prepared for tension tests were subjected
to axial tension–tension load controlled fatigue cycling in a 100 kN MTS-servo-
hydraulic Universal Testing Machine (UTM). Details of the fatigue loading are
given in Table 1. In view of the thinness of the specimen for fatigue loading, these
specimens were fatigue cycled at a mean stress of 85 % of peak stress. Few experi-
ments were carried out on commercially pure copper until complete failure of spec-
imen. Based on the mean life to failure, experiments were repeated till fraction of
failure life (60 and 80 % of failure life) and tension–tension fatigue tests were inter-
rupted. Small SPT samples were extracted from these specimens at the gage section
of dog-bone specimen and were tested for monotonic SPT response.
Results of Monotonic SPT

An important observation that was made relates to the ratio of the slope of initial
linear portion of the load–displacement response which, when normalized with ini-
tial thickness of the specimen is found to be proportional to the elastic modulus of
the material being tested (Fig. 4). Thus, it may be written as,
(1) E ¼ kðS=to Þ
FIG. 4 Force–displacement response for various tested specimens.

where:
E ¼ the elastic modulus in MPa,
S ¼ the slope of initial linear portion in N/mm,
to ¼ the initial thickness of the material, and
k ¼ the proportionality constant.
The value of k was evaluated using the mean of the slope of ten test trials and
elastic modulus obtained from tensile test. The values of k thus obtained are
reported in Table 2. The elastic modulus of copper, brass, and SS-304 as estimated
from conventional tensile test is 110.82, 117.65, and 209.64 GPa, respectively
(Fig. 5). One observation to note is that there was no ratcheting response that was
observed during tensile testing of any of the specimens, which could be due to the
thinness of the specimens tested.
The elastic modulus of damaged specimens was calculated from SPT and ten-
sile tests. In order to quantify the amount of damage based on stiffness change, the
following equation [36] is used.
(2) D ¼ 1 ðEd =E Þ
where:
D ¼ the damage parameter, and
Ed and E* ¼ the elastic modulus of damaged and virgin specimen, respectively.
The damage level thus estimated from monotonic SPT and tensile tests are in
reasonable agreement (Table 3), though consistently the damage predicted by SPT is
lower than what is estimated by a conventional tension test.
Cyclic Small Punch Test

Earlier research on cyclic ABI to estimate fatigue life of materials [9–12] prompted
the use of a novel cyclic small punch test to determine the fatigue response of mini-
ature specimens. As indicated earlier, ABI tests are useful when the thickness of the
sample is considerably higher compared to the depth of indentation. In case of thin
specimens extracted from critical locations, use of cyclic ABI may not provide good
results due to boundary conditions of indentation.
Assessing the fatigue resistance of materials from small punch test involves sub-
jecting the specimen to a constant amplitude load cycling in compression–
compression at a given frequency until failure. The specimens were loaded at pre-
determined clamping torque (from monotonic SPT studies) using scaled down
transducer ranges (1/10 of load, displacement) of 100 kN MTS servo-hydraulic test
system. This also called for tuning of servo parameters of the test system.
During cyclic SPT process, the specimen material is plastically deformed and
the hysteresis response is observed for each cycle of loading. In case of stable
stress–strain response under cycling, the loop width and the energy stored inside
the hysteresis loop would remain constant. However, in reality, depending on the
material response to fatigue cycling, the hysteresis energy stored during cycling
176
TABLE 2 Modulus estimated from SPT and tensile test.
Copper - Virgin Copper 60 % Damaged Copper 80 % Damaged Brass - Virgin SS 304 - Virgin
E60 %, E80 %,
S, Evirgin, S, E60 % GPaTension S, E80 % GPaTension S, E, S,
Trail No. N/mm GPa N/mm GPa Test N/mm GPa Test N/mm GPa N/mm E, GPa
1 1905.23 110.82 990.10 57.72 53.98 591.35 34.40 31.95 2102.02 117.65 3092.69 209.64
2 1900.35 989.92 589.22 2098.52 3091.11
3 1901.16 989.62 592.01 2101.63 3089.98
4 1905.62 991.31 591.05 2099.55 3090.90
5 1901.51 990.33 590.11 2100.06 3088.99
6 1902.33 993.01 593.02 2102.21 3092.55
7 1901.75 992.09 592.89 2103.11 3089.56
8 1905.51 991.32 588.98 2101.10 3091.46
9 1904.40 992.81 590.10 2099.09 3092.99
10 1903.36 992.51 589.75 2100.45 3090.82
Mean value 1903.12 991.30 590.85 2100.77 3091.10
Standard 1.96 1.26 1.45 1.48 1.35
deviation
k 17.47 33.53 56.26 16.80 20.35
FIG. 5 Tensile stress–strain curves of tested specimens.
either increases (due to strain softening) or decreases (due to strain hardening). If

the material strain hardens, one would expect increase in dislocation density, as
well as increase in brittleness of the material. Characterizing dislocation density and
dislocation dipole and its influence on fatigue is important for a complete under-
standing of the physics of the problem, but due to constraints, it was kept outside
the scope of this paper. At some levels of cycling, the material fails due to cumula-
tive strain energy accumulated due to cycling and many researchers have used the
strain energy stored in the system during cycling as a parameter to correlate failure
of specimens. This approach was found to work well for cyclic ABI testing, where
cumulative hysteretic strain energy during cyclic ABI was found to be in close
agreement with the cumulative hysteretic strain energy of a low cycle fatigue speci-
men [37,38]. Thus, it can be said that plastic strain energy accumulates during
cyclic SPT with cycles of punching and leads to fracture of the specimen when a
critical value is reached. Continuous tracking of displacement versus cycles and
hysteresis energy versus cycles provides an insight on failure life of material and to
understand the failure initiation, complete failure during cyclic SPT testing. In
TABLE 3 Damage parameter estimated from monotonic SPT and Tensile test.
Material Elastic Modulus Elastic Modulus Damage Damage Percent

(GPa) SPT (GPa) Tension Test Percent (SPT) (Tension test)
Virgin Copper 110.82 110.82 0.00 0.00

60 % of failure life 57.72 53.98 47.92 51.29
80 % of failure life 34.4 31.95 68.96 71.17
addition to the analysis of displacement trace, record of load–displacement was an-

alyzed for hysteretic energy stored inside a loop. Hysteresis energy is used as a pa-
rameter to quantify failure life of specimens.
The peak loads for the cyclic punch testing were selected from the monotonic
test response. The peak loads for cyclic SPT varied between 92 and 85 % of the
monotonic SPT peak load; a sinusoidal cycling of loads with amplitude of 40 N was
applied to fatigue the specimen (Fig. 6(a)). This resulted in the failure of the speci-
men at around 15 000 cycles.
One of the points that was observed during cyclic SPT testing was that the fre-
quency of loading has a significant effect on the hysteresis loop (Fig. 6(b)); this could
be due to the fact that the plastic energy stored in the material creates heating of the
specimen (as confirmed through infrared thermography studies during fatigue cy-
cling of plain cylindrical specimens [11]). Thus, it was found important to subject
the specimen to a frequency, where there is a thermal balance so that there is little
or no effect on plastic dissipation energy. Moreover, hysteresis loop is a combina-
tion of an-elastic and plastic effects. Only a part of applied load contributes to
actual plastic deformation and the rest is converted to heat [39,40]. In order to
extract only the plastic part of the hysteresis loop which actually causes plastic
deformation, cyclic small punch testing was conducted at various frequencies and
load levels. It is observed that at frequencies above 2 Hz (Fig. 7) the hysteresis
increases due an-elastic effects. At frequencies around less than 1.25 Hz, the hyster-
esis becomes nearly stable indicating that heat and an-elastic effects are negligible.
Based on these findings, cyclic small punch experiments were conducted at a fre-
quency of 1 Hz. The plastic dissipation energy per cycle was then continuously
tracked as a function of number of cycles to quantify the fatigue damage. Repeat
experiments were conducted for each test condition to ascertain the fatigue life and
fatigue response of materials under cyclic SPT.
The hysteresis energy and displacement variation during cyclic punch testing
till failure for all the tested samples is presented in Figs. 8 and 9 for different peak
loads. The response may be divided into three stages: (i) primary stage during
which the hysteresis energy accumulation rate decreases to a certain value, (ii) sec-
ondary stage characterized by constant energy accumulation rate, and (iii) tertiary
stage where the hysteresis energy accumulation rate increases abruptly. A jump in
the hysteresis was observed while transiting from secondary to tertiary stage. This
FIG. 6 (a) Input sinusoidal load during cyclic SPT, (b) hysteresis loops for different
frequencies.
might be an indicator of specimen failure, after which the specimen sustained the
applied loading for fewer number of cycles followed by a sharp rise in hysteresis
energy at final fracture. The displacement variation may also be characterized in a
similar way with the number of cycles to failure.
In order to quantify the amount of fatigue damage, the following relation is
proposed:
X X
(3) D¼1 Wd = Wv
FIG. 7 Influence of frequency on hysteresis during cyclic small punch test.
where:
Wd and Wv ¼ the plastic energy dissipated per cycle (also representative of
work per cycle) of damaged and virgin specimen, respectively, and
D ¼ the damage parameter.
To employ this equation, first the virgin and damaged specimens were sub-
jected to monotonic small punch test. Then at the same capacity of the peak
load, i.e., say at 92 % of peak load of virgin and damaged specimens and at a
FIG. 8 Hysteresis versus number of cycles to failure data for various tested
samples at different loads: (a) virgin copper, (b) copper damaged to 60 % of
total life, (c) copper damaged to 80 % of total life, and (d) virgin stainless
steel-304.
given frequency, say at 1 Hz fatigued until failure. The plastic dissipation energy
is estimated for all cycles till failure and then summed to find out the damage
parameter D. The damage parameter of investigated materials is presented in
Table 4. These values are higher than that estimated from static SPT (as shown
in Table 3). This increased degradation may be due to the fatigue loading. A
framework summarizing the steps followed in this work for assessing the loss of
stiffness and fatigue damage quantification is shown in Fig. 10.
Figure 11 shows the Scanning Electron Microscope (SEM) micrograph of failed
specimens of SPT. It can be seen that multiple secondary cracks were found around
the failure surface in 80 % damaged specimen compared to the failure of virgin
specimen which failed by a single crack growth.
Work is in progress to conduct controlled experiments with on-line monitoring
of crack initiation through visual, acoustic–emission techniques to identify on-set
of crack initiation during cyclic SPT testing. It is also proposed to extract the stria-
tion spacing from the specimens subjected to cyclic SPT to verify possible correla-
tion between striation spacing from SPT specimens and fatigue crack growth rate
from conventional fracture mechanics specimens. Work is in progress on polished,
FIG. 9 Displacement versus number of cycles to failure data for various tested samples
at different loads: (a) virgin copper, (b) copper damaged to 60 % of total life, (c)
copper damaged to 80 % of total life, and (d) virgin stainless steel-304.
TABLE 4 Damage parameter estimation for fatigue tested samples.
Material +W, mJ D, percent
Virgin copper 275.36 0

60 % of failure life 124.99 54.6
80 % of failure life 74.40 72.9
FIG. 10 Flow chart for quantifying fatigue damage through cyclic SPT.
FIG. 11 SEM micrographs of cyclic SPT failed specimens of commercially pure Copper
and SS-304: (a) virgin copper material, (b) copper sample after 60 % of failure
life cycling, (c) copper sample after 80 % of failure life cycling, and (d) SS-304
(Virigin).
etched specimens to understand the local microstructure aspects of deformation

during monotonic and cyclic SPT.
As clamping conditions are one of the important variables that affect the qual-
ity of testing, it is proposed to model the cyclic SPT using numerical tools and
arrive at optimum friction coefficient for experiments.
Conclusions
A novel, cyclic small punch test procedure for assessing the degradation of mechan-
ical properties has been proposed. Optimum clamping torque for static SPT test has
been identified for three different materials. An empirical correlation has been
developed between the elastic modulus of metallic materials derived from tension
tests and data derived from the load–displacement response of materials during
monotonic SPT. An energy based fatigue damage quantification framework for
estimating fatigue failure life through cyclic SPT has been proposed by optimizing
the frequency at which hysteresis is constant. The experimental results reveal that
static and cyclic SPT is an effective method for investigating the fatigue damage of
metallic materials. Further work is needed to establish such correlations for differ-
ent materials as well as to understand the mechanism of failure during cyclic SPT.
References
[1] Fleury, E. and Ha, J. S., “Small Punch Tests to Estimate the Mechanical Properties of
Steels for Steam Power Plant: I. Mechanical Strength,” Int. J. Pressure Vessels Piping,
Vol. 75, No. 9, 1998, pp. 699–706.
[2] Taljat, B., Zacharia, T., and Kosel, F., “New Analytical Procedure to Determine Stress–
Strain Curve From Spherical Indentation Data,” Int. J. Solids Struct., Vol. 35, No. 33, 1998,
pp. 4411–4426.
[3] Murty, K. L., Mathew, M. D., Wang, Y., Shah, V. N., and Haggag, F. M., “Non-Destructive
Determination of Tensile Properties and Fracture Toughness of Cold Worked A36 Steel,”
Int. J. Pressure Vessels Piping, Vol. 75, No. 9, 1998, pp. 831–840.
[4] Murty, K. L., Miraglia, P. Q., Mathew, M. D., Shah, V. N., and Haggag, F. M.,
“Characterization of Gradients in Mechanical Properties of SA-533B Steel Welds Using
Ball Indentation,” Int. J. Pressure Vessels Piping, Vol. 76, No. 6, 1999, pp. 361–369.
[5] Hu, X. Z. and Lawn, B. R., “A Simple Indentation Stress–Strain Relation for Contacts With
Spheres on Bilayer Structures,” Thin Solid Films, Vol. 322, Nos. 1–2, 1998, pp. 225–232.
[6] Gao, Y. F., Xu, H. T., Oliver, W. C., and Phar, G. M., “Effective Elastic Modulus of Film-on-
Substrate Systems Under Normal and Tangential Contact,” J. Mech. Phys. Solids, Vol. 56,
No. 2, 2008, pp. 402–416.
[7] Haggag, F. M., Wang, J. A., Sokolov, M. A., and Murty, K. L., “Use of Portable/In-Situ
Stress–Strain Microprobe System to Measure Stress–Strain Behavior and Damage in
Metallic Materials and Structures,” Non-Traditional Methods of Sensing Stress, Strain and
Damage in Materials and Structures, ASTM STP 1318, G. Lucas and D. Stubbs, Eds., ASTM
International, Philadelphia, PA, 1997, pp. 85–98.
[8] Haggag, F. M., Byun, T. S., Hong, J. H., Miraglia, P. Q., and Murty, K. L., “Indentation–
Energy–to–Fracture (IEF) Parameter for Characterization of Carbon Steels Using Non–
Destructive Automated Ball Indentation (ABI) Technique,” Scriptia Materialia, Vol. 38,
No. 4, 1998, pp. 645–651.
[9] Prakash, R. V. and Bhokardole, P., “An Assessment of the Automated Ball Indentation
Test Method to Evaluate the Mechanical Properties of Materials,” Proceedings of 60th
Annual Technical Meeting, Jamshedpur, India, Nov. 15–16, 2006.
[10] Prakash, R. V., Bhokardole, P., and Shin, C. S., “Investigation of Material Fatigue
Behavior Through Cyclic Ball Indentation Testing,” J. ASTM Int., Vol. 5, No. 9, 2008,
pp. 236–255.
[11] Prakash, R. V., “Evaluation of Fatigue Damage in Materials Using Indentation Testing
and Infrared Thermography,” Trans. Indian Inst. Metals, Vol. 63, Nos. 2–3, 2010, pp.
173–179.
[12] Prakash, R. V. and Ghosh, S., “Creep, Creep-Fatigue Damage Identification Through
Monotonic and Cyclic ABI Testing,” Proceedings of the ASTM Symposium on Creep,
Creep-Fatigue Interactions, San Antonio, TX, Nov. 17, 2010.
[13] Ghosh, S. and Prakash, R. V., “Study of Damage and Fracture Toughness Due to Influ-
ence of Creep and Fatigue of Commercially Pure Copper by Monotonic and Cyclic
Indentation,” Metall. Mater. Trans. A, Vol. 44A, 2012, pp. 224–234.
[14] Manahan, M. P., Argon, A. S., and Harling, O. K., “The Development of Miniaturized Disk
Bend Test for the Determination of Post–Irradiation Mechanical Properties,” J. Nucl.
Mater., Vols. 103–104, 1981, pp. 1545–1550.
[15] Mao, X. and Takahashi, H., “Development of a Further-Miniaturized Specimen of 3 mm

Diameter for TEM Disk (U 3 mm) Small Punch Tests,” J. Nucl. Mater., Vol. 150, 1987, pp.
42–52.
[16] Eto, M., Takahashi, H., Misawa, T., Suzuki, M., Nishiyama, Y., Fukaya, K., and Jitsukawa, S.,
“Development of a Miniaturized Bulge Test (Small Punch Test) for Post-Irradiation
Mechanical Property Evaluation,” ASTM STP 1204, ASTM International, Philadelphia, PA,
1993, pp. 241–255.
[17] Xu, Y. and Zhao, Z., “A Modified Miniature Disk Test for Determining Material Mechanical
Properties,” J. Test. Eval., Vol. 23, No. 4, 1995, pp. 300–306.
[18] Fleury, E. and Ha, J. S., “Small Punch Tests on Steels for Steam Power Plant (II),” KSME
Int. J., Vol. 12, No. 5, 1998, pp. 827–835.
[19] Eskner, M. and Sandstrom, R., “Mechanical Property Evaluation Using the Small Punch
Test,” J. Test. Eval., Vol. 32, No. 4, 2004, 11504.
[20] Isselin, J. and Shoji, T., “Yield Strength Evaluation by Small–Punch Test,” J. Test. Eval.,
Vol. 37, No. 6, 2009, JTE101657.
[21] Bayoumi, M. R. and Bassim, M. N., “Study of the Relationship Between Fracture Tough-
ness (JIc) and Bulge Ductility,” Int. J. Fract., Vol. 23, 1983, pp. 71–79.
[22] Mao, X. and Takahashi, H., “Estimation of Mechanical Properties of Irradiated Nuclear
Pressure Vessel Steel by use of Subsized CT Specimen and Small Punch Specimen,” Scr.
Metall., Vol. 25, 1991, pp. 2487–2490.
[23] Baik, J. M., Kameda, J., and Buck, O., “Development of Small Punch Tests for Ductile–
Brittle Transition Temperature Measurement of Temper Embrittled Ni–Cr Steels,” ASTM
STP 888, 1986, ASTM International, Philadelphia, PA, pp. 92–111.
[24] Kameda, J. and Mao, X., “Small-Punch and TEM-Disc Testing Techniques and Their Appli-
cation to Characterization of Radiation Damage,” J. Mater. Sci., Vol. 27, No. 4, 1992, pp.
983–989.
[25] Ule, B., Sustar, T., Dobes, F., Milicka, K., Bicego, V., Tettamanti, S., Maile, K.,
Schwarzkopf, C., Whelan, M. P., Kozlowski, R. H., and Klaput, J., “Small Punch Test
Method Assessment for the Determination of the Residual Creep Life of Service
Exposed Components: Outcomes From an Interlaboratory Exercise,” Nucl. Eng.

Des., Vol. 192, No. 1, 1999, pp. 1–11.
[26] Yang, Z. and Wang, Z., “Relationship Between Strain and Central Deflection in Small
Punch Creep Specimens,” Int. J. Pressure Vessels Piping, Vol. 80, No. 6, 2003, pp.
397–404.
[27] Dobes, F. and Milicka, K., “Application of Creep Small Punch Testing in Assessment of
Creep Lifetime,” Mater. Sci. Eng. A, Vols. 510–511, 2009, pp. 440–443.
[28] Izaki, T., Kobayashi, T., Kusumoto, J., and Kanaya, A., “A Creep Life Assessment Method
for Boiler Pipes Using Small Punch Creep Test,” Int. J. Pressure Vessels Piping, Vol. 86,
No. 9, 2009, pp. 637–642.
[29] Hou, F., Xu, H., Wang, Y., and Zhang, L., “Determination of Creep Property of
1.25Cr0.5Mo Pearlitic Steels by Small Punch Test,” Eng. Failure Anal., Vol. 28, 2013, pp.
215–221.
[30] Zhao, L., Jing, H., Xu, L., Han, Y., Xiu, J., and Qiao, Y., “Evaluating of Creep Property of
Distinct Zones in P92 Steel Welded Joint by Small Punch Creep Test,” Mater. Des., Vol.
47, 2013, pp. 677–686.
[31] Bai, T., Chen, P., and Guan, K., “Evaluation of Stress Corrosion Cracking Susceptibility of
Stainless Steel 304L With Surface Nanocrystallization by Small Punch Test,” Mater. Sci.
Eng. A, Vol. 561, 2012, pp. 498–506.
[32] Budzakoska, E., Carr, D. G., Stathers, P. A., Li, H., Harrison, R. P., Hellier, A. K., and Yeung,
W. Y., “Predicting the J Integral Fracture Toughness of Al 6061 Using the Small Punch
Test,” Fatigue Fract. Eng. Mater. Struct., Vol. 30, No. 9, 2007, pp. 796–807.
[33] Villarraga, M. L., Edidin, A. A., Herr, M., and Kurtz, S. M., “Multiaxial Fatigue Behavior of
Oxidized and Unoxidized UHMWPE During Cyclic Small Punch Testing at Body Temper-
ature,” J. ASTM Int., Vol. 1, No. 1, 2004, 11218.
[34] ASTM E8/E8M-13a: Standard Test Methods for Tension Testing of Metallic Materials,
[35] Ramesh, T., 2012, “Mechanical Property Evaluation of Pressure Vessel Materials Through
Shear Punch and Small Punch Tests,” Ph.D. thesis, IIT Madras, Chennai, India.
[36] Lemaitre, J. and Desomrat, R., Engineering Damage Mechanics, Springer, New York,
2004.
[37] Aneesh Bangia and Raghu V Prakash, “Energy Parameter Correlation of Failure Life
Data between Cyclic Ball Indentation and Low Cycle Fatigue,”Open Journal of Metals,
Vol. 1, No. 2, pp. 31–36.
[38] Prakash, R. V., T. Pravin, Kathirvel, T., and Balasubramaniam, K., “Thermo-Mechanical
Measurement of Elasto-Plastic Transitions During Cyclic Loading,” Theoretical and
Applied Fracture Mechanics, Vol. 56, No. 1, 2011, pp. 1–6.
[39] Lanteigne, J. and Nguyen-Duy, P., “Energy Balance Approach to Low Cycle Fatigue,” J.
Fract., Vol. 23, No. 4, 1983, pp. RI47–RI49.
[40] Chang, C. S., Pimbley, W. T., and Conway, H. D., “An Analysis of Metal Fatigue Based on
Hysteresis Energy,” J. Exp. Mech., Vol. 8, No. 3, 1968, pp. 133–137.
OTHER MECHANICAL TESTS
E. Lucon,1 C. N. McCowan,1 R. L. Santoyo,1

and J. D. Splett2
Certified KLST Miniaturized

Charpy Specimens for the
Indirect Verification of
Small-Scale Impact Machines
Reference
Lucon, E., McCowan, C. N., Santoyo, R. L., and Splett, J. D., “Certified KLST Miniaturized Charpy
Specimens for the Indirect Verification of Small-Scale Impact Machines,” Small Specimen Test
ABSTRACT
Small specimen test techniques are becoming ever more popular as the need
increases to characterize mechanical properties by use of the smallest possible
amount of material, because of various restrictions on material availability,
irradiation, testing space, and other factors. The National Institute of Standards
and Technology (NIST) recently qualified reference miniaturized Charpy V-notch
(MCVN) specimens for the indirect verification of small-scale impact testing
machines (with capacity in the range 15 J to 50 J and impact velocity around
3.8 m/s). The same materials used for NIST standard verification specimens were
evaluated at three energy levels (low, high, and super-high). The miniaturized
specimen type investigated is denominated KLST (from the German
Kleinstprobe, or “small specimen”). In the first phase of the qualification activity,
several instrumented impact tests on miniaturized KLST specimens of low, high,
and super-high energy were performed and analyzed at NIST. In this part of the
study, which was published elsewhere [Lucon, E., “Certified Miniaturized Charpy
10, 2015.
1
National Institute of Standards and Technology (NIST), Applied Chemicals and Materials Division, Boulder,
CO 80305, United States of America.
2
National Institute of Standards and Technology (NIST), Statistical Engineering Division, Boulder, CO 80305,
United States of America.
3
Houston, TX.
Copyright V
Specimens for the Indirect Verification of Small-Size Impact Machines,” Mater.

Perform. Character., Vol. 2, No. 1, 2013], the variability of MCVN data was
compared to that of full-size Charpy data from the same lot of specimens.
Additional aspects were also investigated, such as the influence of shear lip
symmetry and specimen full or partial fracture on absorbed energy, and the
correlation between miniaturized and full-size Charpy data. In the second phase
of the qualification activity, certified values of absorbed energy and maximum
force were established for each energy level by means of an international round
robin, which involved nine highly qualified laboratories. The results of this round
robin, and the outcome of the statistical analyses performed in accordance with
ASTM E691-13 (ASTM E691-13: Standard Practice for Conducting an
Interlaboratory Study to Determine the Precision of a Test Method, ASTM
International, West Conshohocken, PA) and ISO 5725-2 [ISO 5725-2, 1994,
“Accuracy (Trueness and Precision) of Measurement Methods and Results,”
International Organization for Standardization, Geneva, Switzerland], are
presented in this paper. Complete details are provided by NIST [Lucon, E.,
McCowan, C., Santoyo, R., and Splett, J., “Standard Reference Materials—
Certification Report for SRM 2216, 2218, 2219: KLST (Miniaturized) Charpy
V-Notch Impact Specimens,” NIST Special Publication 260-180, National Institute
of Standards and Technology, Gaithersburg, MD, 2013], which accompanies the
certified KLST.
Keywords
small specimen test techniques, miniaturized Charpy specimens, small-scale
impact machines, KLST, instrumented impact tests, international round robin,
statistical analyses
Introduction
Miniaturization of mechanical test samples, including impact specimens, is becom-
ing increasingly important as material consumption or material availability often
represent limiting factors for mechanical testing. This paper documents the proce-
dures used at NIST to develop certified values of maximum force (Fm) and absorbed
energy (KV) for miniaturized reference Charpy V-notch impact specimens of KLST
type, at three absorbed energy levels (low, high, and super-high).
KLST (from the German Kleinstprobe, or “small specimen”) is the most com-
monly used miniaturized Charpy specimen (see Fig. 1). It was the first type of minia-
turized Charpy V-notch (MCVN) specimen to be standardized in an amendment
(annex D) to ISO 14556 [1]. The use of KLST specimens is also covered by ASTM
E2248-13 [2]. The edge of the instrumented striker used for testing KLST specimens
has a radius of 2 mm; the distance between specimen supports (span) is 22 mm.
Although KLST specimens can also be tested on a conventional, full-size impact
tester (provided anvils and supports are adequately modified), the recommended
procedure is to use a small-scale pendulum with a significantly lower potential
energy (15 J to 50 J instead of 300 J or more) and slightly lower impact speed
LUCON ET AL., DOI 10.1520/STP157620140003 191
FIG. 1 KLST-type MCVN specimen (dimensions in mm).
(3.8 m/s instead of 5 m/s or more). Such small-scale impact testers cannot be indi-
rectly verified by means of standard Charpy reference specimens, for dimensional
and energy reasons. On the other hand, MCVN specimens with certified values of
absorbed energy or maximum force were previously unavailable.
The investigations reported in this paper allowed the National Institute of
Standards and Technology (NIST) to certify KLST reference specimens of three
energy levels, which are available to users for the indirect verification of small-scale
impact testers. These specimens permit verification at room temperature
(21 C 6 5 C) of both the force scale (by means of certified maximum force values,
Fm) and the energy scale (by means of certified absorbed energy values, KV).
Materials and Specimens

The feasibility of fabricating MCVN reference specimens of the same materials and
equivalent energy levels used for existing Charpy V-notch (CVN) reference speci-
mens had been demonstrated in a collaboration between the Institute for Reference
Materials and Measurements (IRMM) (EU reference specimen producer) and
SCKCEN (Belgian Nuclear Research Center) [3].
TABLE 1 Certified values of absorbed energy and maximum force for the full-size impact speci-
mens from which KLST specimens were extracted.
KV at 40 C 6 1 C KV at 21 C 6 1 C Fm at 21 C 6 1 C
Certified Expanded Certified Expanded Certified Expanded

Ref. value Uncertainty Reference Uncertainty Reference Uncertainty
Lot (J) (J) value (J) (J) value (kN) (kN)
LL-103 15.3 0.1 18.2 0.1 33.00 1.86

HH-103 97.5 0.6 105.3 0.6 24.06 0.70
SH-36 – – 239.8 1.2 25.64a –
a
Lot SH-36 has no certified maximum force value. Therefore, we used as a reference value the
average of 51 instrumented tests performed at room temperature (Fm ¼ 25.64 kN with standard
deviation r ¼ 0.09 kN).
TABLE 2 Chemical composition of AISI 4340 steel, used for low-energy and high-energy lots
(weight %).
C Si Mn Ni Cr Mo S P
0.4 0.28 0.66 1.77 0.83 0.28 0.001 0.004
Following in the same footsteps, NIST extracted KLST-type MCVN specimens

from previously tested full-size impact verification specimens of low, high, and
super-high energy. The certified values of absorbed energy and maximum force for
the three lots used are shown in Table 1 with expanded uncertainties.
Low-energy and high-energy specimens were manufactured from AISI 4340
steel bars from a single heat-treated batch to minimize compositional and micro-
structural variations. The chemical composition of the steel is given in Table 2.
Super-high-energy specimens were manufactured from double-vacuum-melted
18Ni-type T-200 maraging steel, adequately forged prior to rolling to minimize
compositional and microstructural variations. The nominal chemical composition
of the steel is shown in Table 3.
Additional information on the materials, the heat treatments, and the machin-
ing processes can be found in Ref 4.
KLST specimens were extracted from broken full-size Charpy samples by the
use of electro-discharge machining (EDM), making sure the orientations of the
specimen and the notch were preserved (Fig. 2). A wire with 0.1 mm (0.004 in.)
diameter was used to cut the notches by EDM.
Note that any plastic deformation ensuing from previous full-size tests (signifi-
cant only in the case of super-high-energy specimens) has no influence on the
results of the miniaturized specimens, because it is limited to the ends of the
MCVN sample, which are ground flat during the machining stage.
Preliminary Qualification of KLST Reference

Specimens: NIST Tests on Low, High, and
Super-High KLST Specimens
Before initiating the round robin, which qualified the KLST reference specimens,
preliminary non-instrumented and instrumented impact tests were conducted at
TABLE 3 Nominal chemical composition of type T-200 steel, used for lot SH-36 (wt. %).
Ni Mo Ti Al C Si Mn S Co P
18.5 3.0 0.7 0.1 0.01 0.01 0.1 0.01 0.5 0.01
FIG. 2 Extraction of four miniaturized specimens from one half of a previously tested,
full-size Charpy sample.
NIST in Boulder, CO. These tests were already documented in a previous publica-
tion by the same authors [4], and only the main results will be recalled here.
The test matrix for this pre-qualification phase is shown in Table 4. Note that
the aim of testing KLST specimens of super-high-energy levels in this phase was
just to establish whether the corresponding KV values were significantly different
from those obtained from high-energy specimens.
The results obtained indicate that the production and the certification of KLST
reference specimens are feasible for the verification of both absorbed energy and
maximum force. The scatter of KV values for MCVN specimens was slightly higher
than for full-size specimens at low energy, and similar at high and super-high
energy. For maximum forces, the data scatter from NIST tests was smaller than
from the round robin that qualified the dynamic force verification specimens from
two of the same batches [5]. The tests performed also demonstrate that super-high-
TABLE 4 Test matrix for the preliminary qualification of KLST reference specimens at NIST.
KLST Specimens
Energy Level Non-Instrumented Instrumented Total
Low 7 25 32
High 7 25 32
Super-high 3 17 20
FIG. 3 Examples of symmetrical and asymmetrical fracture.
energy KLST specimens provide sufficiently different absorbed energy

values with respect to high-energy specimens (on the average, 9.79 J as compared to
5.20 J).
Additional issues were investigated in the pre-qualification activity, and the
main outcomes are summarized below. Please refer to Ref 4 for full details.
(1) Average KV values were correlated between KLST and full-size CVN speci-
mens, and the results compared to normalization/correlation approaches
published in the literature. It was found that none of the published methods
were completely satisfactory at all energy levels, and the actual correlations
depend on energy level and material strength. Empirical relations were
obtained by fitting the data with power-law curves.
(2) The relationship between fracture symmetry (i.e., both shear lips on the
same specimen half or one on either half; see Fig. 3) and absorbed energy
was investigated. At the low-energy level, no influence was detected on
absorbed energy, whereas for high- and super-high-energy specimens, asym-
metrical fracture was found to be statistically associated to higher absorbed
energies.
(3) At low- and high-energy levels, all KLST specimens exited the machine
anvils fully broken. Conversely, at the super-high-energy level, none of the
specimens fully fractured during the test.
Description of the Round-Robin Exercise

An international interlaboratory comparison (round-robin exercise), organized
and coordinated by NIST, was conducted among nine laboratories with
well-documented expertise in testing MCVN specimens. With the exception of

NIST, all of the remaining participants were in Europe.
Each participant received five KLST specimens per energy level, to be tested at
room temperature (21 C 6 3 C) with a small-scale impact tester equipped with a
2-mm instrumented striker, in accordance with ISO 14556 (2000) [1], ASTM
E2248-09 [2], and ASTM E2298-13a [6].4
For every test performed, participants were required to report the following
data:
• characteristic force values: force at general yield, Fgy, and maximum force, Fm;
in addition (only at the low-energy level), force at unstable crack propagation,
Fbf, and crack arrest force, Fa;
• characteristic absorbed energy values: energy at general yield, Wgy, at maxi-
mum force, Wm, and at test termination, Wt; in addition (only at the low-
energy level), energy at unstable crack propagation,5 Wbf, and at crack arrest,
Wa; and
• absorbed energy (KV) provided by the machine dial and/or encoder.
Optionally, participants were also asked to measure and report lateral expan-
sion (LE) and shear fracture appearance (SFA). Finally, labs were requested to pro-
vide force-time and/or force-displacement curves for all tests, in ASCII or Excel
format.
According to the information provided by the participants, all the labs used a
static calibration of their instrumented striker, as recommended by both ASTM
E2298-13a [6] and ISO 14556 (2000) [1].
Statistical Analyses of the Round-Robin Results

FOREWORD
For the statistical analyses of the round-robin results, only the following parameters
were considered: Fgy, Fm, Wt, and KV. In the analyses, participating labs were iden-
tified with numbers from 1 to 9 to guarantee anonymity. The average values
reported by participants are presented in Table 5.
The statistical analyses were conducted primarily in accordance with ISO
5725-2:1994 [7] and ISO/TR 22971:2005 [8]. However, some analytical steps also
reflected the provisions of ASTM E691-14 [9].
INVESTIGATION OF REPEATABILITY AND REPRODUCIBILITY

OF THE TEST METHOD
Repeatability and reproducibility standard deviations were established from data
collected for each of the investigated parameters. However, the presence of
4
The operational procedures of ISO 14556, ASTM E2248, and ASTM E2298 are practically identical. The most
notable differences are in some of the symbols used (for example, the force at unstable crack propagation is
designated Fbf in the ASTM standards and Fiu in the ISO standard).
5
This parameter is designated Wiu in ISO 14556 (2000).
196
TABLE 5 Collection of average values provided by round-robin participants.
Low Energy High Energy Super-High Energy
Laboratory Fgy (kN) Fm (kN) Wt (J) KV (J) Laboratory Fgy (kN) Fm (kN) Wt (J) KV (J) Laboratory Fgy (kN) Fm (kN) Wt (J) KV (J)
1 1.51 2.33 1.32 1.52 1 1.37 1.74 4.88 5.58 1 1.17 1.77 8.98 10.26
2 1.87 2.35 1.43 1.55 2 1.50 1.79 5.64 5.70 2 1.45 1.79 9.96 9.99
3 2.14 2.23 1.43 1.78 3 1.39 1.67 5.41 5.75 3 1.52 1.69 9.81 10.15
4 1.98 2.80 1.52 1.57 4 1.61 1.90 5.80 5.63 4 1.70 1.90 10.40 9.86
5 1.93 2.71 1.48 1.40 5 1.61 1.93 6.33 5.46 5 1.66 1.97 11.11 9.83
6 1.50 2.26 1.49 1.90 6 1.36 1.70 5.51 5.92 6 1.42 1.70 9.95 10.36
7 1.76 2.35 1.42 1.48 7 1.35 1.67 5.44 5.49 7 1.43 1.67 9.72 9.82
8 1.96 2.42 1.64 1.60 8 1.34 1.77 5.74 5.61 8 1.29 1.79 10.22 10.07
9 2.43 1.49 1.52 9 1.36 1.84 5.77 5.62 9 1.34 1.85 10.44 10.11
FIG. 4 Mandel’s h plot for absorbed energy KV, low-energy specimens. The dotted lines
correspond to 95 % confidence levels, whereas the dashed lines indicate the
99 % confidence levels. Lab #6 exhibits a significant positive deviation (at the
95 % level) from the remaining participants.
individual laboratories or values that appear to be inconsistent with the general

trend may affect the calculations, and decisions need to be made with respect to
these suspected outlier values.
ISO 5725-2:1994 [7] introduces two measures, originally proposed by
Mandel [10]:
• the between-laboratory consistency statistic, h (ratio of the difference between
the lab mean and the mean of all labs, and the standard deviation of the means
from all labs); and
• the within-laboratory consistency statistic, k (quotient of the lab standard
deviation and the mean standard deviation for all labs).
For each energy level, based on a graphical inspection of Mandel’s h and k
plots, one can identify individual results for each laboratory that might be consid-
ered different from the expected distribution of results, i.e., exceeding critical values
of Mandel’s statistics at the 95 % and 99 % confidence levels.
In general, Mandel’s statistics indicated no specific laboratories exhibiting pat-
terns of results that are markedly different, as indicated by consistently high or low
within-laboratory or between-laboratory variations across all measured parameters.
However, the presence of a few extreme values of h and k, mostly at the 95 % confi-
dence level (two examples are shown in Fig. 4 and Fig. 5), warranted the investiga-
tion of possible outliers through numerical outlier techniques, as detailed in the
following section.
FIG. 5 Mandel’s k plot for max force Fm, high-energy specimens. The dotted line
corresponds to the 95 % confidence level, whereas the dashed line indicates the
99 % confidence level. Lab #5 exhibits a very significant deviation (at the 99 %
level) from the remaining participants.
OUTLIER DETECTION
ISO 5725-2:1994 [7] recommends confidence levels of 95 % to define values termed
“stragglers” and 99 % to define values termed “statistical outliers.”
In the numerical tests used to detect the presence or absence of outliers, it is
assumed that the results are distributed in a Gaussian or normal manner, or at least
according to a single unimodal distribution. It is also assumed that the number of
tests performed and the number of results provided for each parameter are the
same for all participants. For the round robin under investigation, these hypotheses
were fulfilled with only minor discrepancies, and the interlaboratory comparison
can be considered essentially “balanced” and, furthermore, results were obtained
under repeatability conditions (i.e., following the same experimental procedure).
Most outlier tests compare some measure of the relative distance of the suspect
result from the mean of all results. Many statistical tests are available, but ISO
5725-2:1994 [7] recommends the use of Grubbs’ test for laboratory means [11] and
Cochran’s test for laboratory standard deviations [12].
Grubbs’ Test
Grubbs’ test was used to check whether any of the lab means in Table 5 was excep-
tionally high or low and would have inflated the estimate of the reproducibility
standard deviation, if retained. The statistic used in Grubbs’ test is closely related to
Mandel’s h statistic.
Grubbs’ test was performed to establish whether the highest or lowest mean
value could be identified as a single straggler or outlier, at the 95 % and 99 % confi-
dence levels, respectively. No stragglers or outliers were identified for any measured
parameters at any energy level.
Cochran’s Test
Cochran’s test identifies variances that are greater than the expected variances for
the parameter of interest. In this respect, Cochran’s test is a one-sided test, as only
the laboratory exhibiting the largest variance (but not the one exhibiting the small-
est) is tested. Cochran’s test is closely related to Mandel’s k statistic.
The analyses identified two Fm outlier values (lab #5 at high-energy level and
lab #7 at super-high-energy level), which were then discarded for the calculation of
the repeatability standard deviation. No stragglers or outliers were found among
absorbed energy values (Wt or KV).
CALCULATION OF REPEATABILITY AND REPRODUCIBILITY

The precision of the measurement of maximum force Fm and absorbed energy KV
was assessed with ASTM E691-14 [9], which relies on the assumption that an equal
number of measurements are taken by each laboratory. Although one of the
participants reported only four values of KV for low-energy specimens, this small
deviation in sample size is not expected to significantly influence the results of the
analysis.
Table 6 lists the grand means for maximum force and absorbed energy for each
energy level. The grand mean is the unweighted mean of individual laboratory
means.
Based on the values of Mandel’s statistics h and k previously calculated and on
the outcome of Grubbs’ and Cochran’s outlier tests, we concluded that:
(1) None of the lab means should be removed from the analyses on the grounds
of inconsistency with respect to the other labs.
(2) One data set and its standard deviation (lab #5, Fm at the high-energy level)
returned k > 1.81 (critical value at the 0.5 % significance level) and was classi-
fied by Cochran’s test as a statistical outlier. It should therefore be removed
from the calculation of the repeatability standard deviation Sr.
TABLE 6 Grand means for maximum force and absorbed energy.
Fm (kN) KV (J)
Low-energy specimens 2.43 1.59

High-energy specimens 1.79 5.65
Super-high-energy specimens 1.79 10.03
TABLE 7 Precision statistics calculated from the round-robin results.
Energy Level Parameter Sr r SR R
Low Fm (kN) 0.035 0.099 0.196 0.547

KV (J) 0.063 0.175 0.155 0.433
High Fm (kN) 0.019 0.052 0.095 0.267
KV (J) 0.109 0.305 0.137 0.384
Super-high Fm (kN) 0.013 0.037 0.100 0.281
KV (J) 0.226 0.634 0.226 0.634
The precision statistics calculated for each parameter and energy level are given
in Table 7: repeatability standard deviation (Sr), 95 % repeatability limit (r ¼ 2.8Sr),
reproducibility standard deviation (SR), and 95 % reproducibility limit (R ¼ 2.8SR).
ADDITIONAL ANALYSES PERFORMED ON FORCE AND ABSORBED

ENERGY VALUES
Relationship between Different Measures of Absorbed
Energy (Wt and KV)
It has been contended [13,14] that the difference between the absorbed energy
returned by the machine dial/encoder (KV) and that calculated from the instru-
mented force-displacement curve (Wt) is a good indicator of the overall perform-
ance of an instrumented impact tester. ASTM E2298-13a [6] requires the difference
between KV and Wt to be within 15 % of KV or 1 J, whichever is larger. This corre-
sponds to the following limits for the round-robin tests: 1 J at the low- and high-
energy level, and 1.5 J (15 % of the grand mean of KV) at the super-high-energy
level.
The relationship between the two measures of absorbed energy varies from
machine to machine, and depends on the characteristics of the instrumented striker
(striker design and static calibration) [13,14].
Values of Wt and KV are illustrated in Fig. 6 (low energy), Fig. 7 (high energy),
and Fig. 8 (super-high energy), with bounds corresponding to the ASTM E2298-13a
[6] limits. Examination of the figures shows that trends from individual laboratories
for KV and Wt are consistent across the different energy levels, thus confirming
that the relationship between the two parameters is machine- and striker-depend-
ent. Based on these comparisons, the overall performance of the participants’ equip-
ment appears satisfactory.
It is also worth noting that at the high- (Fig. 7) and super-high-energy level
(Fig. 8), instrumented energy values show more scatter (horizontal spread) than
encoder energy values, as should be expected because of the additional uncertainty
introduced by individual striker calibrations. This effect is less evident for low-
energy data (Fig. 6).
FIG. 6 Instrumented and encoder absorbed energy values from low-energy KLST
verification specimens.
Correlation between Full-Size and KLST Absorbed Energies

The correlations between values of KV yielded by different Charpy specimen types
have been extensively investigated. An overview of various correlations of upper
shelf energy (USE) data was provided by Sokolov and Alexander [15]. An addi-
tional correlation approach was published in Ref 16.
FIG. 7 Instrumented and encoder absorbed energy values from high-energy KLST
verification specimens.
FIG. 8 Instrumented and encoder absorbed energy values from super-high-energy

KLST verification specimens.
A summary of different normalization factors (NF ¼ USEfull-size/USEKLST)

calculated for KLST miniaturized specimens, based on a literature search detailed in
Ref 16, is provided in Table 8.
Considering the KLST tests performed at three energy levels in the framework
of the round robin, experimental normalization factors NFexp were calculated by
dividing the certified/average values of KVfs at room temperature by the average
KLST absorbed energies for every data set. The results are shown in Table 9 and
should be compared with the normalized factors listed in Table 8.
Theoretical (Table 8) and experimental (Table 9) normalization factors are
graphically compared in Fig. 9.
When examining Fig. 9, one must remember that all the approaches found in
the literature address USE values, rather than generic values of absorbed energy. At
the high- and super-high-energy levels, analysis of the instrumented traces
indicates fully ductile behavior; therefore, it is appropriate to assume KV ¼ USE.
TABLE 8 Summary of normalization factors for estimating full-size USE based on KLST USE.
NF1 NF2 NF3 NF4 NF5 NF6
[17] [18] [19] [20] [15] [16]
8.9 26.5 23.7 13 24.9 21.6

TABLE 9 Experimental normalization factors obtained from KLST tests performed by round-robin
participants, with corresponding standard errors corresponding to a 95 % two-sided
uncertainty interval.
KV fs
NFexp ¼
Energy Level Specimen Type KV (J) KV ss Standard Error NFexp
Low Full-size 18.2 (cv) 11.4 1.84

KLST 1.60 (mv)
High Full-size 105.3 (cv) 18.7 10.65
KLST 5.64 (mv)
Super-high Full-size 239.8 (cv) 23.8 12.71
KLST 10.05 (mv)
Note: cv ¼ certified value (at RT); mv ¼ mean value (at RT).
For low-energy specimens, however, the material’s behavior is typical of the ductile-
to-brittle transition regime (shear fracture appearance values, both measured by
some participants and estimated from the analysis of the instrumented test records,
range between 27 % and 71 %), and, therefore, it is not appropriate to assume
KV ¼ USE.
FIG. 9 Comparison between theoretical and experimental normalization factors for

KLST specimens at different energy levels. The dotted curves correspond to the
standard errors of the regression parameters.
FIG. 10 Correlation obtained between KVKLST and KVfs. The dotted curves correspond
to the standard errors of the regression parameters.
The most immediate conclusion is that normalization factors depend on

absorbed energy. None of the approaches considered appears convincing for all
energy levels. Based on test results, the following relationship between KVfs and
KVKLST was obtained (Fig. 10):
1:4147
(1) KVfs ¼ 9:132 KVKLST
This correlation is strictly applicable only to the materials investigated.
Correlation Between Full-Size and KLST Maximum Forces

The relationship observed between the mean values of maximum forces measured
from KLST specimens and the corresponding reference/average values for full-size
specimens is shown in Fig. 11. General trends were found to be consistent between
the two specimen types, i.e., the highest Fm was observed at low energy; super-high-
energy specimens provide slightly higher Fm values than high-energy specimens.
Establishment of Certified Reference Values

and Sample Size
The certified reference values for maximum force Fm and absorbed energy KV were
determined from the round-robin results by the use of the Mandel–Paule method
for computing consensus values [21].
FIG. 11 Correlation obtained between Fm,KLST and Fm,fs.
Certified reference values correspond to consensus values. These are weighted

means in which the weights account for both within-machine and between-
machine variation. Detailed calculations can be found in Ref 22.
For the calculation of expanded uncertainties corresponding to 95 % uncer-
tainty intervals, a coverage factor was used, based on a t-table value for 8 degrees of
freedom (df) and a 95 % two-sided uncertainty interval. The values obtained are
summarized in Table 10.
According to the standard operating procedures of the Charpy Verification
Program at NIST [23], the sample size n is the minimum number of specimens
from a given production lot that should be tested in a verification test. It is
calculated as:
TABLE 10 Certified reference values, standard uncertainties, coverage factors, and expanded
uncertainties.
Energy Certified Standard Coverage Expanded

Level Parameter Reference Value Uncertainty Factor, k Uncertainty
Low Fm (kN) 2.43 0.065 2.306 0.150

KV (J) 1.59 0.051 2.306 0.117
High Fm (kN) 1.78 0.035 2.306 0.081
KV (J) 5.65 0.043 2.306 0.098
Super-high Fm (kN) 1.79 0.034 2.306 0.077
KV (J) 10.03 0.075 2.306 0.173
TABLE 11 Sample-size calculations for KLST verification specimens.
Energy Level sp (J) 5 % KV (J) E (J) Sample Size, n
Low 0.062 0.080 0.255 0.5

High 0.109 0.282 0.282 1.3
Super-high 0.226 0.503 0.503 1.8
2
3sp
(2) n¼
E
where:
sp ¼ the pooled standard deviation (or the machine standard deviation if only
one machine is used), and
E ¼ the greater of 0.255 J or 5 % of the certified reference value.
E ¼ 0.255 J, was obtained by applying the exponential regression function of
Eq 1 to the value E ¼ 1.4 J used in Ref 23 for full-size Charpy specimens.
The results of the sample-size calculations based on KV data from the round
robin are given in Table 11.
The values of n reported in Table 11 would justify, from a purely statistical stand-
point, a sample size of two specimens in a set for all three energy levels. A sample
size larger than 5 would indicate that the material in question is not adequate for
producing reference specimens.
Conclusions
An international round robin involving nine highly qualified laboratories allowed
NIST to certify miniaturized KLST-type Charpy specimens for the indirect verifi-
cation of small-scale impact testing machines (Table 12). The statistical analyses of
the round-robin results, conducted in accordance with ISO and ASTM proce-
dures, allowed establishing reference values of maximum force and absorbed
energy at three distinct energy levels, as well as their associated expanded stand-
ard deviation.
Additional observations emerging from the activities described can be summar-
ized as follows:
(a) Based on the comparisons between encoder and instrumented absorbed ener-
gies, the performance of the test equipment of all participating laboratories
appears satisfactory.
(b) The correlation between absorbed energies measured from miniaturized and
full-size Charpy specimens depends on energy level, and cannot be
represented by a unique normalization factor.
(c) The general trends for maximum forces at the three energy levels are consist-
ent between full-size and MCVN specimens.
TABLE 12 Round-robin participants.
Laboratory Location
VTT Espoo, Finland

KIT Karlsruhe, Germany
CIEMAT Madrid, Spain
SCKCEN Mol, Belgium
NRG Petten, The Netherlands
BAM Berlin, Germany
UJV Rež, Czech Republic
NIST Boulder, CO, United States
HZDR Rossendorf, Germany
NOTE: The order of the participating labs is randomized, and does not correspond to the labora-
tory numbers in the paper.
ACKNOWLEDGMENTS
The writers gratefully acknowledge the collaboration of all the participating laboratories.
[1] ISO 14556:2000, “Steel-Charpy V-Notch Pendulum Impact Test-Instrumented Test

Method,” ISO, Geneva, Switzerland, 2000.
[2] ASTM E2248-13: Standard Test Method for Impact Testing of Miniaturized Charpy V-Notch
Specimens,” ASTM International, West Conshohocken, PA, 2013, www.astm.org.
[3] Lucon, E., Roebben, G., Puzzolante, J.-L., and Lamberty, A., “Impact Characterization of
Sub-Size Charpy V-Notch Specimens Prepared from Full-Size Certified Reference
Charpy V-Notch Test Pieces,” J. of ASTM Int., Vol. 2, No. 7, 2005.
[4] Lucon, E., “Certified Miniaturized Charpy Specimens for the Indirect Verification of Small-
Size Impact Machines,” Materials Performance and Characterization, Vol. 2, No. 1, 2013.
[5] McCowan, C. N., Splett, J. D., and Lucon, E., “Dynamic Force Measurement: Instrumented
Charpy Impact Testing,” NIST-IR 6652, NIST, Gaithersburg, MD, 2008.
[6] ASTM E2298-13a: Standard Test Method for Instrumented Impact Testing of Metallic
Materials, ASTM International, West Conshohocken, PA, 2013, www.astm.org.
[7] ISO 5725-2:1994, “Accuracy (Trueness and Precision) of Measurement Methods and
Results-Part 2: Basic Method for the Determination of Repeatability and Reproducibility
of a Standard Measurement Method,” ISO, Geneva, Switzerland, 1994.
[8] ISO/TR 22971:2005, “Accuracy (Trueness and Precision) of Measurement Methods and
Results- Practical Guidance for the use of ISO 5725-2:1994 in Designing, Implementing
and Statistically Analysing Interlaboratory Repeatability and Reproducibility Results,”
ISO, Geneva, Switzerland, 2005.
[9] ASTM E691-14: Standard Practice for Conducting an Interlaboratory Study to Determine
the Precision of a Test Method, ASTM International, West Conshohocken, PA, 2014,
www.astm.org.
[10] Mandel, J., “A New Analysis of Interlaboratory Test Results,” in ASQC Quality Congress
Transaction – Baltimore, MD, May 6-8, 1985, pp. 360-366.
[11] Grubbs, F. E., “Sample Criteria for Testing Outlying Observations,” Annals of Mathemati-
cal Statistics, Vol. 21, No. 1, 1950, pp. 27-58.
[12] Cochran, W. G., “The Distribution of the Largest of a set of Estimated Variances as a Frac-
tion of Their Total,” Annals of Human Genetics, London, Vol. 11, No. 1, 1941, pp. 47–52.
[13] Lucon, E., Chaouadi, R., and van Walle, E., “Different Approaches for the Verification of
Force Values Measured with Instrumented Charpy Strikers,” in Pendulum Impact
Machines: Procedures and Specimens, ASTM STP 1476, T. Siewert, M. Manahan, and C.
McCowan, Eds., ASTM, West Conshohocken, PA, 2006, pp. 95-102.
[14] Lucon, E., “On the Effectiveness of the Dynamic Force Adjustment for Reducing the
Scatter of Instrumented Charpy Results,” J. ASTM Int., Vol. 6, No. 1, 2009.
[15] Sokolov, M. A. and Alexander, D. J., “An Improved Correlation Procedure for Subsize and
Full-Size Charpy Impact Specimen Data,” NUREG/CR-6379, ORNL-6888, 1997.
[16] Lucon, E., Chaouadi, R., Fabry, A., Puzzolante, J.-L., and van Walle, E., “Characterisation
of Materials Properties by use of Full Size and Subsize Charpy Tests: An Overview of
Different Correlation Procedures,” in Pendulum Impact Testing: A Century of Progress,
ASTM STP 1380, T. A. Siewert, Ed., ASTM, Philadelphia, PA, 2000, pp. 146-163.
[17] Corwin, W. R., Klueh, R. L., and Vitek, J. M., “Effect of Specimen Size and Nickel Content
on the Impact Properties of 12 Cr-1 MoVW Ferritic Steel,” J. Nucl. Mater., Vol. 122, Nos. 1-
3, 1984, pp. 343–348.
[18] Corwin, W. R. and Hougland, A. M., “Effect of Specimen Size and Material Condition on
the Charpy Impact Properties of 9Cr-1Mo-V-Nb Steel,” in The Use of Small-Scale Speci-
mens for Testing Irradiated Material, ASTM STP 888, W. R. Corwin and G. E. Lucas, Eds.,
ASTM, Philadelphia, PA, 1986, pp. 325-338.
[19] Lucas, G. E., Odette, G. R., Sheckherd, J. W., McConnell, P., and Perrin, J., “Subsized Bend
and Charpy V-Notch Specimens for Irradiated Testing,” in The Use of Small-Scale Speci-
mens for Testing Irradiated Material, ASTM STP 888, W. R. Corwin and G. E. Lucas, Eds.,
ASTM, Philadelphia, PA, 1986, pp. 304-324.
[20] Louden, B. S., Kumar, A. S., Garner, F. A., Hamilton, M. L., and Hu, W. L., “The Influence of
Specimen Size on Charpy Impact Testing of Unirradiated HT-9,” J. of Nuclear Mater., Vol.
155-157, 1988, pp. 662-667.
[21] Paule, R. C. and Mandel, J., “Consensus Values and Weighting Factors,” NIST Journal of
Research, 87, No. 5, 1982, pp. 303-308.
R
[22] Lucon, E., McCowan, C., Santoyo, R., and Splett, J., “Standard Reference MaterialsV - Certi-
fication Report for SRM 2216, 2218, 2219: KLST (Miniaturized) Charpy V-Notch Impact
Specimens,” NIST Special Publication 260-180, 2013.
[23] McCowan, C. N., Siewert, T. A., and Vigliotti, D. P., “Charpy Verification Program: Reports
Covering 1989-2002,” NIST Technical Note 1500-9, NIST, Gaithersburg, MD, 2003.
Yuzuru Ito,1 Masahiro Saito,2 Katsunori Abe,2

and Eiichi Wakai3
Effect of Hydrogen on Crack

Growth Behavior in F82H Steel
Using Small-Size Specimen
Reference
Ito, Yuzuru, Saito, Masahiro, Abe, Katsunori, and Wakai, Eiichi, “Effect of Hydrogen on Crack
Growth Behavior in F82H Steel Using Small-Size Specimen,” Small Specimen Test Techniques:
6th Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon, Eds., pp. 209–224, doi:10.1520/
ABSTRACT
Crack growth testing is one of the necessary material tests at the high flux test
module. It is necessary to develop a small-size specimen in order to measure
crack growth. In this study, the effect of hydrogen on the crack growth in F82H
steel was investigated through the use of specimens that were one-quarter the
size of a standard-size specimen and were cathodically hydrogen charged. A
wedge opening load (WOL) specimen was employed for the miniaturized
specimen, because the loading mechanism of WOL specimens is quite simple
relative to those of other notched-specimen designs. The loading bolt and pin
were coated with zirconium oxide film to provide electrical insulation between
the WOL specimen and the loading bolt and pin, in order to avoid the formation
of a local cell in the water. The hydrogen charging process was conducted for
around 600 h in electrolytic solution at a temperature of 35 C with a current
density of 2 mA/cm2. A constant load with an equivalent initial stress intensity
Manuscript received March 5, 2014; accepted for publication September 11, 2014; published online
September 30, 2014.
1
Japan Atomic Energy Agency, IFMIF Target and Test Facilities Development Group, Japan Atomic Energy
Agency, 4002 Narita-Cho, Oarai-machi, Higashiibaraki-gun, Ibaraki Pref., 311-1393, Japan (Corresponding
author), e-mail: ito.yuzuru@jaea.go.jp
2
Hachinohe Institute of Technology, Hachinohe Institute of Technology, 88-1 Ohbiraki, Myo, Hachinohe,
Aomori Pref., 031-8501, Japan.
3
Japan Atomic Energy Agency, IFMIF Target and Test Facilities Development Group, Japan Atomic Energy
Agency, 4002 Narita-Cho, Oarai-machi, Higashiibaraki-gun, Ibaraki Pref., 311-1393, Japan.
4
Houston, TX.
Copyright V
of K ¼ 30 MPaHm was then applied to the specimen for about 337 h at room
temperature in air, and the rate of crack growth was measured. Cleavage
features were observed on grain facets between the pre-cracking area and the
post-cracking area in the hydrogen-charged specimen. It seemed that the
feature formed by the crack proceeded by two steps in one grain facet. It is
believed that the feature formed is typical of hydrogen embrittlement. Therefore,
the small-size specimen used in this study is thought to be a useful one for
evaluating the effect of hydrogen on crack behavior in F82H steel.
Keywords
small-size specimen, crack growth, IFMIF
Introduction
Structural materials and functional materials from fusion nuclear reactors will be
exposed to neutrons with energies up to about 14 MeV at about 3.5 MW/m2
(5.0 MW/m2 at the peak) with a fluence up to more than 10 MW s/m2 during
the operation of a fusion demonstration reactor (DEMO). Radiation damage of
materials in a fusion reactor environment can be characterized by synergistic
effects of displacement damage and nuclear transmutation products such as
hydrogen and helium atoms [1–6]. This damage will induce the degradation of
mechanical properties. In order to safely operate fusion nuclear reactors, the
detailed behavior of material degradation with respect to a 14 MeV neutron dose
must be known. The International Fusion Materials Irradiation Facility (IFMIF)
is a deuterium-lithium neutron source with high intensity for irradiation experi-
ments of candidate fusion reactor materials. The results of these experiments can
be used to prepare for the DEMO’s design and licensing [7]. The irradiation vol-
ume of the IFMIF is about 14 l, and the greatest displacement damage is more
than 20 displacements per atom per year in a volume of 0.5 l. About 1000 speci-
mens need to be irradiated in the IFMIF high flux test module (HFTM) within a
limited volume of about 0.5 l, as recommended in Ref 8. Therefore, we have to
use small-size specimens, and a small-specimen test technique or technology
(SSTT) is very important [9]. In the IFMIF/EVEDA, some SSTTs such as tests
for fracture toughness, fatigue, and fatigue crack growth rate (CGR) measurement
are performed. Because the CGR measurement is one of the material tests at the
HFTM, it is necessary to develop small-size specimens for measuring the crack
growth. However, it is necessary that the size of the “crack” (i.e., the plastic zone)
be small relative to the remaining ligament, a condition referred to as small-scale
yielding. Thus, the developed specimen should satisfy a small-scale yielding con-
dition even when downsized.
It is well known that stress corrosion cracking (SCC)—that is, environmentally
assisted cracking (EAC)—is the one of the important concerns in a fission reactor.
ITO ET AL., DOI 10.1520/STP157620140026 211
Many studies have been performed to examine the mechanisms of EAC in order to
enhance the safety of reactors. In a fusion reactor, it is predicted that EAC is not a
serious concern, because the blanket is designed to be replaced in order to maintain
the reactor’s performance. However, it is reported [10] that research on the suscep-
tibility of the blanket to EAC should be required because it is damaged by the heavy
irradiation in the fusion reactor. For example, it was reported [11] that transgranu-
lar SCC occurred in irradiated F82H specimens during a slow strain rate test
(SSRT) at 300 C in hydrogenated water or in oxygenated water. Moreover, it is
reported [12] that more hydrogen would be generated by nuclear reactions than by
corrosion reactions. This means that hydrogen embrittlement (HE) might be of
concern in fusion reactors. Regardless of whether the hydrogen is generated by cor-
rosion or nuclear reaction, during reactor operation HE will not occur, because the
hydrogen can easily diffuse under such high-temperature conditions. However, the
diffusivity of hydrogen will decrease after the reactor has shut down, and hydrogen
may then accumulate around defects, dislocations, and crack tips in the metal.
Therefore, it is necessary to examine the HE susceptibility of F82H steel for that
situation. The objective of this study was to develop a small-size specimen and a
compatible test technique in order to evaluate the effect of hydrogen on crack
growth in F82H steel, and then to confirm the specimen validity for CGR measure-
ment. Tests on specimens subjected to cathodic hydrogen charging were performed
to examine the effect of hydrogen on crack growth in F82H steel.
Materials and Methods

MATERIAL
F82H steel, supplied by the Japan Atomic Energy Agency, was used to perform
crack growth tests. The F82H steel was from F82H-IEA heat no. 9753 [13]. The
chemical composition of the material is summarized in Table 1. The yield stress of
the F82H steel was 607 MPa at room temperature (RT).
SPECIMENS
For mechanical tests that use small specimens, high accuracy controllability in the
applied stress and displacement is required [9]. Therefore, along with considera-
tions for manipulating the irradiated specimen in the hot cell, it was preferable that
the loading mechanism and method for the specimen be simple in this study. The
advantages and disadvantages of some major specimen shapes are summarized in
Table 2. Although compact tension (CT) specimens are common in fracture
mechanics, wedge opening load (WOL) specimens were used in this study, because
only the loading bolt and pin are required for fixturing, and only tightening of the
loading bolt is necessary to apply a load on the specimen. It would be simpler to
perform the test if the WOL specimen were set up with the loading bolt and pin
before irradiation, as merely tightening the loading bolt would start the crack
growth test.
212
TABLE 1 Chemical composition of F82H-IEA heat (weight percent).
C Si Mn P S Cu Ni Cr Mo V Nb B Total Soluble Co Ti Ta W
Nitrogen Alumi-
(To.N) num
(Sol.Al)
0.09 0.07 0.1 0.003 0.001 0.01 0.02 7.84 0.003 0.19 0.0002 0.0002 0.007 0.003 0.003 0.004 0.04 1.98
ITO ET AL., DOI 10.1520/STP157620140026 213
TABLE 2 Summary of the advantages and disadvantages of the specimens.
Specimen Shape Advantages Disadvantages Remarks
Smooth round Relatively possible to The strain rate (i.e., loading A smooth round bar
bar specimen reduce the testing time rate to the specimen) specimen is usually
because of testing by the affects the SCC test result. employed in a tensile test,
rising load. in this instance in SSRT
testing.
The specimen shape is Unfavorable for assessing
rather simple. hydrogen embrittlement
because of a uniaxial stress
field formed at the
specimen.
CT specimen Common specimen shape The loading device and its A CT specimen is employed
in fracture mechanics. fixtures are necessary to to examine the irradiation
apply a load to the assisted stress corrosion
specimen. cracking (IASCC) by using
the capsule at the Japan
materials testing reactor
(JMTR).
CDCB specimen Possible to measure CGR The loading device and its A contoured double cantile-
under constant K condition. fixtures are necessary to ver beam (CDCB) specimen
apply a load to the is employed to measure the
specimen. SCC growth rate in the
welded metal under light
water reactor (LWR) water
conditions.
Special attention is
necessary to fabricate the
specimen because of its
complex geometry.
CBB specimen The specimen shape is very Preferable for assessing A creviced bent beam
simple. whether SCC occurred. (CBB) specimen is
employed to assess the
resistance against SCC in
the irradiated material.
The fixtures are necessary, Unfavorable for measuring
but they are not so massive. CGR.
WOL specimen The loading mechanism of The loading condition will A WOL specimen is the
the specimen is quite be mild because the value proposed specimen shape
simple. of K decreases with crack in this study.
growth.
Can be used for assessing
whether SCC occurred and
for measuring CGR.
FIG. 1 The 0.25T-WOL specimen with a thickness of 6.35 mm.
The WOL specimens were designed in accordance with the Standard Test
Method for Stress Corrosion Cracking [14]. A 0.25T-WOL specimen with a
thickness of 6.35 mm, shown in Fig. 1, was adapted as a small-size specimen. A stress
intensity factor K of 30.6 MPaHm was applied to the 0.25T-WOL specimens in
this study. K for the WOL specimen was calculated by applying a displacement by
tightening the loading bolt and assessing the crack length at that moment. The dis-
placement at the mouth of the crack was monitored by a crack opening displace-
ment (COD) gauge. The required displacement for applying the stress intensity
factor K0 with crack length a0 is calculated with Eq 1 [14].
2 3
a0
K0 pffiffiffiffiffi6C6 W 7
(1) V0 ¼ a0 4 a 5
E C3
0
W
where:
V0 ¼ displacement at the mouth of the crack and is as same as the COD,
W ¼ width of the specimen,
E ¼ elastic modulus, and
C6(a0/W) and C3(a0/W) ¼ coefficients.
The coefficients C6(a0/W) and C3(a0/W) are calculated from Eqs 2 and 3.
a a a 2 a 3 a 4
(2) C3 ¼ 30:96 195:8 þ 730:6 1186:3 þ 754:6
W W W W W
a a a 2 a 3 a 4
C6 ¼ exp 4:495 16:130 þ 63:838 89:125 þ 46:815
W W W W W
(3)
For WOL specimens, it is noted that K decreases as the crack propagates because of
the constant COD, as shown in Fig. 2. The reduction of K due to crack propagation
ITO ET AL., DOI 10.1520/STP157620140026 215
FIG. 2 Comparison of the variation of K due to crack propagation for a WOL specimen
and a CT specimen.
is estimated to be about 11 % for 1.0 mm of crack propagation. For comparison, the

variation of K due to the crack propagation of a CT specimen tested under constant
force is also plotted in Fig. 2. In the case of CT specimens, CGR test are generally per-
formed under constant force. K increases about 6 % at 1.0 mm of crack propagation.
FATIGUE PRE-CRACKING
After the WOL specimen was fabricated, a fatigue pre-crack of about 0.5 mm
was introduced at RT in air. Because we planned to perform the crack growth
test at K ¼ 25 MPaHm, the maximum K of the fatigue pre-cracking was
20 MPaHm, which would be about 20 % less than that of the crack growth test. The
stress intensity factor of the WOL specimen was calculated with the following
equation [14]:
pffiffiffi a a 2 a 3 a 4
P a
(4) K¼ 30:96 195:8 þ 730:6 1186:3 þ 754:6
BW W W W W
where:
P ¼ applied load,
B ¼ specimen thickness (¼ 6.35 mm), and
W ¼ specimen width (¼ 16.13 mm).
HYDROGEN CHARGING
Cathodic hydrogen charging was used to introduce hydrogen into the WOL speci-
men. The WOL specimen was immersed into the electrolytic solution and acted as
the cathodic electrode in the solution. Hydrogen was generated by the electrolysis
of water; then the hydrogen was adsorbed onto the surface of WOL specimen, and
from there it migrated into the metal.
The hydrogen charging apparatus, as shown in Fig. 3, consisted of (i) a beaker
containing the electrolytic solution into which the WOL specimen was immersed,
(ii) a thermal bath to hold the beaker and maintain a constant testing temperature,
FIG. 3 The cathodic hydrogen-charging apparatus.
(iii) a potentiostat to provide electricity to the WOL specimen and platinum mesh,
and (iv) a gas cylinder to supply nitrogen gas to the electrolytic solution in order to
degas oxygen. The oxygen degassing procedure was necessary during the hydrogen
charging process, because oxygen is generated by the electrolysis of water, as well as
hydrogen. Nickel wire with a diameter of 0.6 mm was connected to the WOL speci-
men by spot welding to conduct electricity from the potentiostat to the specimen.
Platinum mesh and wire were used as the anode in this test. The surfaces of the
WOL specimen without electrodes were coated with a manicure to electrically insu-
late between the specimen and the electrolytic solution. The electrolytic solution
was prepared according to Ref 15, and the solution temperature was maintained at
35 C during the hydrogen charging process.
The hydrogen charging process was conducted with a current density of
2 mA/cm2 with the potentiostat in a galvanic mode. The hydrogen diffusion coef-
ficient at 35 C in F82H steel was calculated as about 3.3 1011 m2/s through
extrapolation of reported data [16]. At this condition, the hydrogen concentration
at the surface of the WOL specimen was estimated as about 3.2 ppm based on the
calculated hydrogen diffusion coefficient. The hydrogen diffusion coefficient was
then used to calculate the time required for the hydrogen concentration to reach
99.99 % at a depth of 6.35 mm below the surface. The hydrogen charging time
in this study was fixed at around 600 h to exceed the required time, which was cal-
culated as around 322 h.
CRACK GROWTH TESTS

Two crack growth tests were conducted, namely, a crack growth test in air following
the hydrogen charging process (test 1) and a crack growth test in air without the
hydrogen charging process (test 2). Test 2 was performed as a baseline for this
study. The details of the testing procedures for test 1 are shown in Fig. 4, and those
ITO ET AL., DOI 10.1520/STP157620140026 217
FIG. 4 Details of the procedure for test 1.
for test 2 can be found in Fig. 5. For test 1, the hydrogen charging process was
conducted for around 600 h in the absence of any force, and then the testing stress
intensity, K ¼ 30 MPaHm, was applied for 337 h (approximately 2 weeks) at RT
in air. In contrast, test 2 was conducted with the specimen submersed in the
electrolytic solution for around 600 h (similar to the hydrogen charging process),
but without charging., Regardless of whether test 1 or test 2 was being performed,
heat tinting of the specimens was performed after crack growth tests to distinguish
the tested area and post-cracking area.
Results
The fracture surfaces obtained in tests 1 and 2 are shown in Figs. 6 and 7, respec-
tively. Intergranular fracture was prominent in both specimens in the pre-cracked
area where the crack was grown at RT in air. In samples from tests 1 and 2, it was
difficult to distinguish the crack growth tested area and the pre-cracked area by
merely observing the features of the fracture surface. In the case of the specimen
tested with the test 1 procedure, however, some “cleavage features” on grain
facets where the crack might proceed by two steps were found in front of the
FIG. 5 Details of the procedure for test 2.

FIG. 6 The fracture surface obtained in the specimen from test 1, observed (a) optically
and (b) via scanning electron microscopy.
post-cracking area. Micrographs of the cleavage features observed in the specimen

from test 1 are shown in Figs. 8 and 9. This type of fracture was not observed in the
specimen from test 2 that had not been subjected to hydrogen charging. Therefore,
this cleavage feature was observed only in the hydrogen-charged specimen. Accord-
ing to photos of the microstructure [13], the grain size of F82H steel, the material
used in this study, was about 25 to 100 lm. Some grains were slightly larger than
FIG. 7 The fracture surface obtained in the specimen from test 2, observed (a) optically
and (b) via scanning electron microscopy.
ITO ET AL., DOI 10.1520/STP157620140026 219
FIG. 8 Photomicrograph of the cleavage features observed in the test 1 specimen.
others, and large grains were rectangular with a size of about 100 by 300 lm. Thus,
the observed grain sizes in the intergranular fracture surface (e.g., Fig. 8) almost cor-
responded to the size shown in the photo of the microstructure [13].
Discussion
POSSIBLE MECHANISM TO GENERATE THE CLEAVAGE FEATURE
One of the mechanisms that could account for the cleavage feature is the HE mech-
anism. It is possible that the hydrogen did not diffuse rapidly enough to keep up
with the crack in the grain facet; that is, it might have been below the hydrogen
concentration necessary to cause fracture beyond the first step. After the critical
hydrogen concentration increased sufficiently to cause fracture in the area due to
FIG. 9 The cleavage features, in detail, obtained in the specimen from test 1.
FIG. 10 Summary of the possible mechanism for generating the cleavage feature.
hydrogen diffusion, the area fractured. This mechanism would generate the cleav-
age feature on grain facets. Figure 10 summarizes the possible mechanism for gener-
ating the cleavage feature. Therefore, the development of the cleavage feature could
be explained by HE during the crack growth test.
DEVELOPMENT OF A SMALL-SIZE SPECIMEN FOR A CRACK GROWTH

TEST IN F82H STEEL
The cleavage feature found in this study implies that the crack propagated during
the crack growth test because of HE. In other words, the crack in the small-size
specimen behaved as a proper crack should. Thus, a small-size specimen of F82H
steel for crack growth was developed in this study, as the HE phenomenon could be
detected by the specimen.
ESTIMATED CRACK GROWTH RATE OF F82H STEEL

The CGR at 30 MPaHm in water at 288 C was about 7 1011 m/s, which was
determined by dividing the measured crack length, 0.26 mm, by the test period,
1027 h [17]. Regarding the CGR for 304 stainless steel, Ref 18 reports a value of
about 1 1010 m/s at 30 MPaHm, measured in a simulated boiling water reactor
environment. This CGR for 304 stainless steel is similar in magnitude to that of
F82H steel; therefore, the value obtained for the CGR in water at 288 C is
reasonable.
Figure 11 shows that the maximum crack length observed on the fracture surface
in the specimen tested in test 1 was about 0.37 mm. Hence, the CGR obtained for
this specimen was about 3 1010 m/s at 30 MPaHm, as this test lasted for 337 h.
Figure 12 shows a graphical comparison of these CGRs obtained at K ¼ 30 MPaHm.
The CGR obtained with test 1 was faster than other CGRs.
INTERGRANULAR FRACTURE IN F82H STEEL

It was found that intergranular fracture occurred even in the fatigue pre-crack
at RT in air. Intergranular fracture at RT in air was also reported in Ref 19 for
ITO ET AL., DOI 10.1520/STP157620140026 221
FIG. 11 The maximum crack length in the specimen from test 1.
thermally treated alloy 690 (Alloy 690TT). Alloy 690TT [20–22] is a nickel-based,
high-chromium, heat-treated alloy that was developed to improve the resistance to
intergranular SCC in high-temperature waters by means of chromium carbide,
Cr23C6, which precipitates along the grain boundaries. F82H steel also contains pre-
cipitates along the grain boundaries that enhance the resistance to radiation and
help maintain toughness under high temperatures. The major precipitate in F82H
steel is Cr23C6 [23], as with Alloy 690TT. Recently, it was reported [24] that the
Cr23C6 in F82H steel may capture hydrogen with relative stability, because the com-
bined energy for hydrogen binding in Cr23C6 is about 0.58 eV. This implies that
FIG. 12 Summary for the estimated CGR.

F82H steel has the potential for hydrogen trapping in the metal even without
hydrogen charging. Therefore, it is proposed that intergranular fracture at RT in air
is due to HE, with the hydrogen supplied from the Cr23C6 precipitates. That is,
some Cr23C6 located in the plastic zone at the crack tip might release captured
hydrogen into the grain boundary or into the metal because of the effect of the
stress of the crack on Cr23C6. The released hydrogen would diffuse along the grain
boundary or in the metal lattice (or both), followed by one or more of the following
scenarios:
(a) Capture in other Cr23C6.
(b) Release from the metal.
(c) Piling up in a region of high triaxial stress where high stress, high strain,
and a high dislocation density coexist.
The hydrogen in case (c) would affect not only HE but also the intergranular
fracture at RT in air. Although Cr23C6 contains less hydrogen initially, it is reported
that the intergranular fracture in F82H steel with P added is induced by grain
boundary segregation of phosphorous [25]. Grain boundary segregation of phos-
phorous, as well as Cr23C6, was also identified on the intergranular fracture surface
in Alloy 690TT [26]. Therefore, this is also one of the reasons that intergranular
fracture was observed at RT in air in F82H steel and in Alloy 690TT.
Conclusions
A crack growth test with cathodic hydrogen charging was performed to examine
the effect of hydrogen on crack growth in F82H steel. The results obtained in this
study are summarized as follows:
1. In the case of the cathodic hydrogen-charged specimen, cleavage features on
grain facets were found in front of the post-cracking area. The cleavage
features are possibly explained by the hydrogen embrittlement (HE) mecha-
nism, which consists of hydrogen diffusion and hydrogen piling up at the
crack tip in the metal. In this scenario, the fracture would be caused by the
hydrogen as a result of the HE mechanism. We conclude that an acceptable
small-size 0.25T-WOL specimen of F82H steel for examining crack growth
was developed in this study because the HE phenomenon could be detected by
the specimen.
2. The estimated crack growth rate (CGR) of 30 MPaHm in water at 288 C
noted for the small-size specimen appears reasonable based on validity data
from the literature that are presented in this paper.
3. Intergranular fracture was observed even in the fatigue pre-crack (i.e., at room
temperature [RT] in air). Chromium carbide (Cr23C6) precipitates along the
grain boundaries in F82H steel and may influence the intergranular fracture
under fatigue crack propagation at RT in air.
ACKNOWLEDGMENTS
This study was performed as a part of the SSTT, which is part of IFMIF/EVEDA
under the Broader Approach (BA) framework.
ITO ET AL., DOI 10.1520/STP157620140026 223
References
[1] Wakai, E., Hashimoto, N., Miwa, Y., Robertson, J. P., Klueh, R. L., Shiba, K., and Jistukawa,
S., “Effect of Helium Production on Swelling of F82H Irradiated in HFIR,” J. Nucl. Mater.,
Vols. 283–287, No. 2, 2000, pp. 799–805.
[2] Wakai, E., Sawai, T., Furuya, K., Naito, A., Aruga, T., Kikuchi, K., Yamashita, S., Ohnuki, S.,
Yamamoto, S., Naramoto, H., and Jistukawa, S., “Effect of Triple Ion Beams in Ferritic/
Martensitic Steel on Swelling Behavior,” J. Nucl. Mater., Vols. 307–311, No. 1, 2002, pp.
278–282.
[3] Wakai, E., Kikuchi, K., Yamamoto, S., Aruga, T., Ando, M., Tanigawa, H., Taguchi, T., Sawai,
T., Oka, K., and Ohnuki, S., “Swelling Behavior of F82H Steel Irradiated by Triple/Dual Ion
Beams,” J. Nucl. Mater., Vol. 318, 2003, pp. 267–273.
[4] Tanaka, T., Oka, K., Ohnuki, S., Yamashita, S., Suda, T., Watanabe, S., and Wakai, E.,
“Synergistic Effect of Helium and Hydrogen for Defect Evolution Under Multi-ion Irradia-
tion of Fe–Cr Ferritic Alloys,” J. Nucl. Mater., Vols. 329–333, 2004, pp. 294–298.
[5] Taguchi, T., Igawa, N., Miwa, S., Wakai, E., Jitsukawa, S., Snead, L. L., and Hasegawa, A.,
“Synergistic Effects of Implanted Helium and Hydrogen and the Effect of Irradiation
Temperature on the Microstructure of SiC/SiC Composites,” J. Nucl. Mater., Vol. 335, No.
3, 2004, pp. 508–514.
[6] Wakai, E., Ando, M., Sawai, T., Kikuchi, K., Furuya, K., Sato, M., Oka, K., Ohnuki, S., Tomita,
H., Tomita, T., Kato, Y., and Takada, F., “Effect of Gas Atoms and Displacement Damage
on Mechanical Properties and Microstructures of F82H,” J. Nucl. Mater., Vol. 356, Nos.
1–3, 2006, pp. 95–104.
[7] International Energy Agency, IFMIF Comprehensive Design Report, International Energy
Agency, Paris, Cedex 15, France, 2004.
[8] Möslang, A., “Development of a Reference Test Matrix for IFMIF Test Modules,” For-
schungszentrum Karlsruhe, Institut für Materialforschung I., Aalen, Germany, 2006, p.
1–34.
[9] Wakai, E., Yamamto, M., Molla, J., Yokomine, T., and Nogami, S., “Design Plan and
Requirement of Test Module and Testing Items in IFMIF,” Fusion Eng. Design, Vol. 86,
Nos. 6–8, 2011, pp. 712–715.
[10] Miwa, Y., Tsukada, T., and Jitsukawa, S., “Material Issues of Blanket Systems for Fusion
Reactors—Compatibility with Cooling Water,” J. Plasma Fusion Res., Vol. 80, No. 7,
2004, pp. 551–557.
[11] Miwa, Y., Jitsukawa, S., and Tsukada, T., “Stress Corrosion Cracking Susceptibility of a
Reduced-activation Martensitic Steel F82H,” J. Nucl. Mater., Vols. 386–388, 2009, pp.
703–707.
[12] Jung, P., “Hydrogen Inventory and Embrittlement in Low Activation Steels,” J. Nucl.
Mater., Vols. 258–263, No. 1, 1998, pp. 124–129.
[13] Shiba, K., Hishinuma, A., Tohyama, A., and Kasamura, K., “Properties of Low Activation
Ferritic Steel F82H IEA Heat—Interim Report of IEA Round-robin Tests (1),” Report No.
JAERI-Tech 97-038, JAERI, Tokai-mura, Naka-gun, Ibaraki-ken, Japan, 1997.
[14] “Standard Test Method for Stress Corrosion Cracking,” 129th Committee on Strength
and Fracture of Advanced Materials, Society of Materials Science, Japan, 1985, pp. 8–17.
[15] Sakai, Y., Yamashita, M., Shiokawa, K., Niu, L. B., Kobayashi, M., and Takaku, H., “SCC
Growth Behavior of Materials for Geothermal Steam Turbine,” Zairyo-to-Kankyo, Vol. 53,
No. 3, 2004, pp. 143–148.
[16] Matsuhiro, K., Ando, M., Nakamura, H., and Takeuchi, H., “Estimation of Tritium Permea-
tion through Reduced-activation Ferritic Steel at IFMIF Target Backwall Damaged by
Neutron Irradiation,” Report No. JAERI-Research 2004-003, JAERI, Tokai-mura, Naka-
gun, Ibaraki-ken, Japan, 2004.
[17] Ito, Y., Saito, M., Abe, K., Yamamoto, M., and Wakai, E., “Development of Crack Growth
Test of F82H Steel at 288 C Using a Small Size Specimen,” 16th International Conference
on Fusion Reactor Materials, Beijing, China, Oct. 20–26, 2013.
[18] Namatame, S., Suzuki, S., Tanaka, N., Itow, M., Kuniya, J., and Shimanuki, S., “SCC Growth
Curve for Austenitic Stainless Steels in BWR In-core Environment,” JSME Annual Meet-
ing 2002(1), University of Tokyo, Tokyo, Japan, Sept 25–27, 2002, Japan Society of Me-
chanical Engineers, Tokyo, Japan, pp. 441–442.
[19] Ito, Y., Lu, Z. P., Miura, H., Yonezawa, T., and Shoji, T., “Effect of Temperature on the
Microscopic Appearance of the Fracture Surface of Alloy 690TT under SSRT Testing,”
JSME Int. J. Ser. A, Vol. 49, No. 3, 2006, pp. 355–362.
[20] Kusakabe, T., Yonezawa, T., and Tokunaga, S., “Research on Corrosion Resistance of
Steam Generator Tube,” Mitsubishi Heavy Industries Technical Review, Vol. 32, No. 3,
1995, pp. 161–164.
[21] Thuvander, M. and Stiller, K., “Microstructure of a Boron Containing High Purity Nickel-
based Alloy 690,” Mater. Sci. Eng. A, Vol. 281, Nos. 1–2, 2000, pp. 96–103.
[22] Yonezawa, T., “Stress Corrosion Cracking in Pressurized Water Reactor,” Kinzoku,
Vol. 73, No. 8, 2003, pp. 727–730 (in Japanese).
[23] Tanigawa, H., Sakasegawa, H., Shiba, K., and Hirose, T., “3. Current Status and Issues in
Manufacturing Technology of Reduced Activation Ferritic/Martensitic Steel,” J. Plasma
Fusion Res., Vol. 87, No. 3, 2011, pp. 167–171.
[24] Watanabe, Y., Iwakiri, H., Murayoshi, N., and Kato, D., “Thermal Structural Analysis for
IFMIF Lithium Target,” 29th Japan Society of Plasma Science and Nuclear Fusion
Research Annual Meeting, Tokyo, Japan, Dec. 3–6, Japan Society of Plasma Science and
Nuclear Fusion Research, Nagoya, Japan, 2013.
[25] Kim, B. J., Kasada, R., Kimura, A., and Tanigawa, H., “Evaluation of Grain Boundary
Embrittlement of Phosphorus Added F82H Steel by SSTT,” J. Nucl. Mater., Vol. 421, Nos.
1–3, 2012, pp. 153–159.
[26] Ito, Y., Lu, Z. P., Miura, H., and Shoji, T., “Fracture Behavior of the Hydrogen Charged
Alloy 690TT and Alloy 600TT,” Proceedings of the 52nd Japan Conference on Materials
and Environments, Sapporo, Japan, Sept 14–16, 2005, Japan Society of Corrosion Engi-
neering, Tokyo, Japan, pp. 231–234 (in Japanese).
NON-DESTRUCTIVE TECHNIQUES
Kathryn H. Matlack,1 Jin-Yeon Kim,2 James J. Wall,3,4

Jianmin Qu,5 and Laurence J. Jacobs2,4
Nonlinear Ultrasonic
Characterization of Radiation
Damage Using Charpy Impact
Specimen
Reference
Matlack, Kathryn H., Kim, Jin-Yeon, Wall, James J., Qu, Jianmin, and Jacobs, Laurence J.,
“Nonlinear Ultrasonic Characterization of Radiation Damage Using Charpy Impact Specimen,”
Small Specimen Test Techniques: 6th Volume, STP 1576, Mikhail A. Sokolov and Enrico Lucon,
Eds., pp. 227–243, doi:10.1520/STP157620140007, ASTM International, West Conshohocken,
PA 2015.6
ABSTRACT
Radiation damage occurs in reactor pressure vessel (RPV) steel, causing
microstructural changes such as point defect clusters, changes in dislocation
density, and precipitates. These radiation-induced microstructural changes cause
material embrittlement. Radiation damage is a crucial concern in the nuclear
industry because many nuclear plants throughout the United States are entering
the first period of life extension, and older plants are currently undergoing
assessment of technical basis to operate beyond 60 years. The result of extended
operation is that the RPV and other components will be exposed to higher levels
Manuscript received January 21, 2014; accepted for publication April 3, 2014; published online September 8,
2014.
1
G.W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United
States of America (Corresponding author), e-mail: katie.matlack@gmail.com
2
School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United
States of America.
3
Electric Power Research Institute, Charlotte, NC 28262, United States of America.
4
G.W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332,
United States of America.
5
Dept. of Civil and Environmental Engineering, Northwestern Univ., Evanston, IL 60208, United States of
America.
6
Houston, TX.
Copyright V
of neutron radiation than they were originally designed to withstand. There is

currently no nondestructive evaluation technique with which to unambiguously
assess the amount of radiation damage in RPV steels. The development of such a
technique would enable the assessment of the integrity of a vessel, allowing
operators to determine whether reactors can continue to operate safely, and
would directly support the nuclear industry Long Term Operation and U.S.
Department of Energy Light Water Reactor Sustainability initiatives. Nonlinear
ultrasound (NLU) is a nondestructive evaluation technique that is sensitive to
microstructural features such as dislocations, precipitates, and their interactions in
metallic materials. The physical effect monitored via NLU is the generation of
higher harmonic frequencies in an initially monochromatic ultrasonic wave, arising
from the interaction of the ultrasonic wave with microstructural features. Recent
research has demonstrated that NLU is sensitive to radiation-induced
microstructural changes in RPV steel. NLU measurements were made on various
Charpy specimen sets of typical RPV material to investigate the applicability of
NLU in characterizing radiation damage over a range of fluence levels, irradiation
temperatures, and material compositions. These previous experimental results are
interpreted with a newly developed analytical model that combines irradiation-
induced precipitate and vacancy contributions to the nonlinearity parameter.
Keywords
nonlinear ultrasound, nondestructive evaluation, reactor pressure vessel steel,
radiation damage
Introduction
Nonlinear ultrasonic nondestructive evaluation (NDE) methods have the potential
to characterize material damage before macroscopic damage and component fail-
ure. These methods are based on the fact that when a pure sinusoidal ultrasonic
wave propagates through a nonlinear medium, higher harmonic wave components
are produced, including a second harmonic wave. This phenomenon can be quanti-
fied using the acoustic nonlinearity parameter b. The magnitude of the acoustic
nonlinearity parameter depends on defect structures present in the material micro-
structure [1–3]. Nonlinear ultrasound (NLU) has been used previously to monitor
microstructural evolution throughout fatigue damage [4–7] and thermal aging
[2,8–11] and has recently been shown to be sensitive to radiation damage [12,13].
This has important implications in the NDE of nuclear reactor components, partic-
ularly in reactor pressure vessel (RPV) steels: because NLU techniques can detect
microstructural changes that are caused by radiation damage in RPV steels, there is
the potential to use NLU for the NDE of RPV embrittlement. There is currently no
NDE method with which to unambiguously monitor RPV embrittlement.
This paper interprets current experimental results of measured b in RPV steels
with a variety of irradiation conditions [12,14] with newly developed models to
explain measured differences in b in low-Cu steel relative to that in medium-Cu
steel. The measurement technique is currently applied to Charpy and broken
MATLACK ET AL., DOI 10.1520/STP157620140007 229
Charpy impact samples, as much of the irradiated material specimen database is in

this form. Further, surveillance specimens, typically with geometry similar to that
of Charpy specimens, were placed in nuclear reactors at the start of operation and
removed periodically for testing. Thus, there is the potential to utilize these devel-
oped methods on these samples as reactors are continually operated for longer
times so as to give insight into low-flux, high-fluence effects.
In this paper, current knowledge of the microstructural evolution of irradiated
RPV steels is reviewed, and then current models describing the contribution to b of
different microstructural features are discussed. Experimental results are then pre-
sented and analyzed to show expected dominating microstructural features in the
nonlinear response.
Radiation-induced Microstructural Changes

in RPV Steels
COPPER-RICH PRECIPITATES
In materials with copper contents greater than about 0.1 wt. %, copper has been
shown to be more efficient in forming precipitates in irradiated steels than other
alloying elements [15]. At RPV operating temperatures of around 290 C, the dis-
solved copper in the matrix is supersaturated. Copper-rich precipitates (CRPs)
nucleate and grow as a result of radiation-enhanced diffusion (RED), and they pin
dislocations, causing hardening [16,17]. CRPs can also be enriched with Mn, Ni, Si,
and P [18], and larger amounts of these elements in RPV material lead to increased
nucleation rates and number densities of CRPs [18]. These precipitates have been
shown to dominate early embrittlement in RPV steels containing more than about
0.05 to 0.1 wt. % Cu, and their nucleation and growth saturates at higher fluence as
Cu is depleted in the matrix [16]. It has been shown that an increase in flux shifts
the saturation of the CRP contribution to hardening to higher fluence levels [19]; in
other words, higher flux slows the rate of precipitation of CRPs. RPV steels with Ni
and Mn contents contain CRPs that are alloyed with Mn and Ni, both of which act
to increase the volume of the precipitates [16]. It has also been shown that the
addition of P acts to increase the number density of precipitates [18]. Atom
probe tomography has shown, for example, that irradiation up to a fluence of
5 1023 n/m2 (E > 1 MeV) in ASTM A533 steel with 0.14 wt. % Cu produces a
large number density of CRPs, on the order of 3 1023 n/m2, with an average radius
of about 1 nm [20]. The composition of these precipitates was found to consist of
Fe and Cu, enriched with Ni, Mn, Si, and P. These CRPs have also been visualized
through small-angle neutron scattering (SANS) techniques [21].
MATRIX FEATURES
Matrix features that form during damage cascades are divided into two separate cat-
egories: unstable matrix defects (UMDs) and stable matrix features (SMFs). UMDs
are primarily vacancy clusters that can form in steels even with little to no Cu con-
tent, and they dissolve quickly at typical reactor operating temperatures of 290 C
[16]. However, at high fluxes typical of test reactors (>1016 n/m2 s), these features
can become significant. The increased amount of UMDs at high fluxes actually act
as sinks, reducing RED and thus delaying CRP nucleation and growth [16]. They
can further act as nucleation sites for SMFs. SMFs can include dislocation loops;
vacancy-solute clusters; dislocation atmospheres; and small precipitates enriched
with Mn, Ni, Si, or P (or some combination of these) [18]. These matrix features
contribute to embrittlement, which generally increases with decreasing irradiation
temperature and increases with the square root of the fluence [22], as they are
weaker obstacles to dislocation motion.
MANGANESE–NICKEL PRECIPITATES
In materials with elevated levels of Mn and Ni, or with low levels of Cu,
manganese–nickel-rich precipitates (MNPs) have been shown to form in the irradi-
ated microstructure [18,23,24]. Lower irradiation temperatures promote the forma-
tion of these MNPs [16], which have been found in larger volume fractions than
CRPs [25]. Pure MNPs form at a much slower rate than CRPs, and are thus referred
to in the literature as “late-blooming phases” [18,22]. Characterization of these fea-
tures and their embrittling effects is still an ongoing area of research, as it has yet to
be determined whether these phases could contribute to increased embrittlement at
low flux but very high fluence, which would be a concern for the extended opera-
tion of nuclear reactors.
PHOSPHOROUS PHASES AND SEGREGATION

It has been shown that phosphorous can contribute to radiation embrittlement
[17,18,23,26,27]. RED promotes P diffusion and nucleation of phosphide phases as
a result of its low solubility in RPV steels, even though P is typically found in very
small quantities (<0.05 wt. %) [17,18]. There is strong interaction between P and
Mn elements, which can potentially form Mn3P phases. Phosphorous has been
shown to increase the CRP contribution to irradiation hardening in some cases, but
this effect becomes less pronounced in RPV steels with much greater Cu contents
[18]. It has also been shown that irradiation causes phosphorous segregation to dis-
locations and grain boundaries, which can contribute to irradiation hardening
[18,28]. However, P segregation to grain boundaries leads to brittle intergranular
fracture, which has not been seen as a dominant failure mechanism in light water
reactor RPV steels (typical U.S. RPV materials).
Theoretical Models for Nonlinear Ultrasound

It is known that the acoustic nonlinearity parameter b is a measurable material
parameter. When a single-frequency sinusoidal wave propagates through a nonlin-
ear medium, a second harmonic wave is generated. This second harmonic wave is
related to b. Metallic materials have a weak but nonzero and measurable material
nonlinearity. This parameter is known to relate to acoustic properties of a propagat-
ing wave.
8A2
(1) b¼ 2 2
A1 xk
where:
A2 ¼ amplitude of the generated second harmonic wave,
A1 ¼ amplitude of the first harmonic wave,
x ¼ wave propagation distance, and
k ¼ wavenumber of the propagating wave.
The parameter b depends on the crystalline structure of the material, and also
on the localized strain present in the material. This strain arises from microstruc-
tural features such as dislocations and precipitates. This section provides a compre-
hensive review of different microstructural contributions to the magnitude of the
acoustic nonlinearity parameter that are relevant to radiation damage in RPV steel
materials at relevant levels of neutron fluence. Theoretical derivations of contribu-
tions of dislocation pinning [1,29], precipitate-pinned dislocations [2,8,29,30], and
vacancies [31] are reviewed in the following sections.
DISLOCATION PINNING
The effect of dislocation pinning on the magnitude of the acoustic nonlinearity
parameter was initially based on work by Hikata [1] and was expanded on later by
Cantrell [29]. The interested reader is referred to this literature for more in-depth
details.
Consider a dislocation line segment pinned between two points a distance 2L
apart. These pinning points can be grain boundaries, other dislocations, or point
defects in the material. Assume that a small but nonzero longitudinal stress r with
a shear component s such that s ¼ Rr, where R is the resolving shear factor, is then
applied to this dislocation segment so that it bows out between the two pinning
points. This geometry is depicted in Fig. 1, where the radius of curvature r of the
FIG. 1 Diagram showing geometry of a bowed dislocation segment of length 2L

between two pinning points and under an applied shear stress. The radius of
curvature r and angle h are also depicted.
bowed segment and the angle h are noted. Note that this stress can be thought
of as either an internal residual stress or an externally applied stress, but it is
small enough that the dislocation segment does not break away from the pinning
points.
It is further assumed that the dislocation density is small enough that bowed
dislocations act independently of each other. An ultrasonic stress is then superim-
posed on the internal stress. The change in magnitude of the ultrasonic nonlinearity
parameter due to a distribution of pinned dislocations bd has been shown to be
24 XKL4 R3 E2
(2) Dbd ¼ jrj
5 G3 b2
where:
X ¼ conversion factor between shear and longitudinal strain,
K ¼ dislocation density,
L ¼ dislocation loop length or pinning length,
R ¼ resolving shear factor,
E ¼ Young’s modulus,
G ¼ shear modulus,
b ¼ Burger’s vector, and
r ¼ residual or internal stress.
PRECIPITATE CONTRIBUTION
Now consider the effect of a distribution of precipitates on the magnitude of b. Pre-
cipitates themselves do not have a significant effect on b [30], but their interaction
with dislocations has been shown to give rise to a significant change in the magni-
tude of b [2,8,30]. As in previous work [29], we can approximate this as a spherical
precipitate embedded in an isotropic medium, exerting a nonzero internal pressure
p0 on the matrix. We then assume there is a distribution of these spherical precipi-
tates embedded in a microstructure with dislocations present. The precipitates exert
a local stress field as described above in the vicinity of the precipitate. Because a dis-
location line is assumed to follow a contour of minimum energy, it is assumed that
two precipitates a distance L/2 away from each dislocation segment act on each dis-
location segment, and contributions from other nearby precipitates are negligible.
This scenario is depicted in Fig. 2.
The stress on a dislocation segment a distance r from the precipitate is given by

4G 3Bp rp 3
(3) rp ¼ dp
3Bp þ 4G r
where:
Bp ¼ Bulk modulus of the precipitate,
rp ¼ radius of the precipitate, and
dp ¼ misfit parameter of the precipitate.
FIG. 2 Diagram of a dislocation bending around a distribution of precipitates, with

spacing of L.
This stress is evaluated at r ¼ L/2, and it is assumed that two precipitates act on
a single dislocation segment. The resulting change in the magnitude of ultrasonic
nonlinearity due to a distribution of precipitates in a dislocated microstructure
Dbpd is then [2,8,30]

1536 XKR3 E2 3Bp
(4) Dbpd ¼ 3 2
dp Lrp3
5 Gb 3Bp þ 4G
where:
rp ¼ average precipitate radius.
Note that the loop length L is related to the number density of precipitates in
the microstructure as L / 1=N 1=3 , such that Dbpd depends on both the radius and
the number density of precipitates.
VACANCY CONTRIBUTION
Similar to precipitates, the stress field surrounding a vacancy and vacancy clusters
rv can interact with the stress field of the dislocation, causing a change in the mag-
nitude of the acoustic nonlinearity parameter [3]. The corresponding stress due to a
vacancy acting on the dislocation at a distance r has been shown to be [32]

4Gð1 þ Þ rv 3
(5) rv ¼ dv
3ð1 Þ r
where:
¼ Poisson’s ratio,
rv ¼ radius of the vacancy, and
dv ¼ vacancy misfit parameter.
The change in magnitude of the ultrasonic nonlinearity due to a distribution of
vacancies in a dislocated microstructure Dbvd is then [3]

512 XKR3 E2 1 þ 3
(6) Dbvd ¼ Lrv dv
5 G3 b2 1
TABLE 1 Chemical composition (wt. %) of JRQ and JFL steel.
Material C Si Mn Cr Mo Ni P Cu S
JRQ 0.18 0.24 1.42 0.12 0.51 0.84 0.017 0.14 0.004
JFL 0.17 0.25 1.42 0.16 0.52 0.75 0.004 0.01 0.002
Experimental Methods
MATERIAL SAMPLES
Two low-alloy RPV steels were investigated: (1) a medium-copper steel, ASTM
A533B class 1 (JRQ), and (2) a low-copper steel, forged ASTM A508 class 3 (JFL).
Chemical compositions (in weight percent) are given in Table 1; note the difference
in Cu content between the two materials, as well as the greater P content in JRQ.
One specimen set consisted of broken Charpy samples of JRQ material that
had been irradiated at a temperature of 290 C up to a neutron fluence of
5.0 1023 n/m2 (E > 1 MeV) at a flux of about 5.0 1016 n/m2 s. These samples
were previously irradiated in the 10 MW (t) SAPHIR reactor and were reported on
previously [28,33]. Two samples in this set (IAR,0.5 and IAR,1.7) were annealed at
50 % of the total neutron fluence, with an annealing schedule of 460 C for 18 h.
This means that, for example, sample IAR,0.5 was irradiated to 0.25 1023 n/m2,
then annealed at 460 C for 18 h, and then re-irradiated to 0.25 1023 n/m2, for a
total neutron fluence of 0.5 1023 n/m2.
The second specimen set consisted of intact Charpy samples of JRQ and JFL
material irradiated at a temperature of 255 C up to nearly 10 1023 n/cm2. These
samples were irradiated at the Rheinsberg power reactor and were part of a previous
International Atomic Energy Agency investigation [34,35]. The specific conditions
for each specimen are given in Table 2. An unirradiated sample from each
specimen set, and material was also measured to provide the baseline measurement
of b0.
TABLE 2 Irradiation conditions for material samples investigated.
Irradiation Sample Total Neutron Annealing at Neutron

Material Temperature, C Designation Fluence (E > 1 MeV), n/m2 50 % Fluence Flux, n/m2 s
JRQ 290 IAR,0.5 0.5 1023 460 C/18 h 5.0 1016

IAR,1.7 1.7 1023 460 C/18 h 5.0 1016
I,5 5.0 1023 — 5.0 1016
JRQ 255 I,5.4 5.4 1023 — 3.0 1016
I,9.8 9.8 1023 — 5.4 1016
JFL 255 I,5.1 5.1 1023 — 2.8 1016
I,8.6 8.6 1023 — 4.7 1016
NONLINEAR ULTRASONIC MEASUREMENTS

To measure the acoustic nonlinearity parameter in the irradiated Charpy samples,
longitudinal waves were excited through the thickness of the specimens. Two trans-
ducers were mounted on opposite sides of the Charpy sample and accurately
aligned and clamped on the specimen with a specially designed fixture [13]. A high-
power gated amplifier (RITEC GA2500A) was used to amplify a sinusoidal wave
from a function generator (Agilent 33250A) into the transmitting transducer. Exci-
tation frequencies ranged from 2.25 MHz to 3.3 MHz, and the transmitting trans-
ducer used was a narrowband transducer with a center frequency around that of
the transmitted wave. The ultrasonic wave propagated through the specimen where
the nonlinearities (microstructural features) in the material generated the second
harmonic wave. On the opposite side of the specimen, a second transducer tuned to
the second harmonic wave frequency received the signal, which consisted of both
the first and the second harmonic wave. The signal was transferred to an oscillo-
scope (Tektronix TDS 5034B) and then to a computer for post-processing. An
experimental schematic of this setup is shown in Fig. 3. The frequency response
was calculated using the fast Fourier transform (FFT) of a Hanning-windowed por-
tion of the signal, and from this the first and second harmonic wave amplitudes, A1
and A2, respectively, could be measured. A representative time signal (with points
representing the bounds of the Hanning window) and frequency response are
shown in Fig. 4. This process was repeated with different output voltage levels of
the amplifier to obtain ten separate measurements of A1 and A2, which gave a single
measurement of the relative acoustic nonlinearity parameter b. Note that b is pro-
portional to A2/A1 [2], such that the linear fit of the measured data for A2 versus A1
[2] is proportional to b. A representative linear fit is shown in Fig. 5. This process
was repeated three times for each sample, with the sample and transducers com-
pletely removed from the measurement fixture and remounted, and the
FIG. 3 Schematic of experimental setup for nonlinear ultrasonic measurements.

FIG. 4 Representative time signal showing the Hanning window bounds (left) and FFT
showing first and second harmonic wave amplitudes (right).
measurement results were averaged. Multiple samples were measured at each irradi-
ation condition.
Experimental Results
Current experimental values of measured b over neutron fluence for JRQ and JFL
material irradiated at 290 C and 255 C are shown in Fig. 6. These results were pre-
sented previously [12,14] and are shown here for clarity in terms of comparison
with the developed analytical models. Data points indicate the average measure-
ment, and the error bars shown represent 1 standard deviation from the mean.
Measurements on JRQ and JFL at 255 C were averaged over three different sam-
ples, and measurements on JRQ at 290 C were averaged over two samples as well
FIG. 5 Linear fit of measured A2 and A1 [2], where the slope of the linear fit is
proportional to b.
FIG. 6 Measured b for JRQ and JFL Charpy specimens for increasing neutron fluence
[14].
as two different locations on each sample. The error bars on JRQ at 290 C are diffi-
cult to see because they are very small (maximum standard deviation of 64 %).
The error bars on JRQ and JFL at 255 C have been adjusted for surface roughness
effects, as described and detailed previously [13]. The measurement of b0 for each
sample set was determined from a nonlinear ultrasonic measurement on an unirra-
diated sample from each sample set, and the data for the three sample sets were
normalized independently. Thus, the results from different sample sets can be com-
pared in terms of the change in b.
The experimental results showed an increase in b up to about 5 1023 n/m2
(E > 1 MeV) for both JRQ and JFL and for the different irradiation temperatures.
This increase was stronger in JRQ than JFL and in the lower temperature irradia-
tions than the higher temperature ones. At the higher fluence levels of JRQ and JFL
specimens irradiated at 255 C, b decreased. Previous SANS measurements on the
JRQ and JFL specimens irradiated at 255 C showed an increasing volume fraction
of precipitates (fp) with increasing neutron fluence, with an average radius (rp) of
about 1 nm. From these data, the number density of precipitates can be calculated
through the relation Np ¼ fp =Vp , where Vp is the volume of a precipitate, which is
assumed to be spherical. Measurements of Charpy transition temperature shifts and
changes in yield stress have also been characterized previously for these specimens.
The NLU results, along with SANS results, DT41J, and changes in yield stress, are
tabulated in Table 3. Note that the volume fractions given in this table are in units
of percent volume.
It should be pointed out that the three datasets presented in Fig. 6 are
extremely difficult to compare because they include different irradiation tempera-
tures and materials—in particular a forged material (JFL) and a rolled plate (JRQ).
The results are presented on the same plot to show, in a concise manner, the differ-
ent trends measured for b.
TABLE 3 Summary of data for JRQ and JFL irradiated at 290 C and 255 C.
Change in Yield
Sample b/b0a Stress,b MPa DT41J, Cb fv,c vol. % rp ,c nm Np,c (1/m3)
JRQ/290C/IAR,0.5 1.07 1 27 n/a n/a n/a

JRQ/290C/IAR,1.7 1.09 20 56
JRQ/290C/I,5.0 1.18 111 96
JRQ/255C/I,5.4 1.97 286 179 0.34 1.0 8.12 1023
JRQ/255C/I,9.8 1.61 358 221 0.50 1.0 11.94 1023
JFL/255C/I,5.1 1.65 117 52 0.02 1.0 0.48 1023
JFL/255C/I,8.6 1.04 170 78 0.09 1.0 2.15 1023
a
References 12 and 14.
b
Reference 33.
c
Reference 21.
Analysis and Discussion

The nonlinear ultrasonic results have been analyzed previously in terms of the gen-
eral dislocation pinning model as given in Eq 2 and the precipitate-pinned disloca-
tion model as given in Eq 4. It has been suggested that these models can be
combined to describe the change in b with an increasing number density of precipi-
tates, and that there is a critical number density of precipitates needed in order for
all dislocations to be pinned by the precipitates. Below this critical number density,
b can increase with an increasing number density of precipitates. The details of this
model can be found elsewhere [14]. This analysis attempts to explain the different
magnitudes of measured b for different irradiation temperatures, and also the
increasing and decreasing trends of b over increasing neutron fluence; essentially it
is suggested that this critical number density of precipitates occurs around a neu-
tron fluence of 5 1023 n/m2. Preliminary analysis suggested that the smaller
increase in b in JFL could be attributed to the matrix features likely to have formed
in this low-copper steel, which act as pinning points to the dislocations. This is in
comparison to the much higher number density of precipitates that formed in JRQ.
The following considers the effects on b of precipitates relative to vacancy-type
features.
VACANCIES VERSUS PRECIPITATES

The contributions to the acoustic nonlinearity parameter for precipitates and
vacancies are given in Eqs 4 and 6, respectively. To isolate the effects of each of
these microstructural features on the magnitude of the ultrasonic nonlinearity
parameter, we compare a dislocated microstructure with purely vacancies to one
with purely precipitates, assuming the same number density of defects in each case.
In this way, and assuming a random distribution of defects, we can assume the
dislocation pinning or loop length in Eqs 4 and 6 is equivalent. Note that both these
equations are based on Eq 2. Thus, to compare the contributions to b of vacancies
and precipitates, it is only necessary to consider the difference in stress produced by
these microstructural features (i.e., the ratio of Eqs 3 and 5). To evaluate the relative
contributions of these two features, the ratio of |rv| to |rp| is considered, which is
!
rv ð1 þ Þ 3Bp þ 4G jdv j rv3
(7) ¼
rp 9ð1 ÞBp dp rp3
In this work we are considering low-alloy steels for which ¼ 0.33 and
G ¼ 80 GPa. The misfit parameter for vacancies is known to range from 0.1 to 0
[32], and the radius is rv b/2 ¼ 0.1435 nm. In RPV material, copper precipitates
have been shown to form during radiation damage [16,25]. If we assume the precip-
itate is purely copper, the bulk modulus of the precipitate is thus the bulk modulus
of copper, Bp ¼ 140 GPa, and the misfit parameter is dp ¼ 0.024 [36]. The value of
the first bracketed term in Eq 7 representing the material properties is not much
more than 1 even if the precipitate composition is somewhat softer than copper.
Therefore the ratio of the misfit parameters will vary from 0 to 5, depending on the
vacancy misfit parameter. However, the ratio in Eq 7 is clearly dominated by the
relative ratios of the radii of the vacancies and precipitates. The precipitate radius
has been reported to be on the order of 1 nm in irradiated RPV steel, so the ratio
(rv/rp) [3] is on the order of 0.001. So, the precipitate contribution to b will domi-
nate over the vacancy contribution.
TWO-FEATURE PINNING MODEL

Assume now that a microstructure contains dislocations and two types of disloca-
tion pinning features: vacancies and precipitates. Assume all pinning points are uni-
formly distributed, such that the distance between pinning points is roughly the
same throughout the microstructure. A diagram of this microstructure is depicted
in Fig. 7, with the particle spacing L indicated. Both features will exert a local stress
FIG. 7 Two-feature pinning model: diagram of dislocation pinning by two different

microstructural features, precipitates and vacancies.
field in the vicinity of a dislocation segment; rp is exerted by the precipitates, and rv

is exerted by the vacancies.
Assuming some number density of precipitates Np and some number density of
vacancies Nv, the average stress ravg on a dislocation segment can be expressed as
Np rp þ Nv rv
(8) ravg ¼
Np þ Nv
The previous analysis showed that the stress field of a precipitate will likely be
much larger than that of a vacancy, because of the larger radius of the precipitate
relative to the vacancy. So, replacing some precipitate pinning points with vacancy
pinning points will reduce the internal stress acting on the dislocation segment.
In low-Cu materials, the dominant hardening mechanism is matrix features
such as vacancy complexes [18,25,37]. The effects on b of these features would
more closely approximate the vacancy contribution than the precipitate contribu-
tion. In these materials, in which the number density of CRPs is significantly low,
one might expect a stronger contribution to b than in a medium-Cu material with a
significantly greater number density of CRPs. However, this is unlikely to be the
case, as other matrix features are likely present in the low-Cu material that (1) pro-
vide pinning points to dislocations, which will decrease the magnitude of b, and (2)
would result in a lower local stress field relative to the precipitates, as evident in the
ratio in Eq 7 and the average stress field given in Eq 8. This is supported by experi-
mental evidence: a greater increase in acoustic nonlinearity was measured in JRQ
(0.14 wt. % Cu) than in JFL (0.01 wt. % Cu), even though the measured number
density of precipitates was an order of magnitude greater in JRQ than in JFL [21].
Because b is driven by changes in microstructure, it is important to understand
the microstructural changes that drive the changes measured with b in the irradi-
ated material. The ultimate goal of a nonlinear ultrasonic NDE system would not
necessarily be to resolve the effects of vacancies and precipitates; the ultimate goal
is to relate the measurement of b to the amount of radiation-induced embrittle-
ment. Having a clear understanding of the microstructural influences of b in the
irradiated material under a variety of conditions is the first step in that direction.
Summary and Conclusions

Nonlinear ultrasonic methods have been shown to be sensitive to radiation damage
in RPV steels. Currently, the operating life of nuclear reactors throughout the
United States is being extended, yet there is a lack of an appropriate NDE technique
with which to monitor radiation-induced embrittlement of an RPV. Nonlinear
ultrasound (NLU) is sensitive to microstructural changes such as dislocations, pre-
cipitates, and vacancies, which are related to generated higher harmonic waves,
quantified using the acoustic nonlinearity parameter b. This work reviews current
experimental results of measured b over increasing neutron fluence in Charpy
specimens of different RPV materials and over different irradiation temperatures.
Analysis of the influence of precipitates and vacancies on the magnitude of b shows

that because precipitates generate a higher local stress field in the vicinity of the dis-
location, the influence on b is stronger. This effect results in a greater measured
value of b for precipitate features than for vacancy-type features, which is shown
experimentally in the results for JRQ (medium-copper steel) relative to those for
JFL (low-copper steel).
ACKNOWLEDGEMENTS
This research is being performed using funding received from the DOE Office of Nu-
clear Energy’s Nuclear Energy University Programs. Additional funding has been pro-
vided by the Electric Power Research Institute (EPRI). This work was also supported
by the National Science Foundation through a Graduate Research Fellowship to
Kathryn Matlack.
References
[1] Hikata, A., Chick, B. B., and Elbaum, C., “Dislocation Contribution to the Second
Harmonic Generation of Ultrasonic Waves,” J. Appl. Phys., Vol. 36, No. 1, 1965, pp.
229–236.
[2] Cantrell, J. H. and Yost, W. T., “Determination of Precipitate Nucleation and Growth
Rates from Ultrasonic Harmonic Generation,” Appl. Phys. Lett., Vol. 77, No. 13, 2000, pp.
1952–1954.
[3] Cantrell, J. H., “Substructural Organization, Dislocation Plasticity and Harmonic Genera-
tion in Cyclically Stressed Wavy Slip Metals,” Proc. R. Soc. London Ser. A Math. Phys.
Eng. Sci., Vol. 460, No. 2043, 2004, pp. 757–780.
[4] Cantrell, J. H. and Yost, W. T., “Nonlinear Ultrasonic Characterization of Fatigue Micro-
structures,” Int. J. Fatigue, Vol. 23, 2001, pp. S487–S490.
[5] Frouin, J., Sathish, S., Matikas, T. E., and Na, J. K., “Ultrasonic Linear and Nonlinear
Behavior of Fatigued Ti-6Al-4V,” J. Mater. Res., Vol. 14, No. 4, 1998, pp. 1295–1298.
[6] Kim, J.-Y., Jacobs, L. J., Qu, J., and Littles, J. W., “Experimental Characterization of
Fatigue Damage in a Nickel-base Superalloy Using Nonlinear Ultrasonic Waves,”
J. Acoust. Soc. Am., Vol. 120, No. 3, 2006, pp. 1266–1273.
[7] Walker, S. V., Kim, J.-Y., Qu, J., and Jacobs, L. J., “Fatigue Damage Evaluation in
A36 Steel Using Nonlinear Rayleigh Surface Waves,” NDT E Int., Vol. 48, 2012, pp.
10–15.
[8] Hurley, D. C., Balzar, D., and Purtscher, P. T., “Nonlinear Ultrasonic Assessment of Precipi-
tation Hardening in ASTM A710 Steel,” J. Mater. Res., Vol. 15, No. 9, 2000, pp.
2036–2042.
[9] Lara, N. O., Ruiz, A., Rubio, C., Ambriz, R. R., and Medina, A., “Nondestructive Assessing
of the Aging Effects in 2205 Duplex Stainless Steel Using Thermoelectric Power,” NDT E
Int., Vol. 44, No. 5, 2011, pp. 463–468.
[10] Baby, S., Kowmudi, B. N., Omprakash, C. M., Satyanarayana, D. V. V., Balasubramaniam,
K., and Kumar, V., “Creep Damage Assessment in Titanium Alloy Using a Nonlinear Ultra-
sonic Technique,” Scr. Mater., Vol. 59, 2008, pp. 818–821.
[11] Ruiz, A., Ortiz, N., Medina, A., Kim, J.-Y., and Jacobs, L. J., “Application of Ultrasonic
Methods for Early Detection of Thermal Damage in 2205 Duplex Stainless Steel,” NDT E
Int., Vol. 54, 2013, pp. 19–26.
[12] Matlack, K. H., Wall, J. J., Kim, J.-Y., Qu, J., Jacobs, L. J., and Viehrig, H.-W., “Evaluation
of Radiation Damage Using Nonlinear Ultrasound,” J. Appl. Phys., Vol. 111, No. 5, 2012,
p. 054911.
[13] Matlack, K. H., Wall, J. J., Kim, J.-Y., Qu, J., and Jacobs, L. J., “Nonlinear Ultrasound to
Monitor Radiation Damage in Structural Steel,” C. Boller, Ed., 6th European Workshop
on Structural Health Monitoring, Dresden, Germany, July 3–6, 2012, DGZfP e. V. and
Fraunhofer IZFP, Berlin, Germany, Vol. 1, pp. 138–145.
[14] Matlack, K. H., Kim, J.-Y., Wall, J. J., Qu, J., Jacobs, L. J., and Sokolov, M. A., “Sensitivity of
Ultrasonic Nonlinearity to Irradiated, Annealed, and Re-irradiated Microstructure
Changes in RPV Steels,” J. Nucl. Mater., Vol. 448, 2014, pp. 26–32.
[15] Cumblidge, S. E., Motta, A. T., Catchen, G. L., Brauer, G., and Böhmert, J., “Evidence for
Neutron Irradiation-induced Metallic Precipitates in Model Alloys and Pressure-vessel
Weld Steel,” J. Nucl. Mater., Vol. 320, No. 3, 2003, pp. 245–257.
[16] Odette, G. R., Lucas, G. E., and Klingensmith, D., “On the Effect of Neutron Flux
and Composition on Hardening of Reactor Pressure Vessel Steels and Model Alloys,” R.
G. Elliman, M. A. Kirk, Jr., G. E. Lucas, and L. L. Snead, Eds., Materials Research Society
Symposium Proceedings, Pittsburgh, PA, Materials Research Society, Warrendale, PA,
Vol. 650, 2001, pp. R6.4.1–R6.4.6.
[17] Odette, G. R., Lucas, G. E., Klingensmith, D., Wirth, B. D., and Gragg, D., “The Effects of
Composition and Heat Treatment on Hardening and Embrittlement of Reactor Pressure
Vessel Steels,” NUREG/CR-6778, U.S. Nuclear Regulatory Commission, Washington,
D.C., 2003.
[18] Eason, E. D., Odette, G. R., Nanstad, R. K., and Yamamoto, T., “A Physically Based
Correlation of Irradiation-induced Transition Temperature Shifts for RPV Steels,” ORNL/
TM-2006/530, U.S. Nuclear Regulatory Commission, Washington, D.C., 2006.
[19] Odette, G. R., Yamamoto, T., and Klingensmith, D., “On the Effect of Dose Rate on Irradi-
ation Hardening of RPV Steels,” Philos. Mag., Vol. 85, Nos. 4–7, 2005, pp. 779–797.
[20] Miller, M. K., Nanstad, R. K., Sokolov, M. A., and Russell, K. F., “The Effects of Irradiation,
Annealing and Reirradiation on RPV Steels,” J. Nucl. Mater., Vol. 351, Nos. 1–3, 2006, pp.
216–222.
[21] Ulbricht, A., Bohmert, J., and Viehrig, H.-W., “Microstructural and Mechanical Characteri-
zation of Radiation Effects in Model Reactor Pressure Vessel Steels,” J. ASTM Int., Vol. 2,
No. 10, 2005, pp. 1–14.
[22] Eason, E. D., Odette, G. R., Nanstad, R. K., and Yamamoto, T., “A Physically Based Corre-
lation of Irradiation-induced Transition Temperature Shifts for RPV Steels,” J. Nucl.
Mater., Vol. 433, Nos. 1–3, 2013, pp. 240–254.
[23] Odette, G. R., “Neutron Irradiation Effects in Reactor Pressure Vessel Steels and Weldments,”
IAEA-IWG-LMNPP-98/3, International Atomic Energy Agency, Vienna, Austria, 1998.
[24] Was, G. S., Hash, M., and Odette, G. R., “Hardening and Microstructure Evolution in
Proton-irradiated Model and Commercial Pressure-vessel Steels,” Philos. Mag., Vol. 85,
Nos. 4–7, 2005, pp. 703–722.
[25] Odette, G. R. and Lucas, G. E., “Recent Progress in Understanding Reactor Pressure Vessel
Steel Embrittlement,” Radiat. Eff. Defects Solids, Vol. 144, 1998, pp. 189–231.
[26] Williams, T. J. and Ellis, D., “A Mechanistically-based Model of Irradiation Damage in

Low Alloy Steel Submerged Arc Welds,” Effects of Radiation on Materials:
20th International Symposium, ASTM STP 1405, Vol. 2, S. T. Rosinski, M. L. Grossbeck, T. R.
Allen, and A. S. Kumar, Eds., ASTM International, West Conshohocken, PA, 2001, p. 8.
[27] Miller, M. K., Pareige, P., and Burke, M. G., “Understanding Pressure Vessel Steels: An
Atom Probe Perspective,” Mater. Charact., Vol. 44, 2000, pp. 235–254.
[28] Nanstad, R. K., Niffenegger, M., Kalkhof, R. D., Miller, M. K., Sokolov, M. A., and Tipping, P.,
“Fracture Toughness, Thermo-electric Power, and Atom Probe Investigations of JRQ
Steel in I, IA, IAR, and IARA Conditions,” J. ASTM Int., Vol. 2, No. 9, 2005, pp. 1–17.
[29] Cantrell, J. H., “Fundamentals and Applications of Nonlinear Ultrasonic Nondestructive

Evaluation,” Ultrasonic Nondestructive Evaluation, T. Kundu, Ed., CRC Press LLC, Boca
Raton, FL, 2004, pp. 363–434.
[30] Cantrell, J. H. and Zhang, X.-G., “Nonlinear Acoustic Response from Precipitate-Matrix
Misfit in a Dislocation Network,” J. Appl. Phys., Vol. 84, No. 10, 1998, pp. 5469–5472.
[31] Cantrell, J. H., “Quantitative Assessment of Fatigue Damage Accumulation in Wavy Slip Met-
als from Acoustic Harmonic Generation,” Philos. Mag., Vol. 86, No. 11, 2006, pp. 1539–1554.
[32] Hull, D. and Bacon, D. J., Introduction to Dislocations, 4th ed., Butterworth-Heinemann,
Oxford, UK, 2001.
[33] Nanstad, R. K., Tipping, P., Kalkhof, R. D., and Sokolov, M. A., “Irradiation and Post-
annealing Reirradiation Effects on Fracture Toughness of RPV Steel Heat JRQ,” The
Effects of Radiation on Materials: 21st International Symposium, ASTM STP 1447, M. L.
Grossbeck, T. R. Allen, R. G. Lott, and A. S. Kumar, Eds., ASTM International, West
Conshohocken, PA, 2004, pp. 149–163.
[34] Zurbuchen, C., Viehrig, H.-W., and Weiss, F.-P., “Master Curve and Unified Curve Applic-
ability to Highly Neutron Irradiated Western Type Reactor Pressure Vessel Steels,” Nucl.
Eng. Des., Vol. 239, 2009, pp. 1246–1253.
[35] Brumovsky, M., Davies, L. M., Kryukov, A., Kyssakov, V. N., and Nanstad, R. K., “Reference
Manual on the IAEA JRQ Correlation Monitor Steel for Irradiation Damage Studies,”
IAEA-TECDOC-1230, International Atomic Energy Agency, Vienna, Austria, 2001.
[36] Liu, C. L., Odette, G. R., Wirth, B. D., and Lucas, G. E., “A Lattice Monte Carlo Structure in
Irradiated Simulation Pressure of Nanophase Compositions and Vessel Fe-Cu-Ni-Mn-Si
Steels,” Mater. Sci. Eng. A, Vol. 238, 1997, pp. 202–209.
[37] Odette, G. R. and Wirth, B. D., “A Computational Microscopy Study of Nanostructural

Evolution in Irradiated Pressure Vessel Steels,” J. Nucl. Mater., Vol. 251, 1997, pp. 157–171.
Kun Mo,1,2 Hsiao-Ming Tung,3 Xiang Chen,4 Di Yun,2

Yinbin Miao,1 Weiying Chen,1 Jonathan Almer,2
April Novak,1 and James F. Stubbins1
In-Situ Synchrotron X-Ray Study

of the Elevated Temperature
Deformation Response of SS 316L
Pressurized Creep Tubes
Reference
Mo, Kun, Tung, Hsiao-Ming, Chen, Xiang, Yun, Di, Miao, Yinbin, Chen, Weiying, Almer, Jonathan,
Novak, April, and Stubbins, James F., “In-Situ Synchrotron X-Ray Study of the Elevated
Temperature Deformation Response of SS 316L Pressurized Creep Tubes,” Small Specimen Test
ABSTRACT
A high-energy diffraction technique is presented that uses synchrotron X-rays to
characterize the in situ deformation response of pressurized creep tubes at elevated
temperature. In addition to the X-ray diffraction measurement, the technique allows
the macroscopic creep strain to be measured simultaneously during X-ray exposure.
We demonstrated this technique in two areas at different temperatures in the tube
specimen. From the X-ray diffraction patterns, we obtained a typical creep curve
with identifiable secondary and tertiary creep response at the high temperature area,
and only observed the secondary creep response at the low temperature area. The
diffraction peak broadening analysis directly showed the development of the
dislocation structures and lattice strain during deformation and make it possible to
track the development of creep void nucleation, growth and coalescence.
Manuscript received February 7, 2014; accepted for publication July 18, 2014; published online September
26, 2014.
1
Department of Nuclear, Plasma, and Radiological Engineering, Univ. of Illinois at Urbana–Champaign,
104 South Wright Street, Urbana, IL 61801.
2
Argonne National Laboratory, Argonne, IL 60439.
3
Institute of Nuclear Energy Research, Longtan, Taoyuan 32546, Taiwan.
4
Oak Ridge National Laboratory, Oak Ridge, TN 37831.
5
Houston, TX.
Copyright V
MO ET AL., DOI 10.1520/STP157620140014 245
Keywords
synchrotron radiation, pressurized creep tubes, creep
Introduction
Time-dependent deformation, or creep, is one of the most significant material
issues limiting the current design of advanced power systems, e.g., the next genera-
tion nuclear plant (NGNP), advanced fossil energy systems, or longer term fusion
power systems [1,2]. The components in these systems are expected to operate reli-
ably at elevated temperatures for long-term service life. This poses a considerable
challenge for structure materials which require higher creep resistance for extended
service lives.
The pressurized creep tube is a unique experimental approach for creep study.
It was developed primarily for irradiation testing, e.g., neutron damage, for reactor
materials, and creep testing for fusion materials [3–5]. The miniature design enables
multiple tubes to fit in a limited experimental space under extreme exposure condi-
tions, e.g., under neutron irradiation in a reactor or in a high temperature corrosive
atmosphere in an experimental loop [6]. Specimens are compact and rely only on
internal pressure to produce desired stresses. The biaxial creep deformation in the
tube can be measured periodically using high precision laser measuring systems.
However, even though creep strain, rupture lives, and partial environmental effects
can be studied from the ex-situ measurements on the tube, no microstructural in-
formation can be assessed during the creep study without employing destructive ex-
amination during or after the experiments. This limits the understanding of the
constitutive behavior of the material during creep. Mechanisms-based constitutive
characterization requires time-dependent microstructural information, e.g., disloca-
tion density or volume faction of voids and precipitates.
To overcome these shortcomings, recently developed high-energy synchrotron X-
ray techniques provide an in situ measurement capability to study internal creep mech-
anisms during deformation. To date, most of synchrotron studies on creep are focused
on the X-ray tomography to evaluate the creep voids evolutions [7–10]. Pyzalla et al.
[11] reported an experiment approach which combines X-ray tomography and diffrac-
tion to investigate in situ creep void growth and microstructure development simulta-
neously. The current work presents a multi-scale measurement methodology to study
internal creep mechanisms using synchrotron X-ray diffraction to take advantage of
the thin wall nature and special geometry of pressurized creep tubes.
Experimental
MATERIAL AND SPECIMENS
The creep tube specimens of type 316L stainless steel were manufactured by Cen-
tury Tubes Inc. The dimensions and photograph of the specimen were shown in
Fig. 1. Using the electron beam welding technique, a specimen was sealed by two
end plugs with a small hole drilled in one of the two end plugs. For internal pressur-
izing, a long tubing of the same material was welded on the cap with the hole.
HIGH-ENERGY X-RAY DIFFRACTION MEASUREMENTS

The in situ biaxial creep test with high-energy X-ray diffraction were carried out at
the 1-ID beamline at the Advanced Photon Source (APS) at Argonne National Lab-
oratory (ANL). The experimental setup is shown in Fig. 2. Similar setups have been
used to analyze various metallic materials [12–18], but this setup was unique in that
the stress was applied internally with high-pressure Ar gas instead of applying uni-
axial tensile forces. X-ray diffraction measurements were performed continuously
during the creep test by using a monochromatic 86 keV (k ¼ 0.01442 nm) X-ray
beam with a beam size of 300 300 lm2. The diffracted X-ray was collected by a
GE angio-type area detector. The distance between the sample and the detector was
1.366 m. In order to measure the local features of creep tube during experiment,
X-ray scans with a step distance of 0.5 mm were conducted along the axial direction
with penetration through the tube specimen.
CREEP TUBE PRESSURIZING

Before pressurizing, the air within the tube was pumped out by a mechanical pump
in order to minimize oxidation of the tube inner wall surface during thermal expo-
sure. Then the three-zone infrared furnace was heated to the target temperature of
700 C. Three thermal couples were spot-welded on one side of the tube specimen
FIG. 1 Schematic diagram and photograph of the pressurized creep tube.

MO ET AL., DOI 10.1520/STP157620140014 247
FIG. 2 Schematic of the synchrotron experimental setup.
to measure the actual temperature of X-ray diffraction scan region. (Fig. 3) The tem-
perature of the scan region was found to be uneven: the temperature at the center
of scan region was measured to be 100 C lower than the target temperature, while
the two sides of scan region attained the target temperature. After the temperature
was stabilized, we pressurized the specimen with high purity argon gas (99.995 %).
The internal gas pressure was monitored by a high pressure gauge during the creep
test. With the internal pressure, the creep stress can be determined by the middle
wall effective stress, reff, which is calculated based on von Mises criterion:
pffiffiffi 2
3R0
(1) reff ¼ 2 rh
R0 þ R2m

pR2 R2
(2) rh ¼ 2 i 2 1 þ 20
R0 Ri Rm
where:
Ro, Ri, and Rm ¼ the outer, inner, and mid-wall radii, respectively,
p ¼ the internal gas pressure, and
reff ¼ the mid-wall hoop stress.
HIGH-ENERGY X-RAY DIFFRACTION ON CREEP TUBE

The procedures to interpret the diameter of the pressurized tubes from diffraction
patterns are shown in Fig. 4. The basic steps for the creep strain analysis are: firstly,
measure the distance, a, between two diffraction rings for a single reflection, e.g.,
311 reflection; secondly, calculate the diameter of the tube, D, based on a tangent
angle relation with a; and finally, evaluate the creep strain by comparing the value
of diameter before and after various stages of creep deformation (Fig. 4(a)).
FIG. 3 X-ray diffraction scan region on the tube. TC1, TC2, and TC3 represent three
thermocouples welded on one side of the tube. During the creep test, the
readings for TC1, TC2, and TC3 were stabilized at 700, 600, and 700 C,
respectively.
However, due to elastic anisotropy, the distance between the two diffraction rings is
not a constant value at various azimuths. Therefore, a polar rebinning processing of
pixel intensity data into radius/angle (R/l) with respect to the pattern center was
performed. This was followed by a peak fitting in R to determine the center of each
diffraction peak in pixel (Fig. 4(b)). Then the distance of two diffraction rings, which
is also the mean distance between two diffraction peaks for azimuth from 0 to 360 ,
can be calculated by:
1X n
(3) a ¼ 0:2 mm ðP1ðgÞ P2ðgÞÞ
n n¼1
where n is the number of bins used during the polar rebinning processing (n ¼ 36
in this study), P1 and P2 are radii in pixel of fitted peak 1 and peak 2, respectively.
The distance of two diffraction rings has a factor of 0.2 mm, which is the dimension
of one pixel in the GE detector. Then we can obtain the diameter of the tube, D, by:
(4) D ¼ a=tanð2hÞ
MO ET AL., DOI 10.1520/STP157620140014 249
FIG. 4 Diameter measurements of the pressurized creep tube: (a) schematic

representations of the measurement; (b) radius peak position versus azimuth for
dual 311 diffraction rings (in circles) and the 2D fitting [21] on the radius profile
for lattice strain analysis (in lines).
Relative changes in the tube diameter during the experiment can be calculated
with respect to the initial state denoted by subscript 0, and which can be converted
to the mid-wall hoop strain by:
D D0
(5) e¼
D0
where:
D0 ¼ the diameter obtained from Eq 4 for the initial point, and
D* ¼ the diameter during creep.
The assumption for the Eq 5 is that the material is incompressible and
deformed uniformly during creep [19]. More details for converting the diameter
into the strain can be found in Ref. [20]. Since tan(2h) is almost constant during the
measurement, D0 and D* in Eq 5 can be replaced with their values from Eq 4, to
yield the much simpler equation:
a a0
(6) e¼
a0
Therefore, the macroscopic creep strain can be found directly by using the dis-
tance between two diffraction rings obtained in Eq 3.
In this study, the (311) reflection was selected to study both macroscopic creep
strain and diffraction peak broadening because this reflection was considered a suit-
able representation for characterization of macroscopic stresses and strains for face-
centered cubic (FCC) metals [22,23].
Results and Discussion

A typical diffraction pattern from the pressurized creep tube is shown in Fig. 5. The
enlarged image, sampled from a part of the diffraction pattern (in the square of
FIG. 5 Representative diffraction patterns of the pressurized creep tube.
Fig. 5), shows that each reflection forms two Debye–Scherrer diffraction rings which
are generated from two sides of the tube. The changes in diameter of the tube were
calculated by Eq 4, and are shown in Fig. 6, in which six different time steps were
selected. The impact of temperature gradient on the local creep strain development
(reflected by the changes in diameter of the tube) was observed: in the region at the
low temperature of 600 C, the changes in diameter were very small even after 4 h
of thermal exposure; in contrast, the changes in diameter were significant in the
region at the high temperature of 700 C. In other words, the tube swelled much
FIG. 6 Time-dependence of the diameter of creep tube during the creep test.
MO ET AL., DOI 10.1520/STP157620140014 251
more severely in two sides of the tube due to a higher local temperature. This inho-
mogeneous radial strain development along the tube, which is originally caused by
the temperature gradient, produced some localized strained areas in axial direction.
The most concentrated strained area in axial direction is in the range of 5 to 7 mm
from the starting point of X-ray diffraction scan (Fig. 6). The creep tube was also
found to crack in this area after the experiment.
In order to study the localized creep behavior for different temperatures in the
tube, the macroscopic creep strain, was calculated by Eq 6 for two specific areas.
The first area is located in 1 mm from the starting point of X-ray diffraction scan
(Fig. 6). The temperature of this area was measured to be 700 C. A typical creep
curve was obtained from the analysis of the time dependence of the separation of
the (311) reflections (Fig. 7). The secondary creep behavior, characterized by a con-
stant strain rate, can be seen from starting point to 200 min. Primary creep was
not observed due to rapid deformations as a result of the high temperature and
stress in the experiment. Following the secondary creep, tertiary creep was observed
with an increased strain rate, and failure occurred at 5 h. These results show that
time dependent creep strain can be measured from pressurized tubes using X-ray
diffraction patterns with much better continuity compared to interrupted strain
measurements employed in previous studies [6,19].
In addition to creep strain measurements, the X-ray data allow for an assess-
ment of the development of the dislocation population and any phase changes.
FIG. 7 Time-dependence of the macroscopic creep strain (in solid line) and the relative
width of the 311 reflection (in dotted line) during creep at 700 C. The solid and
dash lines are curves fitted though the data points to serve to guide the eye. The
time of 0 min was initialized to the completion of pressurization.
According to the Williamson–Hall’s model [24], the diffraction peak broadening is

basically determined by two factors: the grain size (corresponding to size broaden-
ing) and dislocation density (corresponding to strain broadening). At the studied
temperatures and strain rate of the present study, the creep mechanism falls in the
power-law creep field in the Ashby deformation map [25]. The dynamic recrystalli-
zation to refine the grains requires a higher temperature (>800 C). Thus, the grain
growth is not likely to occur in the conditions of this study. Therefore, the diffrac-
tion peak sharpening and broadening during creep is mostly likely attributed to the
dislocation density evolution. For the area at 700 C, as the creep progresses, the full
width at half maximum (FWHM) of the 311 reflection decreases and reaches a
static level indicating a decrease in dislocation density in the early stages of creep
(Fig. 7). This observation agrees well with the creep study by Pyzalla et al., and indi-
cates that small residual stresses may result during manufacturing the tube in the
present study [11]. After 45 min, the decrease in the FWHM almost stopped, and
then the diffraction peak width maintained a low value for 2 h. This creep period
(before 180 min in Fig. 7) with the characteristic of minimum changes in dislocation
density is considered to be the secondary creep regime. In this regime, the disloca-
tion movement, especially dislocation climb, is active, while the dislocation density
within the material does not increase. After this regime, the FWHM increases until
the tube fractures. This creep period is considered to be the tertiary creep regime,
where dislocation density significantly increases as the results of the development of
FIG. 8 Time-dependence of the macroscopic creep strain (in solid line) and the relative
width of the 311 reflection (in dotted line) during creep at 600 C. The solid and
dash lines are curves fitted though the data points to serve to guide the eye. The
time of 0 min was initialized to the completion of pressurization.
MO ET AL., DOI 10.1520/STP157620140014 253
localized stress concentrations. The localized stress concentrations are due to the
nucleation, growth and coalescence of a large volume of creep voids. From the dif-
fraction peak broadening analysis, the division of secondary and tertiary creep
regimes can be refined from the direct observations on the creep curve (noted that
the start of tertiary creep was 200 min observed from the creep curve in Fig. 7),
and the more accurate time of the initiation of the tertiary creep can be determined.
The other selected area for analyzing the creep strain and diffraction peak broaden-
ing is 7.5 mm from the starting point of X-ray diffraction scan (Fig. 6). The temperature
of this area was measured to be 600 C. Unlike the creep strain development in the
area at 700 C, we did not observe a complete creep curve in this area (Fig. 8). Although
the creep strain increased with time, the typical feature of the tertiary creep: the
increased strain rate was not observed. This indicates the tertiary creep regime was not
reached in this area during the creep test. The observation on the creep strain develop-
ment agrees well with the diffraction peak broadening analysis in which the FWHM
maintained a low value for the entire creep test (Fig. 8). However, the regions of recov-
ery and dislocation movement both of which were observed in the area at 700 C were
found in this low temperature area, indicating the secondary creep was dominant.
Conclusions
In this study, we applied the high-energy X-ray diffraction to study the creep behav-
ior of a pressurized tube. During the creep test, temperature gradient was observed
and developed a localized strain in axial direction in the tube. The creep strain devel-
opment and diffraction peak width evolution were studied in two specific areas in the
tube at the temperatures of 600 and 700 C. For the higher temperature area, we
obtained a typical creep curve with identifiable secondary and tertiary creep response,
while only a secondary creep response was observed for the lower temperature area.
The diffraction peak broadening analysis agrees well with the observations on creep
strain development, and provides additional information to track the in situ develop-
ment of the dislocation densities and creep voids during creep deformation.
ACKNOWLEDGMENTS
The work was supported by the U.S. Department of Energy under grant DE-FC07-
07ID14819 and DOE NEUP 09-516. Argonne National Laboratory’s work was sup-
ported under U.S. Department of Energy contract DE-AC02-06CH11357.
References
[1] Guérin, Y., Was, G. S., and Zinkle, S. J., “Materials Challenges for Advanced Nuclear
Energy Systems,” MRS Bull., Vol. 34, 2009, pp. 10–14.
[2] Kurtz, R., “Recent Progress on Development of Vanadium Alloys for Fusion,” J. Nucl.
Mater., Vols. 329–333, 2004, pp. 47–55.
[3] Gilbert, E. and Bates, J., “Dependence of Irradiation Creep on Temperature and Atom
Displacements in 20% Cold Worked Type 316 Stainless Steel,” J. Nucl. Mater., Vol. 65,
1977, pp. 204–209.
[4] Vitek, J., Braski, D., and Horak, J., “Effect of Preinjected Helium on the Response of V-
20Ti Pressurized Tubes to Neutron Irradiation,” J. Nucl. Mater., Vols. 141–143, 1986, pp.
982–986.
[5] Tsai, H., Matsui, H., Billone, M., Strain, R., and Smith, D., “Irradiation Creep of Vanadium-
Base Alloys,” J. Nucl. Mater., Vol. 258–263, 1998, pp. 1471–1475.
[6] Tung, H.-M. Mo, K., and Stubbins, J. F., “Biaxial Thermal Creep of Inconel 617 and Haynes
230 at 850 and 950 C,” J. Nucl. Mater., Vol. 447, Nos. 1–3, 2014, pp. 28–37.
[7] Sket, F., Isaac, A., Dzieciol, K., Sauthoff, G., Borbely, A., and Pyzalla, A. R., “In Situ Tomo-
graphic Investigation of Brass During High–temperature Creep,” Scr. Mater., Vol. 59,
No. 5, 2008, pp. 558–561.
[8] Beckmann, F., Grupp, R., Haibel, A., Huppmann, M., Noethe, M., and Pyzalla, A. R., “In-
Situ Synchrotron X–Ray Microtomography Studies of Microstructure and Damage Evolu-
tion in Engineering Materials,” Adv. Eng. Mater., Vol. 9, No. 11, 2007, pp. 939–950.
[9] Sket, F., Dzieciol, K., Isaac, A., Borbely, A., and Pyzalla, A. R., “Tomographic Method for
Evaluation of Apparent Activation Energy of Steady–state Creep,” Mater. Sci. Eng. A,
Vol. 527, Nos. 7–8, 2010, pp. 2112–2120.
[10] Cheong, K. S., Stevens, K. J., Suzuki, Y., Uesugi, K., and Takeuchi, A., “The Effects of
Microstructure on Creep Behaviour—A Study through Synchrotron X-Ray Tomography,”
Mater. Sci. Eng. A, Vols. 513–514, 2009, pp. 222–227.
[11] Pyzalla, A., Camin, B., Buslaps, T., Di Michiel, M., Kaminski, H., Kottar, A., Pernack, A., and
Reimers, W., “Simultaneous Tomography and Diffraction Analysis of Creep Damage,”
Science, Vol. 308, No. 5718, 2005, pp. 92–95.
[12] Pan, X., 2008, “Tensile Fracture Mechansims of Ferritic/Martensitic Structural Materials,”
Ph.D. thesis, University of Illinois at Urbana–Champaign, Urbana, IL.
[13] Pan, X., Wu, X. L., Chen, X., Mo, K., Almer, J., Haeffner, D. R., and Stubbins, J. F.,
“Temperature and Particle Size Effects on Flow Localization of 9–12%Cr Ferritic/Marten-
sitic Steel by In Situ X–ray Diffraction and Small Angle Scattering,” J. Nucl. Mater., Vol.
398, Nos. 1–3, 2010, pp. 220–226.
[14] Pan, X. A., Wu, X. L., Mo, K., Chen, X., Almer, J., Ilavsky, J., Haeffner, D. R., and Stubbins,
J. F., “Lattice Strain and Damage Evolution of 9–12%Cr Ferritic/Martensitic Steel during
In Situ Tensile Test by X–ray Diffraction and Small Angle Scattering,” J. Nucl. Mater., Vol.
407, No. 1, 2010, pp. 10–15.
[15] Hedstrom, P., Han, T. S., Lienert, U., Almer, J., and Oden, M., “Load Partitioning Between
Single Bulk Grains in a Two–Phase Duplex Stainless Steel During Tensile Loading,” Acta
Mater., Vol. 58, No. 2, 2010, pp. 734–744.
[16] Hedstrom, P., Lindgren, L. E., Almer, J., Lienert, U., Bernier, J., Terner, M., and Oden, M.,
“Load Partitioning and Strain-Induced Martensite Formation During Tensile Loading of a
Metastable Austenitic Stainless Steel,” Metall. Mater. Trans. A, Vol. 40, No. 5, 2009, pp.
1039–1048.
MO ET AL., DOI 10.1520/STP157620140014 255
[17] Young, M. L., Almer, J. D., Daymond, M. R., Haeffner, D. R., and Dunand, D. C., “Load Par-
titioning Between Ferrite and Cementite During Elasto-Plastic Deformation of an
Ultrahigh-Carbon Steel,” Acta Mater., Vol. 55, No. 6, 2007, pp. 1999–2011.
[18] Alvarez, M. A. V., Santisteban, J. R., Vizcaino, P., Flores, A. V., Banchik, A. D., and Almer,
J., “Hydride Reorientation in Zr2.5Nb Studied by Synchrotron X-Ray Diffraction,” Acta
Mater., Vol. 60, No. 20, 2012, pp. 6892–6906.
[19] Li, M., Nagasaka, T., Hoelzer, D., Grossbeck, M., Zinkle, S., Muroga, T., Fukumoto, K., Mat-
sui, H., and Narui, M., “Biaxial Thermal Creep of Two Heats of V4Cr4Ti at 700 and 800 C
in a Liquid Lithium Environment,” J. Nucl. Mater., Vols. 367–370, 2007, pp. 788–793.
[20] Gilbert, E. R. and Blackburn, L. D., “Creep Deformation of 20 Percent Cold-Worked Type
316 Stainless-Steel,” J. Eng. Mater. Technol., Vol. 99, No. 2, 1977, pp. 168–180.
[21] He, B. B., Two-Dimensional X–Ray Diffraction, Wiley, New York, 2009.
[22] Clausen, B., Lorentzen, T., and Leffers, T., “Self-Consistent Modeling of the Plastic Defor-
mation of FCC Polycrystals and its Implications for Diffraction Measurements of Internal
Stresses,” Acta Mater., Vol. 46, No. 9, 1998, pp. 3087–3098.
[23] Mo, K., Tung, H.-M., Li, M., Almer, J., Chen, X., Chen, W., Hansen, J. B., and Stubbins, J. F.,
“Synchrotron Radiation Study on Alloy 617 and Alloy 230 for VHTR Application,” J. Pres-
sure Vessel Technol., Vol. 135, No. 2, 2013, 021502.
[24] Williamson, G. K. and Hall, W. H., “X-Ray Line Broadening From Filed Aluminium and
Wolfram,” Acta Metall., Vol. 1, No. 1, 1953, pp. 22–31.
[25] Frost, H. J. and Ashby, M. F., Deformation-Mechanism Maps: The Plasticity and Creep of
Metals and Ceramics, Pergamon Press, Oxford, UK, 1982.
257
Author Index
A I
Abe, K., 209-224 Ito, Y., 209-224

Almer, J., 244-255
Arunkumar, S., 168-186 J
Jacobs, L. J., 227-243

B
Jitsukawa, S., 3-11
Brumovsky, M., 160-167
K
Busby, J. T., 31-49
Byun, T. S., 121-142
Kim, J.-Y., 227-243
Kinoshita, H., 3-11
C Konopı́k, P., 12-30
Kopriva, R., 160-167
Chen, W., 244-255 Kytka, M., 160-167
Chen, X., 70-87, 244-255
L
D
Lambrecht, M., 53-69
Džugan, J., 12-30 Lasan, M., 160-167
Lucon, E., 189-208
E
M
Efsing, P., 88-109
Mabuchi, Y., 53-69
Maloy, S. A., 121-142
F
Matlack, K. H., 227-243
Matocha, K., 145-159, 160-167
Field, K. G., 31-49
May, J., 88-109
McCowan, C. N., 189-208
G Miao, Y., 244-255
Miura, N., 53-69
Gray, S. E., 31-49 Mo, K., 244-255
Gussev, M. N., 31-49
N
H
Nanstad, R. K., 70-87
Hein, H., 88-109 Novak, A., 244-255
258
O Stubbins, J. F., 244-255

Suzuki, M., 3-11
Ogawa, T., 53-69 Suzuki, S., 3-11
Onizawa, K., 53-69
T
P
Tanigawa, H., 3-11
Planman, T. K., 110-120 Tung, H.-M., 244-255
Prakash, R. V., 168-186
Procházka, R., 12-30
V
Q
Valo, M., 53-69, 88-109
Valo, M. J., 110-120
Qu, J., 227-243
Viehrig, H.-W., 53-69
R
W
Rouden, J., 88-109
Wakai, E., 209-224
S Wall, J. J., 227-243
Wallin, K. R., 110-120
Saito, M., 209-224
Santoyo, R. L., 189-208 Y
Sato, S., 3-11
Siegl, J., 160-167 Yamamoto, M., 53-69
Sokolov, M. A., 31-49, 70-87 Yoon, J. H., 121-142
Soneda, N., 53-69 Yoshimoto, K., 53-69
Splett, J. D., 189-208 Yun, D., 244-255
259
Subject Index
C J
component integrity, 160-167 J-R curve testing, 70-87

copper, 168-186
crack growth, 209-224 K
creep, 244-255
cyclic small punch test, 168-186 KLST, 189-208
L
D
load-deflection curve, 145-159
DCPD, 70-87
DC(T), 70-87 M
digital image correlation, 31-49
master curve, 88-109
dynamic testing, 12-30
Master Curve method, 53-69
material testing, 88-109
F mechanical testing, 160-167
micro tensile tests, 12-30
fatigue damage, 168-186 miniature CT specimen, 110-120
FATT, 145-159 miniature samples, 31-49
fracture, 168-186 miniature specimen, 1-13
fracture toughness, 70-87, 88-109 miniaturized Charpy specimens, 189-
fracture toughness evaluation, 53-69 208
fracture toughness test, 110-120
N
H nondestructive evaluation, 227-243

nonlinear ultrasound, 227-243
high dose HT9 steel, 121-142 normalization method, 70-87
high temperature testing, 70-87 nuclear reactors, 160-167
hot cell facility, 160-167
hysteresis energy, 168-186 O
off-load line deflection, 110-120

I
P
IFMIF, 209-224
instrumented impact tests, 189-208 plant life extension, 160-167
international round robin, 189-208 precracking, 121-142
irradiation embrittlement, 88-109 pressurized creep tubes, 244-255
260
R small-punch fracture energy ESP,

145-159
radiation damage, 227-243 small-punch test, 145-159
radiation effect, 121-142 small-scale impact machines,
reactor pressure vessel, 88-109 189-208
reactor pressure vessel material, small-size specimen, 209-224
53-69 SS-1, 31-49
reactor pressure vessel steel, 227-243 SS-2, 31-49
reconstitution technique, 88-109 SS-3, 31-49
rotation point, 110-120 SS-J3, 31-49
static fracture testing, 121-142
S statistical analyses, 189-208
surface finishing, 3-11
scale factor, 31-49 surface roughness, 3-11
sheet tensile specimens, 3-11 synchrotron radiation,
small punch test, 160-167 244-255
small sample techniques, 12-30
small sized specimens, 88-109
small specimen reuse technique, T
121-142
small specimen test techniques, thermal annealing effect, 121-142
189-208 true curves, 31-49
ASTM INTERNATIONAL
Helping our world work better
ISBN: 978-0-8031-7597-6
Stock #: STP1576
www.astm.org

STP 1576-2015

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STP 1576-2015

Uploaded by

Copyright:

Available Formats

ASTM INTERNATIONAL

Selected Technical Papers

Editors: Mikhail A. Sokolov, Enrico Lucon

Small Specimen Test Techniques:

ASTM Stock #STP1576

Peer Review Policy

Printed in Bay Shore, NY

Inﬂuence of Surface Roughness on Tensile Strength of Reduced-Activation

Micro-Tensile Test Technique Development and Application to Mechanical

Role of Scale Factor During Tensile Testing of Small Specimens 31

Fracture Toughness Testing

International Round Robin Test on Master Curve Reference Temperature

J-R Curve Determination for Disk-Shaped Compact Specimens Based on the

Extended Mechanical Testing of RPV Surveillance Materials Using Reconstitution

Rotation Point and KJC Estimations for Miniature CT-Specimens Based on

Small Punch Testing

Small-Punch Testing for Tensile and Fracture Behavior: Experiences

Application of Miniature Small Punch Test Specimen in Determination of

Evaluation of Damage in Materials Due to Fatigue Cycling Through Static

Other Mechanical Tests

Effect of Hydrogen on Crack Growth Behavior in F82H Steel Using

In-Situ Synchrotron X-Ray Study of the Elevated Temperature Deformation

Author Index 257

Subject Index 259

STP 1576, 2015 / available online at www.astm.org / doi: 10.1520/STP157620140019

Shigekazu Suzuki,1 Shinnosuke Sato,1 Miho Suzuki,1

Influence of Surface Roughness

abrasive paper polishing followed by alumina suspension polishing, and (4)

FIG. 1 Shape and dimensions of small specimen (SS-J3 type).

SMALL SPECIMEN PROCESSING METHOD

TENSILE TEST CONDITIONS

SPECIMEN SAMPLING POSITIONS

FIG. 3 Specimen sampling positions and orientations.

FIG. 4 Surface roughness and tensile properties.

FIG. 5 Typical stress–strain curve.

RESULTS OF TENSILE TEST

FIG. 6 Dispersion of strain.

FIG. 7 Dispersion of tensile strength.

FIG. 8 SEM observation of the surface of the gage part.

STP 1576, 2015 / available online at www.astm.org / doi: 10.1520/STP157620140022

J. Dzugan,1 R. Procházka,2 and P. Konopı́k2

Micro-Tensile Test Technique

FIG. 1 M-TT specimen.

FIG. 2 Model of testing grips for M-TT.

FIG. 3 Testing setup with mechanical extensometer.

FIG. 5 Testing setup with print screen of videoextensometer measurement.

FIG. 6 Comparison of tensile curves obtained on standard-size specimens and M-TT

Determination of Cross-section Reduction

FIG. 9 Cross-section reduction determination for M-TT specimens.

TABLE 1 Comparison of cross-section reduction measurement with M-TT and standard-size

53.4 53.2 49.9 46.2 0.04 12.25 51.84

FIG. 10 Elongation determination for M-TT.

Tensile Tests results

Experimental LCS_Standard 4.84 209.4 1143.4 1258.0 3.6 15.6 71.1

FIG. 11 Macro of the investigated heterogeneous weld.

Strain Rate Sensitivity Measurement

[4] Procházka, R., 2013, “Determination of Mechanical Properties of Structural Materials by

[8] Zhang, D. and Chu, J., “Mechanical Characterization of Post-buckled Micro-bridge

STP 1576, 2015 / available online at www.astm.org / doi: 10.1520/STP157620140013

Maxim N. Gussev,1 Jeremy T. Busby,1 Kevin G. Field,1

Role of Scale Factor During

TABLE 1 Element composition of the investigated materials.

Material Type C Mn Si S P Cr Ni Cu Mo N Others

context of the presented work, possible negative effects caused by annealing at

J. Dzugan,1 R. Procházka,2 and P. Konopı́k2

specimen and clevis was at most 1 C. Organization B monitored and controlled