September, 2013
Vol.12, No.3
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
rib
ut
incorporating
The Journal of Pipeline Integrity
io
n
Journal of
Pipeline Engineering
Great Southern Press
Clarion Technical Publishers
Journal of Pipeline Engineering
Editorial Board - 2013
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
rib
ut
io
n
Dr Husain Al-Muslim, Pipeline Engineer, Consulting Services Department, Saudi Aramco,
Dhahran, Saudi Arabia
Mohd Nazmi Ali Napiah, Pipeline Engineer, Petronas Gas, Segamat, Malaysia
Dr-Ing Michael Beller, Rosen Engineering, Karlsruhe, Germany
Jorge Bonnetto, Operations Director TGS (retired), TGS, Buenos Aires, Argentina
Dr Andrew Cosham, Atkins, Newcastle upon Tyne, UK
Dr Sreekanta Das, Associate Professor, Department of Civil and Environmental Engineering,
University of Windsor, ON, Canada
Leigh Fletcher, Welding and Pipeline Integrity, Bright, Australia
Daniel Hamburger, Pipeline Maintenance Manager, Kinder Morgan, Birmingham, AL, USA
Dr Stijn Hertele, Universiteit Gent – Laboratory Soete, Gent, Belgium
Prof. Phil Hopkins, Executive Director, Penspen Ltd, Newcastle upon Tyne, UK
Michael Istre, Chief Engineer, Project Consulting Services,
Houston, TX, USA
Dr Shawn Kenny, Memorial University of Newfoundland – Faculty of Engineering and Applied
Science, St John’s, Canada
Dr Gerhard Knauf, Salzgitter Mannesmann Forschung GmbH, Duisburg, Germany
Prof. Andrew Palmer, Dept of Civil Engineering – National University of Singapore, Singapore
Prof. Dimitri Pavlou, Professor of Mechanical Engineering,
Technological Institute of Halkida , Halkida, Greece
Dr Julia Race, School of Marine Sciences – University of Newcastle,
Newcastle upon Tyne, UK
Dr John Smart, John Smart & Associates, Houston, TX, USA
Jan Spiekhout, DNV Kema, Groningen, Netherlands
Prof. Sviatoslav Timashev, Russian Academy of Sciences – Science
& Engineering Centre, Ekaterinburg, Russia
Patrick Vieth, President, Dynamic Risk, The Woodlands, TX, USA
Dr Joe Zhou, Technology Leader, TransCanada PipeLines Ltd, Calgary, Canada
Dr Xian-Kui Zhu, Principal Engineer, Edison Welding Institute, Columbus, OH, USA
пїЅпїЅпїЅ
3rd Quarter, 2013
141
The Journal of
Pipeline Engineering
incorporating
io
n
The Journal of Pipeline Integrity
Volume 12, No 3 • Third Quarter, 2013
rib
ut
Contents
вќ–вќ–вќ–
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
Guest Editorial..............................................................................................................................................143
A special issue of Journal of Pipeline Engineering on fracture-toughness testing, evaluation, and application
for pipeline steels
Dr Xian-Kui Zhu...........................................................................................................................................145
Fracture-toughness (K, J) testing, evaluation, and standardization
Dr William R Tyson, Dr Guowu Shen, Dr Dong-Yeob Park, and James Gianetto...........................................157
Low-constraint toughness testing
Dr Su Xu, Dr William R Tyson, and Dr C H M Simha..................................................................................165
Testing for resistance to fast ductile fracture: measurement of CTOA
Dr Robert Eiber............................................................................................................................................175
Drop-weight tear test application to natural gas pipeline fracture control
Dr Brian N Leis ............................................................................................................................................183
The Charpy impact test and its applications
Dr Robert L Amaro, Dr Jeffrey W Sowards, Elizabeth S Drexler, J David McColskey,
and Christopher McCowan............................................................................................................................199
CTOA testing of pipeline steels using MDCB specimens
Prof. Claudio Ruggieri and Leonardo L S Mathias......................................................................................... 217
Fracture-resistance testing of pipeline girth welds using bend and tensile fracture specimens
Dr He Li, Qiang Chi, Jiming Zhang, Yang Li, Lingkang Ji, and Chunyong Huo..............................................229
Fracture-toughness evaluations by different test methods for the Chinese Second West-East gas transmission
X-80 pipeline steels
Dr Philippa Moore and Dr Henryk Pisarski...................................................................................................237
CTOD and pipelines: the past, present, and future
Dr Rudi M Denys, Dr Stijn HertelГ©, and Dr Antoon A Lefevre......................................................................245
Use of curved-wide-plate (CWP) data for the prediction of girth-weld integrity
Dr Xian-Kui Zhu and Dr Brian N Leis...........................................................................................................259
Ductile-fracture arrest methods for gas-transmission pipelines using Charpy impact energy or DWTT energy
OUR COVER PHOTO shows a single-edge-notched specimen in tension
(SENT, or equivalently SE(T)) under load in CANMET’s laboratory. For further
details, see the paper by William Tyson et al. on pages 157-163. The photo is
reproduced with kind permission of CANMET; it was originally published in
one of the research organization’s recent technical reports.
The Journal of Pipeline Engineering
has been accepted by the Scopus
Content Selection & Advisory
Board (CSAB) to be part of the
SciVerse Scopus database and
index.
142
The Journal of Pipeline Engineering
T
HE Journal of Pipeline Engineering (incorporating the Journal of Pipeline Integrity) is an independent, international,
quarterly journal, devoted to the subject of promoting the science of pipeline engineering – and maintaining
and improving pipeline integrity – for oil, gas, and products pipelines. The editorial content is original papers
on all aspects of the subject. Papers sent to the Journal should not be submitted elsewhere while under
editorial consideration.
rib
ut
io
n
Authors wishing to submit papers should do so online at www.j-pipeng.com. The Journal of Pipeline Engineering
now uses the Aires Editorial Manager manuscript management system for accepting and processing manuscripts,
peer-reviewing, and informing authors of comments and manuscript acceptance. Please follow the link shown
on the Journal’s site to submit your paper into this system: the necessary instructions can be found on the
User Tutorials page where there is an Author's Quick Start Guide. Manuscript files can be uploaded in text
or PDF format, with graphics either embedded or separate.
Please contact the editor (see below) if you require any assistance.
The Journal of Pipeline Engineering aims to publish papers of quality within six months of manuscript acceptance.
4. Back issues: Single issues from current and past
volumes are available for US$87.50 per copy.
or
d
1. Disclaimer: While every effort is made to check
the accuracy of the contributions published in The
Journal of Pipeline Engineering, Great Southern Press
Ltd and Clarion Technical Publishers do not accept
responsibility for the views expressed which, although
made in good faith, are those of the authors alone.
is
t
Notes
ot
f
5. Publisher: The Journal of Pipeline Engineering
is published by Great Southern Press Ltd (UK and
Australia) and Clarion Technical Publishers (USA):
Great Southern Press, PO Box 21, Beaconsfield
HP9 1NS, UK
• tel: +44 (0)1494 675139
• fax: +44 (0)1494 670155
• email:[email protected]
• web:www.j-pipe-eng.com
• www.pipelinesinternational.com
py
-n
2. Copyright and photocopying: В© 2013 Great
Southern Press Ltd and Clarion Technical Publishers.
All rights reserved. No part of this publication may
be reproduced, stored or transmitted in any form or
by any means without the prior permission in writing
from the copyright holder. Authorization to photocopy
items for internal and personal use is granted by the
copyright holder for libraries and other users registered
with their local reproduction rights organization. This
consent does not extend to other kinds of copying such
as copying for general distribution, for advertising and
promotional purposes, for creating new collective works,
or for resale. Special requests should be addressed to
Great Southern Press Ltd, PO Box 21, Beaconsfield
HP9 1NS, UK, or to the editor.
co
Editor: John Tiratsoo
• email: [email protected]
pl
e
Clarion Technical Publishers, 3401 Louisiana,
Suite 110, Houston TX 77002, USA
• tel:
+1 713 521 5929
• fax: +1 713 521 9255
• web: www.clarion.org
Associate publisher: BJ Lowe
• email:[email protected]
Sa
m
3. Information for subscribers: The Journal of Pipeline
Engineering (incorporating the Journal of Pipeline
Integrity) is published four times each year. The
subscription price for 2013 is US$350 per year (inc.
airmail postage). Members of the Professional Institute
of Pipeline Engineers can subscribe for the special
rate of US$175/year (inc. airmail postage). Subscribers
receive free on-line access to all issues of the Journal
during the period of their subscription.
v
6. ISSN 1753 2116
v
v
www.j-pipe-eng.com
is available for subscribers
3rd Quarter, 2013
143
Guest Editorial
A special issue of Journal of Pipeline Engineering on fracture-toughness
testing, evaluation, and application for pipeline steels
T
Guest Editor delivers a brief review of historical efforts
as well as recent advances in the development of
the critical K and J testing, resistance-curve testing,
experimental estimation, and standardization by the
American Society for Testing and Materials (ASTM).
The ASTM standard specimens include the three-point
bend and compact-tension specimens.
More than ten experts in this area were invited to
write papers for this special issue, including selected
organization representatives, influential scientists,
engineers, and academics who have made significant
contributions to the development of fracture-toughness
test methods and are recognized internationally in
the area of fracture-toughness testing and evaluation,
and its application to pipeline steels. These authors
come from different countries and regions in the
world, including North America (USA and Canada),
Europe (UK and Belgium), South American (Brazil),
and Asia (China). Eleven papers were selected to
cover fracture-toughness test methods, test procedures,
experimental techniques, experimental evaluations, and
standard developments, as well as their applications
to transmission pipelines for the six fracture-toughness
parameters of CVN, DWTT, K, J, CTOD, and CTOA.
In order to better characterize dynamic ductile fracture
toughness for pipeline steels, the CTOA was proposed
as an alternative fracture parameter, and different
CTOA test methods have been developed with use
of different laboratory specimens. In the third paper,
Dr Su Xu et al. describe the CTOA test procedures
and �round-robin’ results using the DWTT specimens
and a simplified single-specimen method developed at
Canmet Materials, with a discussion on the application
to determine a critical CTOA with typical results and
step-by-step procedures. The sixth paper, by Dr Robert
Amaro et al., summarizes the CTOA testing of pipeline
steels at quasi-static and dynamic rates using modified
double-cantilever-beam (MDCB) specimens that have
been done at NIST at its Boulder, Colorado, facility
between 2006 and 2012.
rib
ut
io
n
HIS SPECIAL ISSUE of the Journal of Pipeline
Engineering is dedicated to the topic of fracturetoughness testing, evaluation, and application for pipeline
steels. It is well known that fracture toughness is the
most important material property required by fracturemechanics’ methods. For pipeline steels, the commonlyused fracture-toughness parameters are Charpy-V notch
(CVN) energy, drop-weight tear test (DWTT) energy,
stress-intensity factor K, J-integral, crack-tip-opening
displacement (CTOD), and crack-tip-opening angle
(CTOA). These parameters have been extensively used
in the oil and gas pipeline industry for engineering
design and structural-integrity assessment, including
material selection; material-performance evaluation;
defect assessment; fatigue-life estimation; crack, leak,
or rupture determination; engineering-critical analysis;
fitness-for-service analysis; fracture-initiation control;
and fracture-propagation control.
Thus, fracturetoughness testing and evaluation are critical to pipeline
steel manufacture and to structural-integrity assessment
and management.
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
Fracture toughness is known to depend on the
crack-tip constraint due to geometry and loading
configurations. To provide a more meaningful measure
of toughness, �low-constraint’ tests are developing using
single-edge notched tensile (SENT) specimens. In the
second paper, Dr William Tyson et al. describe the
development of SENT tests in terms of J and CTOD
for determining crack-growth-resistance curves that
can be used to assess the tolerance of weld flaws to
tensile loads. The seventh paper by Professor Claudio
Ruggieri presents the J-integral resistance-curve testing
for pipeline girth welds using the conventional bending
and SENT specimens.
Among the six fracture parameters, the K and J are true
fracture parameters and have special significance. The K
factor, proposed in 1957 to describe the elastic cracktip field, symbolized the birth of linear-elastic fracture
mechanics, and the J concept, proposed in 1968 to
characterize the elastic-plastic crack-tip field, symbolized
the birth of elastic-plastic fracture mechanics. Over a
subsequent half century, numerous efforts have been
made to develop valid fracture-toughness test methods
and standards. In the first paper, the undersigned
The DWTT energy is an apparent toughness parameter
that has been used in the pipeline industry since the
1960s. In the fourth paper, Robert Eiber reviews the
need, development, and application of DWTT energy
for controlling fracture propagation in natural gas
transmission pipelines. He summarizes the incidents
that started the research leading to the development
of the DWTT from 1960 to present. The initial goal
of the DWTT was to define the ductile-to-brittle
transition temperature of pipeline steels to facilitate
the specification of transition temperature below the
144
The Journal of Pipeline Engineering
io
n
rib
ut
It is expected that this special issue will serve as a
technical document for tracking the historical efforts and
developments of fracture-toughness testing, evaluation,
and application to pipeline steels, for understanding/
using appropriate fracture-toughness parameters as well
as the corresponding test methods, and also for further
improving the fracture-toughness test standards in the
future. As such, it will provide a useful technical source
for researchers and engineers in the oil and gas pipeline
industry.
ot
f
The eighth paper, by Dr He Li et al., presents fracturetoughness evaluations using the CVN and DWTT
energies for the X80 pipeline steels used for the Chinese
second west-east gas transmission pipeline. Based on a
variety of test data, these authors compare the 2-mm
striker and 8-mm striker CVN energies over a range
of temperatures, and the DWTT energy with the
8-mm striker CVN energy at the room temperature,
and thus obtain useful relationships between these
toughness parameters.
The last paper, by the Guest Editor and Dr Brian Leis,
discusses the applications of CVN impact energy and
DWTT energy on ductile-fracture-propagation control,
and reviews the existing ductile-fracture arrest methods
in terms of CVN and DWTT toughness parameters for
predicting the arrest-fracture toughness of gas transmission
pipelines, including modern high-strength pipeline steels.
is
t
In the 1950s, the CVN impact energy was the first
apparent fracture-toughness parameter to be used
to characterize toughness of linepipe steels. In the
fifth paper, Dr Brian Leis reviews the CVN test
and standard development, and assesses its use to
characterize fracture resistance in applications from
vintage to modern toughness pipeline steels. He
concludes that where tough materials are involved,
alternative testing practices are needed that are better
adapted to the specific loading and failure response
of the structure of interest.
In addition to the fracture-toughness tests, Professor
Rudi Denys developed a curved-wide-plate (CWP)
specimen for conducting a quasi-structure test to provide
a rational basis for predicting girth-weld integrity for both
stress- and strain-based designs, for establishing material
requirements, and for validating numerical models or
fracture-mechanics’-based defect assessments. These are
described in the tenth paper.
or
d
operating temperature range for linepipe. Today, the
need for a measure of the steel toughness has emerged
to control ductile-fracture-propagation arrest, leading
to examination of the DWTT energy as a substitute
for the CVN energy.
Sa
m
pl
e
co
py
-n
The CTOD is an engineering-fracture parameter that
was proposed in 1963, and this parameter has been
widely used for structural integrity analysis in the oil
and gas industry ever since. In the ninth paper, Drs
Philippa Moor and Henryk Pisarski present a review
of CTOD testing and its application to pipelines
in the past, present, and future, and describe the
development of standardized fitness-for-service assessment
procedures from the use of the CTOD design curve
to the failure-analysis diagram approach in the CTOD
British Standard.
Thanks are given to the editor who made this special
issue possible. He provided helpful advice in paper
solicitation, organization, review, and final submission.
His exceptional support made this special issue a recordbreaker in the history of this Journal.
Dr Xian-Kui Zhu
Fellow of ASME
Principal Engineer, Structural Integrity and Modelling
Edison Welding Institute
Columbus, OH 43221, USA
Email: [email protected]
3rd Quarter, 2013
145
Fracture-toughness (K, J) testing,
evaluation, and standardization
by Dr Xian-Kui Zhu
io
n
Battelle Memorial Institute, Columbus, OH, USA
T
ot
f
or
d
is
t
rib
ut
HE STRESS-INTENSITY factor K and the J-integral are the two most important parameters in
fracture mechanics, serving as the material properties to quantify the toughness or resistance of
materials against fracture in linear elastic and elastic-plastic conditions, respectively.The fracture-toughness
characterized by K and J has been widely used as the essential material properties in fracture-mechanics’
design and structural-integrity assessment for pipelines and other structures. Experimental testing and
evaluation has played a central role in providing reliable fracture-toughness for fracture-mechanics’
analysis of structures containing cracks. Since the K-factor and the J-integral concepts were proposed,
numerous investigations have been made to develop valid experimental test methods, test techniques,
evaluation procedures, and test-method standardization, as evident in ASTM E399 and ASTM E1820 –
two commonly used fracture-toughness test standards. In recent years, important improvements for the
KIc testing and significant progresses for the J-integral testing have been achieved. To better understand
these two fracture-toughness parameters and to properly use the associated test standards, the present
paper delivers a brief review of historical efforts as well as recent advances in the development of the
K-factor and the J-integral experimental estimation and standard testing.
F
Fracture-toughness measurement and evaluation is
required by the fracture-mechanics’ methods for
quantifying material toughness and for predicting crack
failure in structural design, fitness-for-service analysis,
material-performance evaluation, and structural-integrity
assessment for a variety of engineering components or
structures, such as nuclear pressure vessels and piping,
chemical plant vessels, petroleum storage tanks, and oil
and gas pipelines. Fracture-toughness testing, experimental
estimation, test procedure, and data evaluation all are
essential to experimental fracture mechanics, and have
been investigated extensively for more than a half of
century by numerous scientists and researchers around
the world.
Sa
m
pl
e
co
py
-n
RACTURE TOUGHNESS IS the capacity of a material
containing a crack to absorb applied energy or to resist
fracture, and is one of the most important properties
of any material for many design applications. Fracture
resistance is a measure of fracture toughness to describe
the increasing resistance of the material against crack
growth. Several fracture parameters have been proposed
to characterize the fracture toughness or resistance of
metallic materials. The stress-intensity factor K and the
J-integral are the two most often used fracture-toughness
parameters [1], where the K-factor was proposed in 1957
by Irwin [2] for describing the intensity of an elastic
crack-tip field, and the J-integral was proposed by Rice [3]
in 1968 for describing the intensity of an elastic-plastic
crack-tip field. Now, the K-factor is the symbol of the
linear-elastic fracture mechanics, whereas the J-integral is
the symbol of the elastic-plastic fracture mechanics. All
fracture-mechanics’ methods require fracture toughness
for crack assessments, and thus fracture-toughness testing
and evaluation are very important. Typically, KIc and JIc
represent fracture toughness at the onset of a mode-I
crack in plane-strain conditions, while K-R and J-R curves
describe the material resistance against crack growth
during loading, respectively, for elastic-dominated cracks
and for ductile tearing cracks.
Author’s contact details:
Note: Since writing this paper, the author has moved to EWI in Columbus, OH, USA, where
his contact information is:
tel: +1 614 688 5135
email: [email protected]
Over the years, a large number of experimental
investigations have been performed to measure
reliable fracture toughness, including development of
experimental methods, test techniques, test devices
and specimens, test procedures, data instrumentation,
estimation methods, experimental evaluation, and testmethod standardization. In recent years, improvements
in K testing and significant progress for the J-integral
testing have been achieved, as evident in the fracturetoughness test standards ASTM E399 [4] and ASTM
E1820 [5] that are updated periodically. ASTM E399
is a standard test method for linear-elastic plane-strain
fracture-toughness KIc testing, and ASTM E1820 is a
standard test method for the measurement of combined
fracture-toughness parameters including J and K. These
146
The Journal of Pipeline Engineering
Fracture toughness KIc (GIc) testing
-n
io
n
ot
f
In the fracture-mechanics’ community, the work of
Griffith [6] in 1920 has been regarded as the start of
the fracture-mechanics’ method, whereas actually the
work of Irwin [2] and Williams [7] in 1957 led to the
development of linear-elastic fracture mechanics’ theory
based on the crack-tip field analysis. Griffith introduced
the concept of the elastic energy-release rate G, while
Irwin proposed the concept of the stress-intensity factor
K. For an elastic crack in model-I loading, Irwin [2]
established a simple relationship between the stressintensity factor KI and the elastic-energy release rate G as:
In 1958, ASTM established a technical committee E24
on fracture testing of metals (now E08 on fatigue and
fracture mechanics) to study the fracture-toughness problem
and to develop a standard fracture-test method for metallic
materials. Srawley and Brown [11], the two pioneers who
made eminent contributions to fracture-toughness testing,
drafted the first standard method for the KIc testing.
The draft test method included the details of specimen
design, specimen size requirement, fatigue pre-cracking, test
device, fixture design, crack-opening-displacement gauge,
data instrumentation, load transducer, test procedure,
and the K calibration and calculation method. This
draft method became the first widely recognized fracturetoughness standard ASTM E399-70T for the KIc testing,
and ASTM E399 became the model for all subsequent
standard test methods of fracture-toughness measurement
in ASTM and other standard organizations in the world.
E399 has undergone many revisions and updates over
the years, and the current version is ASTM E399-09e2
[4]. In this standard, a conditional fracture toughness
KQ is defined as:
rib
ut
K (G) testing
The lower-bound value KIc (or GIc) is called �plane-strain
fracture toughness’ for the mode-I cracks, denoting the
minimum fracture resistance of the material.
is
t
To better understand the two most important fracturetoughness parameters and their test standards, the
present paper delivers an overview of historical efforts
and recent advances in the development of the K-factor
and the J-integral experimental test methods and
standards in ASTM. This includes the review of K
(G) testing, early J experimental estimation and testing,
advances of J experimental estimation and evaluation,
and development of the K-factor and the J-integral
based fracture-test standardization.
Thus, the thickness dependence of the critical Kc is
similar to that for the critical Gc.
or
d
standards were developed by ASTM (American Society
for Testing and Materials), and have been used worldwide for measuring the critical value of K and J at
the onset of fracture, and resistance curves (R-curves)
during crack growth.
is for the plane stress conditions, and
is for the plane-strain conditions.
co
where E’ = E
E’ = E (1 – ν2)
py
K I2 / E ' = G(1)
Sa
m
pl
e
In order to predict fracture failure, Irwin and Kies [8]
first introduced the concept of fracture resistance and
used it as a material property based on the energy
approach. Krafft et al. [9] obtained the earliest R-curves
using a set of centre-cracked thin plates for 7075-T6
aluminium alloy in terms of the elastic energy release
rate G and the absolute crack extension О”a. Fracture
instability was determined at a critical value of Gc,
where a drive force intercepts or tangents the resistance
curve. Irwin et al. [10] showed that the Gc depends
on the initial crack size and specimen thickness, and
observed that the Gc decreases with increasing thickness,
and reaches a lower bound value denoted by GIc for a
large thickness where the plane-strain conditions prevails.
In the 1960s, R-curves were first developed in terms
of the energy-release rate G; later, the stress-intensity
factor K was often used in lieu of G. Basically, a K-R
curve and a G-R curve are equivalent to each other for
an elastic material due to their relationship in Eqn 1.
KQ =
PQ
B W f (a / W ) (2)
where PQ is a measured critical load, and f(a/W) is a
geometry function of a/W with a the crack size and
W the specimen width. Different fracture specimens
have been developed, but this paper is central to the
commonly-used compact tension (CT) and single-edgenotched bend (SENB) specimens. In order to validate
the conditional toughness KQ as a fracture toughness KIc
result, the initial crack size of 0.45 ≤ a0/W ≤ 0.55 and
the following two validity requirements must be met:
B, b ≥ 2.5 ( K Q / σ ys )
2
(3a)
Pmax ≤ 1.1PQ (3b)
where b = W-a is the specimen ligament, Pmax is the
maximum load, and Пѓys is the yield stress of tested
material. In general, the size requirements in Eqn 3a
make it difficult to measure KIc, because the material
must be either relatively brittle or very large with a
specimen width exceeding 1 m for the 95% secant offset
procedure in ASTM E399 to accurately estimate the load
PQ at the crack initiation. The second requirement, in
Eqn 3b, is an attempt to assure that the nonlinearity
3rd Quarter, 2013
147
io
n
Applications showed that the fracture toughness KIc is
good to use only for cracks in elastic conditions, and
thus a new facture-toughness parameter was sought from
the 1960s to the 1970s to characterize fracture toughness
of a crack in elastic-plastic conditions. Such a fracture
parameter was proposed by Rice [3] in 1968 based on
the deformation theory of plasticity, and is known as
the J-integral, a path-independent mechanics’ parameter.
Originally, this parameter was used to measure the
intensity of the HRR singular crack-tip field developed
by Hutchinson [16], and Rice and Rosengren [17], for
elastic-plastic hardening materials. Finite-element analysis
showed that the J-integral can well describe the stresses,
strains, and other mechanics’ behaviours at the crack
tip for ductile metals.
The analytical and numerical results encouraged the
early experimental investigations on the J-integral
testing to develop a viable test method for evaluating
its critical value. Among the early investigators, Begley
and Landes [18, 19] successfully measured the J-integral
and its critical value at the onset of ductile-fracture
tearing using multiple laboratory specimens in mode-I
loading. Since then, the J-integral has become a
measurable material parameter and obtained extensive
applications in characterizing the fracture toughness of
ductile materials.
-n
ot
f
Similarly, Joyce [60] discussed more KIc issues. In addition,
Wallin [13] suggested determining KQ at a fixed amount
of crack growth of 0.5 mm or 2% ligament size of 1T
specimen to eliminate the size effect on the KIc value.
Further discussions of the validation requirements and
Wallin’s suggestions to improve the KIc testing are
continuing in the ASTM E08 committee.
J-integral estimation for stationary cracks
rib
ut
a. the ligament size b is the controlling dimension
in a standard specimen, and KIc is a function
of b because of the 95% secant procedure;
b. the coefficient 2.5 in Eqn 3a may be reduced
to 1.5 or smaller; and
c. the Pmax/PQ requirement in Eqn 3b only ensures
that the resistance curve between PQ and Pmax
has a �power’ of less than 0.1, and it has no
significance to define KIc.
Early J experimental estimations
is
t
The issues related to the validation requirements in
Eqns 3 have been discussed by the ASTM Technical
Committee E08 on fatigue and fracture mechanics for
many years with an aim to improve the KIc testing and
to determine a size-independent KIc value. Recently,
Wallin [12] analysed a large amount of the KIc test
data for different metals including aluminium, steel,
alloy, and titanium, and concluded that:
sheet metallic materials if the thickness requirement
for maintaining crack-tip plane-strain conditions
is disregarded. It was concluded that a J-R curve
is more general and useful than a K-R curve for
metallic materials.
or
d
observed in a load-displacement record relates to crack
initiation and not just to the plastic zone. However,
experiments have showed that the Pmax/PQ criterion in
Eqn 3b mostly invalidates good test data [1, 60].
K-R curve testing
Sa
m
pl
e
co
py
In the 1970s, after a substantial progress was made
on the KIc testing of thick-section specimens, the
attention was refocused on the critical Kc testing for
thin-section specimens, where plane stress prevails.
Utilizing the K-factor, R-curves were able to obtain for
the thin-sections using small-scale laboratory specimens,
such as the CT specimens and central cracked panels.
A fracture test method of K-R curves using small
specimens was standardized and published in 1974
by ASTM with designation E561-74, and the current
version is ASTM E561-10e1 [14]. This standard was
developed particularly for K-R curve measurements of
mode-I cracks for thin-sheet materials under plane-stress
crack-tip conditions. Materials that can be tested for
the K-R curve development are not limited by strength,
thickness, or toughness, so long as specimens are of
sufficient size to remain predominately elastic to the
effective crack extension value of interest.
Recently, Zhu and Leis [15] showed that ASTM E561
can be equivalent to ASTM E1820 (to be discussed next
in detail) to determine a size-independent R-curve for
the thin-section specimen of low-toughness aluminium
alloy, and ASTM E1820 would be applicable to thin-
In the earliest experimental evaluation, the J-integral
was interpreted as a strain-energy release rate, or work
done to the specimen per unit fracture surface area
in a material given by:
J= в€’
dU (4)
Bda
where U is the strain energy, a is the crack length,
and B is the specimen thickness. Begley and Landes
[18] tested a series of fracture specimens of the same
geometry with different crack sizes and instrumented
load-displacement data. From their test data, the energy
absorbed by each specimen was determined, and the J was
calculated from Eqn 4. This approach was rudimentary
and has shortcomings, such as multiple specimen tests
and complicated experimental analysis for determining
a critical Jc. Therefore, a simple experimental technique
was sought for estimating the J-integral only from a
single-specimen test.
148
The Journal of Pipeline Engineering
io
n
ot
f
where A is the total area under a load versus load-line
displacement (LLD) that represents the work done to
the specimen or the energy absorbed by the specimen
as a result of the presence of a crack. О· is a geometry
factor that is a function of a/W. Clarke and Landes
[23] and Sumpter [24] obtained expressions of the
О· factor using the limit-analysis method for CT and
SENB specimens.
The J-integral estimation Equns 5 or 8 are valid only
for stationary cracks in an experimental evaluation
of the J-integral to obtain its critical value at ductilefracture initiation. However, the earliest J-R curves were
constructed simply using the J-integral values that were
calculated by Eqn 5 in terms of the original crack
size and crack extension that was measured using an
unloading compliance technology proposed by Clark
et al. [26]. The resulting resistance curve tends to
overestimate J for a growing crack because the crackgrowth correction was not taken into account. To
allow crack growth, Eqn 5 or 8 has been extended in
different ways to obtain a crack-growth-corrected J as
needed in an accurate J-R curve evaluation. Two typical
improved equations for considering the crack-growth
correction are of incremental functions, where test
data are spaced at small intervals of crack extension
and the J is evaluated from the previous step. The
first J-integral incremental equation was proposed by
Garwood et al. [27] and improved by Etemad and
Turner [28], for a single-edge-notched bend specimen
with a deep crack. At the nth step of crack growth,
the total J-integral was estimated by:
rib
ut
О· A(5)
Bb
J-integral estimation for growing cracks
is
t
J=
where BN is a net thickness for the specimen with side
grooves, О· denotes a plastic geometry factor, and Apl
is the plastic area under the load-LLD curve obtained
in a fracture test. Equations 6-8 were adopted in the
first ASTM fracture-toughness test standard E813-81
[25] that was published in 1982, and now are used in
the basic procedure of the current version E1820-09e2
[4] to evaluate the plain-strain initiation toughness JIc,
when a crack-growth resistance is not desired.
or
d
Among many others, Rice et al. [20] showed that
the J-integral can be simply determined from a loaddisplacement curve obtained in a single-specimen
test using an approximate evaluation formula. They
proposed several simple J evaluation equations for
different specimens they considered. Here, only
the single-edge-notched bend (SENB) specimen and
compact-tension (CT) specimen in mode-I loading are
discussed. Through further investigations by Landes et
al. [21] and Merkle and Corten [22], a more general
equation for estimating the J-integral in a singlespecimen fracture test by use of the SENB and CT
specimens was developed as:
py
-n
For convenience, a total load-line displacement is
often separated into an elastic component and a
plastic component. On this basis, the total J-integral
can be split into elastic and plastic components that
are determined separately:
=
J J el + J pl (6)
co
пЈ«
 ηn ∆U n ,( n −1)
g (О· ) n
=
J n J n в€’1 пЈ¬1 +
(an в€’ an в€’1 ) пЈ· +
(9)
пЈ­ (W в€’ an )
пЈё B(W в€’ an )
The objective of the above separation is to improve the
m
pl
e
accuracy of the J-integral estimation, and to obtain the
consistent value of J when deformation is near linear
elastic conditions. In Eqn 6, the elastic component
Jel can be calculated directly from the K-factor for a
plane strain crack:
K 2 (1 в€’ОЅ 2 ) (7)
E
Sa
J el =
where E is Young’s modulus and ν is Poisson’s ratio.
For a stationary crack, the plastic Jpl is determined
from Eqn 5 as:
J pl =
О· Apl
BN b
The second J-integral incremental equation was
proposed by Ernst et al. [29] based on the principle of
variable separation. Since the J-integral was developed
in reference to the deformation theory of plasticity,
it was shown that the J is independent of the
loading path leading to the current values of loadline displacement and crack size in the J-controlled
crack-growth conditions. As a result, the deformation
J-integral is a unique function of two independent
variables: load-line displacement and crack length. On
these bases, Ernst et al. [29] obtained an incremental
equation to evaluate the total J-integral at the ith step
of crack growth in the form of:
пЈ®
пЈ№пЈ«
О·
пЈ¶
Оі
J i пЈЇ J i в€’1 + i в€’1 Ai в€’1,i пЈє пЈ¬1 в€’ i в€’1 (ai в€’ ai в€’1 ) пЈ· (10)
(8) =
Bb
b
пЈ°
i в€’1
пЈ»пЈ­
i в€’1
пЈё
3rd Quarter, 2013
149
(11)
io
n
rib
ut
where the CMOD-based plastic geometry factor was
obtained in Ref.31 from their FEA results.
Advances in J experimental estimations
More accurate J-integral incremental equations
In the experimental evaluation of J-R curves, the
LLD-based J-integral incremental Eqn 11 has been
used widely as an �accurate’ expression for more than
30 years because it considers crack-growth correction
and was adopted by ASTM E1820. In contrast, the
other incremental Eqn 9 did not receive much attention
until 2008 when two similar equations were proposed
by Neimitz [32] and Kroon et al. [33]. However, Zhu
and Joyce [34] showed that the two �new’ equations
are similar and equivalent to the Garwood-type Eqn 9.
In addition, Tyson and Park [35] proposed a modified
ASTM E1820 incremental J-integral equation in order
to allow larger crack-growth increments between any
two unloading-reloading cycles in a fracture test using
the elastic-compliance method. In comparison to Eqn
11, it is seen that their expression is too complicated
to be used in practice.
-n
i в€’1,i
where the incremental plastic area Apl is calculated by:
i в€’1,i
A=
pl
pl
K 2 (1 в€’ОЅ 2 ) О·CMOD ACMOD
+
(13)
E
Bb
ot
f
пЈ®
пЈ№пЈ« Оі
пЈ¶ О·
=
J pl (i ) пЈЇ J pl (i в€’1) + i в€’1 Aiplв€’1, i пЈє пЈ¬1 в€’ i в€’1 (ai в€’ ai в€’1 ) пЈ·
BN bi в€’1
пЈ°
пЈ» пЈ­ bi в€’1
пЈё
J=
is
t
Based on the incremental equation (10) for a J-integral
evaluation, the first J-R curve test standard ASTM
E1152-87 [30] was standardized and published in 1987.
This standard separated the J-integral into elastic and
plastic components and determined them separately and
incrementally at each loading step. The elastic component
of J is calculated directly from the K-factor using
Eqn 7, and the plastic component of J is determined
incrementally from Eqn 10 in the form of:
specimen. Following the basic idea of Sumpter, Kirk, and
Dodds [31] studied several possible J-integral estimation
approaches for shallow cracked SENB specimens using
detailed elastic-plastic finite-element analyses (FEA). They
found that the LLD-based J-estimation equation could
give inaccurate results for hardening materials because
the LLD-based plastic О· factor is very sensitive to the
strain-hardening exponent for SENB specimens with
shallow cracks of a/W < 0.3. In contrast, for the same
geometry, the CMOD-based plastic О· factor is nearly
insensitive to the strain-hardening exponent, when a
similar О·-factor equation was used with the plastic
area being obtained under a load-CMOD curve. Thus,
Kirk and Dodds [31] concluded that the CMOD-based
J-estimation is the most reliable, and suggested use of
the following equation in a J-integral evaluation for
SENB specimens:
or
d
where Оі is a geometry factor related to the plastic О·
factor, Ai-1,i is the incremental area under an actual
load-displacement record from step i-1 to i. Both
incremental equations in Eqns 9 and 10 consider
the crack-growth correction on the J-integral from the
last step. Moreover, Eqn 10 makes the correction
on the incremental work done to the specimen, but
Eqn 9 does not. Consequently, a larger J is likely
estimated from Eqn 9 than from Eqn 10, as shown
experimentally in Ref. 29. In general, these two
incremental formations of the J-integral equation are
applicable to any specimen, provided the two geometry
factors are known for each specimen.
1
( Pi + Pi −1 ) ( ∆ pl (i ) − ∆ pl (i −1) ) (12)
2
Sa
m
pl
e
co
py
where О”pl is the plastic component of load-line
displacement. Accurate estimation of the plastic
component Jpl(i) at each loading step using Eqn 11
requires small and uniform crack-growth increments.
Accordingly, a loading increment between the two
loading-unloading cycles must be small, and usually
30 to 60 cycles are sufficient if the elastic-compliance
method prescribed in E1820 is used. Equivalently, a
crack-growth increment is required be less than 1% of
the crack ligament size in order to use Eqn 11 for an
accurate J-integral evaluation [34]. With the calculated
values of Ji and the measured values of crack extension
(О”a = ai - a0), where a0 is an original crack length, a
J-R curve is obtained by plotting J against successive
increments of crack growth from a single-specimen test.
CMOD-based J estimation for stationary cracks
Experiments showed that an accurate measurement of
load-line displacement (LLD) is more difficult than that
of crack-mouth-opening displacement (CMOD) for the
SENB specimens in three-point bending, particularly for
a shallow crack. Sumpter [24] first used the load-CMOD
data directly in a J-integral evaluation using a bending
To obtain a more-accurate J-integral incremental equation
for a growing crack, Zhu and Joyce [34] developed
different mathematical models and physical models,
and obtained the corresponding incremental J-integral
equations. Three physical models were assumed to
approximate the integration path of a differential of
the J-integral along the actual load-displacement curve
obtained in a fracture test for a growing crack. For
convenience, these physical models were referred to as
the upper-step-line approximation (USLA), the lowerstep-line approximation (LSLA), and the mean-step-line
150
The Journal of Pipeline Engineering
For the LSLA model, the J-integral incremental equation is:
пЈ« Оі
пЈ¶ О·
=
J pl (i ) J pl (i в€’1) пЈ¬1 в€’ i (ai в€’ ai в€’1 ) пЈ· + i Aiplв€’1,i
пЈ­ bi
пЈё BN bi
(15)
For the MSLA model, the J-integral incremental equation is:
пЈ¶
пЈ« 1пЈ«Оі
Оі пЈ¶
=
J pl (i ) J pl (i в€’1) пЈ¬пЈ¬1 в€’ пЈ¬ i в€’1 + i пЈ· (ai в€’ ai в€’1 ) пЈ·пЈ·
2
b
b
i пЈё
пЈ­ i в€’1
пЈё
пЈ­
пЈ® 1 пЈ« О·i в€’1 О·i пЈ¶ i в€’1,i пЈ№ пЈ« 1 пЈ« Оі i в€’1 Оі i
+
+пЈЇ
+ пЈ· Apl пЈє пЈ¬пЈ¬1 в€’ пЈ¬
пЈ¬
пЈ° 2 BN пЈ­ bi в€’1 bi пЈё
пЈ» пЈ­ 4 пЈ­ bi в€’1 bi
пЈ¶ (16)
пЈ¶
пЈ· (ai в€’ ai в€’1 ) пЈ·пЈ·
пЈё
пЈё
Typically, to obtain an adequate normalization function,
a blunted crack size is used, and measured loads are
normalized:
PNi =
Pi
WB [1 в€’ abi / W ]
О·
(17)
where i refers to the i-th loading point, PNi is a
normalized load and abi is the blunting corrected
crack length. In the same time, the measured plastic
displacement is normalized:
-n
ot
f
Comparisons of Eqn 14 with Eqn 11, and Eqn 15 with
Eqn 9, show that the J-integral incremental equation for
the USLA model is the same as the Ernst-type equation,
and the incremental equation for the LSLA model is
identical to the Garwood-type equation. Equation 16
for the MSLA model is a new incremental equation
that is equivalent to the average of Eqns 14 and 15.
Furthermore, Zhu and Joyce [34] and Zhu [36] showed
that:
io
n
(14)
rib
ut
пЈ®
пЈ№пЈ« Оі
пЈ¶
О·
=
J pl (i ) пЈЇ J pl (i в€’1) + i в€’1 Aiplв€’1,i пЈє пЈ¬1 в€’ i в€’1 (ai в€’ ai в€’1 ) пЈ·
B
b
b
N i в€’1
i в€’1
пЈ°
пЈ»пЈ­
пЈё
is
t
For the USLA model, the J-integral incremental equation is:
The concept of the normalization method was proposed
by Herrera and Landes [37] for determining a J-R
curve directly from a load-displacement record in a
single-specimen test. This method requires an adequate
calibration function to fit the relation between the
normalized load and the normalized plastic displacement.
Different calibration functions were investigated, such
as a power-law function, and a combined function of
power law and straight line. A three-parameter LMN
function proposed by Landes et al. [38] was found to
be appropriate. Joyce [39] improved the LMN function
as a four-parameter normalization function, and the
corresponding normalization method was finally accepted
by ASTM E1820-01 and all later versions in Annex
15 Normalization data reduction technique.
or
d
approximation (MSLA). For each physical model, the
authors developed an incremental equation for estimating
the J-integral with considering the crack growth correction.
pl
e
co
py
a. the Garwood-type incremental equation (15) could
overestimate a true J-R curve;
b. the Ernst-type incremental equation (14) always
underestimates the true J-R curve; and
c. the new Eqn 16 determines a J-R curve that
well matches the true curve with much higher
accuracy than the two existing incremental
J-integral equations.
m
Normalization method
Sa
The two conventional test techniques, i.e. the elasticunloading-compliance method and the electric-potentialdrop method, are often used for growing-crack-size
measurement. They can be difficult or impractical to
implement under severe test conditions, such as high
loading rate, high temperature, or aggressive conditions.
A normalization method was then developed as an
alternative approach for directly estimating instantaneous
crack lengths from a load vs load-line-displacement
curve in conjunction with the use of initial and final
measurements of physical crack sizes. This method does
not require any test devices for online monitoring crack
growth, and thus the test costs are reduced.
∆=
pli
∆ pli ∆ i − PC
i i
=
(18)
W
W
where Ci is the specimen load-line compliance using
the blunting corrected crack length abi. Using Eqns
17 and 18, the measured load and displacement
data, up to but not including the maximum load,
are normalized. The final load-displacement pair is
normalized using the same equations except for the
final crack length without blunting correction. From
the final normalized point, a tangent line is drawn
to the normalized load-displacement curve to define a
tangent point. Using the normalized load-displacement
pair PNi, ∆ pli, a normalization function can be fitted
using the least-squares regression in the form of:
PN =
c1 + c2 ∆ pl + c3 ∆ 2pl
c4 + ∆ pl
(19)
where c1, c2, c3, and c4 are the fitting coefficients.
With this normalization function, an iterative procedure
is further used to force all (PNi, ), and ai data at
each loading point to lie on the fitted function as in
Eqn 19 by adjusting ai. In this way, crack lengths
at all data points can be determined, the J is
calculated from Eqns 7, 8, and 11, and a J-R curve
is obtained.
3rd Quarter, 2013
151
Modified-basic method
is
t
The elastic component and total value of the J-integral
are still calculated by Eqns 6 and 7, respectively. Note
that an equation similar to Eqn 21 was recently proposed
by Cravero and Ruggieri [45] in a different analysis
for a single-edge-notched tension (SENT) specimen. For
a special case with an equal LLD and CMOD, such
as for CT specimens where LLD could be estimated
directly from CMOD gauges mounted at the loadline, the two incremental Eqns 11 and 21 become
identical to each other. In general, Eqn 21 can be
used for any specimen, provided that the corresponding
geometry factors О·CMOD and ОіCMOD are known a priori for
that specimen.
ot
f
J pl (i ) (a0 )
 α − m  ∆a (20)
1+ пЈ¬
пЈ·в‹…
пЈ­ О± + m пЈё b0
1
( Pi + Pi в€’1 ) (Vpli в€’ Vpliв€’1 ) (22)
2
-n
=
J i ( ∆a ) J el (i ) (a0 ) +
i в€’1,i
A=
V pl
or
d
To unify the different fracture-test standards developed
in Europe and in the USA, Wallin and Laukkanen
[43] proposed a new evaluation procedure to correct
ductile-crack growth in a J-R curve evaluation. This
procedure is regarded as an improved basic method of
ASTM E1820, and so it is referred to as a �modifiedbasic’ method. In this method, four steps are needed
to determine a final crack-growth corrected J-R curve:
rib
ut
io
n
For SENB specimens in three-point bending, successful of a crack-growth corrected J-R curve. To this end, Zhu et
applications of the normalization method were al. [44] developed a CMOD-based J-integral incremental
demonstrated by the present author and his coauthors: equation for determining the plastic component of the
Zhu and Joyce [40] for HY80 steel, Zhu and Leis J-integral that is similar to ASTM E1820 LLD-based
[41] for X-80 pipeline steel, and Zhu et al. [42] for J-integral incremental equation:
A285 carbon steel. They compared experimental J-R
i в€’1
i в€’1
curves obtained using the normalization method with
пЈ«
пЈ¶
О·CMOD
Оі CMOD
i в€’1, i пЈ¶ пЈ«
A
=
J
J
+
1
(ai в€’ ai в€’1 ) пЈ· (21)
в€’
those obtained using the elastic-compliance method or
пЈ¬ pl (i в€’1)
пЈ·пЈ¬
V pl
pl ( i )
bi в€’1 BN
bi в€’1
пЈ­
пЈёпЈ­
пЈё
the electrical-potential method. Combined with other
comparisons, they concluded that the normalization
method is equivalent to the unloading-compliance In this equation, О·CMOD and ОіCMOD are two CMOD-based
method and the potential-drop method in a J-R curve plastic geometry factors, denotes the incremental area
under the P-Vpl curve (where Vpl is plastic CMOD),
evaluation from a single-specimen test.
and is calculated by:
py
where О± =1 for SENB specimens and О± = 0.9 for
CT specimens; m is a curve-fitting parameter from
experimental data.
m
pl
e
co
The new correction procedures have been developed for
standard CT and SENB specimens, and are generally
valid for both LLD-based and CMOD-based J-integral
calculations. The procedures are applicable to both
single-specimen tests and multiple-specimen tests, and
have the same or better accuracy as the crack-growth
correction used in the present ASTM E1820. Therefore,
this modified-basic method was adopted by ASTM
E1820-05 and its later versions in Annex A16 Evaluation
of crack growth corrected J-integral values.
Sa
CMOD-based J-integral incremental equations
Since CMOD measurements are generally more
accurate than LLD measurements, a fracture test using
SENB specimen favours CMOD gauges for measuring
displacement and specimen compliance. Using loadCMOD data, a crack-growth corrected J-R curve can be
determined using the modified-basic method outlined
above. However, the suggested correction procedure is
indirect and involves multiple steps in determining a
crack-growth corrected J-R curve. A direct CMOD method
has been desired for a long time for the determination
Due to the more-accurate CMOD data are used in Eqns
21 and 22, this new J-integral incremental equation
is able to determine a more-accurate J-R curve in a
single-specimen test. Moreover, because LLD data are
not needed in the CMOD-based J-integral estimation,
LLD gauges are not required. Thus, the fracture test
becomes more cost-effective.
Plastic geometry factor determination for SENB
specimens
In the LLD- and CMOD-based J-integral incremental
Eqns 11 and 21, two plastic geometry factors О· and Оі
are involved. An accurate J-R curve evaluation needs
accurate functions of these geometry factors, and thus
their determination become very important. A brief
review of determining these geometry factors was given
by Zhu and his co-workers [40, 44]. The slip-line solution
and the elastic-plastic finite-element calculation have
been used to determine these geometry factors for the
conventional fracture specimens. However, inaccurate
functions of О· and Оі were found in the existing solutions
for the SENB specimens in both LLD- and CMOD-based
formulations. More-accurate functions of these geometry
factors were thus determined by Zhu and Joyce [40] for
152
The Journal of Pipeline Engineering
io
n
With the development of experimental estimation
methods and experimental techniques, numerous
efforts were contributed to standardize the K-factor
and J-integral-based fracture-test methods. Zhu and Joyce
[1] presented more detailed reviews on the K and J
test method standardization. Landes [54] presented an
interesting review of historical development of J-integral
fracture mechanics’ and experimental testing at ASTM
that involved important events, places, and people. A
related review was also made in Ref.55.
pl
e
co
py
-n
ot
f
In the 1980s, the constraint-effect or size-dependence
of J-R curves was not fully understood and it was
thought that the deformation J-integral was incapable
of describing fracture resistance of ductile materials
at large crack extensions. Accordingly, Ernst [51-52]
proposed a so-called modified J-integral, JM, for use
in characterization of fracture resistance at large crack
extensions beyond the limits of the deformation J-R
curves, and this author showed that JM-R curves were
consistent at the early stages with the JM vs tearingmodulus curves correlated closely for C(T) specimens.
However, in the late 1980s and in the early 1990s,
many experimental results showed that the modified
JM-R curves were still size-dependent and may even
behave worse than the deformation J-R curves. As a
result, the deformation J, after correction for crack
growth, continues to be used in E1820 for the J-R
curve testing today.
Development of fracture-toughness
test standards
rib
ut
Extensive experiments [46-48] showed that the crack-tip
constraint has significant effects on J-R curves, with
lower curves for standard deep cracks in bending and
elevated curves for non-standard shallow cracks in
bending or any cracks in tension. The present author
[49, 50] proposed a constraint-correction method to
determine a constraint-corrected J-R curve in terms of
fracture-toughness testing, the two-parameter fractureconstraint theories, and the finite-element calculations.
This constraint-correction method is a viable technology
to solve the transferability of the J-R curves using
small laboratory specimens to those for cracks of real
structures. More detailed review on this topic can be
found in Ref.1.
From these observations, it should be recognized that
the constraint effect on resistance curves is a natural
consequence of the crack-tip constraint or triaxiality
effect on the crack-tip field. Thus, the deformation J-R
curve is adequate to describe the constraint effect, and
it should continue to be used.
is
t
Constraint effect and modified JM resistance curves
3. A JM-R curve is always higher than the deformation
J-R curve, and may become upward hooking beyond
30% crack extension for small or deeply cracked
specimens. Such behaviour may lead to nonconservative results for the integrity assessment.
or
d
SENB specimens with a wide range of crack lengths
in pure bending conditions. Zhu et al. [44] obtained
more-accurate functions of both LLD- and CMOD-based
О· and Оі factors for SENB specimens with deep and
shallow cracks in three-point bending conditions. The
newer functions of the plastic geometry factors were
adopted in the current version of ASTM E1820-11 [5].
Sa
m
Recently, the present author [53] obtained an incremental
equation for calculating the modified J-integral in
development of a resistance curve and re-evaluated the
deformation vs modified J-integral resistance curves using
historical test data and newly developed experimental data
for different structural steels. The results showed that:
1. A JM-R curve is essentially the corresponding J-R
curve without crack-growth correction, and thus
is not a result corresponding to the deformation
plasticity theory.
2. The JM-R curves are also dependent on specimen
size, geometry type or loading mode, and specimen
length. Thus, they are not �size-independent’.
It is found that the process of ASTM standardization
for the first KIc test method was excessively long and
it took about 10 years to draft and to publish the test
standard. Srawley and Brown [11] took many years to
draft/revise the first KIc test method and published it
in 1965 [11]. ASTM assigned it a temporary standard
designation of ASTM E399-70T in 1970, and officially
published it in 1972 [1]. This standard became the model
for all subsequent fracture-toughness test standards in
ASTM. In contrast, the first ASTM R-curve test standard
E561-74 took a shorter time and was published in
1974. These two standards are in the present versions
of ASTM E399-09e2 and ASTM E561-10e1. In fact,
the latest version of E399 is E399-12e1 as shown only
on the ASTM website. In this newest version, the
Pmax/PQ criterion might be removed and some other
improvements may be updated for the KIc testing.
Similar to ASTM E399, the standardization of the first
JIc test method took about 10 years to draft and to
publish [1]. The first JIc test standard was published
in 1982 with a designation of ASTM E813-81 [25], in
which only the test result of the critical J-integral was
accepted as the fracture toughness of materials. Similarly,
the first J-R curve test standard ASTM E1152-87 [30]
also took about 10 years to be developed and got
published in 1987. Then about 10 years later, ASTM
E1737-96 [57] was formed in 1996 by merging E813
for the JIc testing and E1152 for the J-R curve testing.
At the same time, a common combined fracture-test
standard ASTM E1820-96 [58] was published for
3rd Quarter, 2013
153
Conclusions
Sa
m
pl
e
co
py
-n
io
n
rib
ut
ot
f
This paper reviewed the historical efforts and recent
advances in development of the K-factor and J-integralbased fracture-toughness testing, experimental estimation,
and standardization at ASTM in the USA. While the
KIc and JIc are used to characterize fracture-initiation
toughness, a K-R curve is used to characterize the
fracture resistance for thin-sheet materials with the
elastic deformation dominated at the crack tip and
a J-R curve is used to characterize the fracture
resistance for thicker specimens of ductile materials.
The traditional J-R curve evaluation was LLD-based,
and has been used for more than 30 years. A moreaccurate J-R curve estimation method was recently
developed by use of CMOD only. In addition, this
review discussed the KIc issues including the Pmax/PQ
criterion used in ASTM E399, and the developments
of the normalization method, modified-basic method,
more-accurate J-integral incremental equations, CMODbased J-R curve evaluation, more-accurate functions
of the plastic-geometry factors, and other progress
recently made for ASTM E1820. It is anticipated that
this review will help users to better understand and
use ASTM E399 for K testing, and ASTM E1820
for the J testing.
1. X.-K.Zhu and J.A.Joyce, 2012. Review of fracture
toughness (G, K, J, CTOD, CTOA) testing and
standardization. Eng Fract. Mech., 85, pp 1-46.
2. G.R.Irwin, 1957. Analysis of stresses and strains
near the end of a crack traversing a plate. J.Applied
Mechanics, 24, pp 361-364.
3. J.R.Rice, 1968. A path independent integral and
the approximate analysis of strain concentration
by notches and cracks. Idem, 35, pp 379-386.
4. ASTM, 2012. ASTM E399-09e2: Standard test
method for linear-elastic plane-strain fracture
toughness KIc of metallic materials. American Society
for Testing and Materials, West Conshohocken, PA,
USA.
5. Ibid., 2012. ASTM E1820-11: Standard test
method for measurement of fracture toughness.
American Society for Testing and Materials, West
Conshohocken, PA, USA.
6. A.A.Griffith, 1920. The phenomena of rupture and
flow in solids. Philosophical Transactions of the Royal
Society of London, Series A, 221, pp 163-197.
7. M.L.Williams, 1957. On the stress distribution at
the base of a stationary crack. J.Applied Mechanics,
24, pp 109-114.
8. G.R.Irwin and J.A.Kies, 1954. Critical energy rate
analysis for fracture strength, Welding Journal - Research
Supplement, 19, pp 193-198.
9. J.M.Krafft, A.M.Sullivan, and R.W.Boyle, 1961.
Effect of dimensions on fast fracture instability of
notched sheets. Cranfield Symposium, I, pp 8-28.
10.G.R.Irwin, J.A.Kies, and H.L.Smith, 1958. Fracture
strengths relative to onset and arrest of crack
propagation. Proc. American Society for Testing
and Materials, 58, pp 640-660.
11. J.E.Srawley and W.F.Brown, 1965. Fracture toughness
testing methods. In: Fracture toughness testing and
its applications. ASTM STP 381, American Society
for Testing and Materials, pp 133-145.
12. K.Wallin, 2012. E399 size requirements and relevance
of the Pmax/PQ criterion, ASTM E08 Workshop
on KIc testing, May, Phoenix, Arizona, USA.
13. Ibid., 2005. Critical assessment of the standard ASTM
E399. Journal of ASTM International, 2, JAI12051.
14.ASTM, 2012. ASTM E561-10e1: Standard test
method for K-R curve determination. American
Society for Testing and Materials.
15.X.-K.Zhu and B.N.Leis, 2009. Revisit of ASTM
round robin test data for determining R curves of
thin sheet materials. Journal of ASTM International,
Paper ID JAI102510.
16.J.W.Hutchinson, 1968. Singular behavior at the
end of a tensile crack in a hardening material. J.
Mechanics of Physics and Solids, 16, pp 13-31.
17.J.R.Rice and G.F.Rosengren, 1968. Plane strain
deformation near a crack tip in a power law
hardening material. Idem, 16, pp 1-12.
is
t
The different versions of ASTM E399 and E1820
reviewed here reflect the improvements and updates of
these fracture-toughness test standards made by ASTM
over the past 50 years or so. Experimental technique
and development of fracture-toughness testing are not
reviewed in this paper, but can be found in an ASTM
manual by Joyce [59].
References
or
d
measuring the critical values of J, K, and Оґ (the cracktip-opening displacement) as well as J-R curve and
Оґ-R curve. Due to the E1820 publication, E1737 was
withdrawn in 1998. The latest version ASTM E182011 [5] incorporates many recent updates, such as the
normalization method in annex A15, the modifiedbasic method in A16, the CMOD-based О·-equation for
basic procedure in A1, the CMOD-based incremental
equation for the resistance curve procedure in A1, and
the more-accurate expressions of the plastic О· and Оі
factors in A1. Most recently, the E1820 also added
two new annexes: A17, a fracture-toughness test method
at impact loading rates using pre-cracked Charpy-type
specimens, and X2, the guidelines for measuring the
fracture toughness of materials with shallow cracks.
Acknowledgements
The author is grateful to Professor James Joyce in
the US Naval Academy for his useful discussions on
the historic efforts of fracture-toughness testing and
standardization in ASTM, and to Jesse Zhu for his
helpful manuscript editing.
154
The Journal of Pipeline Engineering
or
d
is
t
rib
ut
io
n
34.
X.-K.Zhu and J.A.Joyce, 2010. More accurate
approximation of J-integral equation for evaluating
fracture resistance curves. Journal of ASTM
International, 7, 1, Paper ID JAI102505.
35.W.R.Tyson and D.-Y.Park, 2009. Modified E1820
J-integral equation (A1.8) to allow larger increments
between unloading. ASTM E08.07.05 Task Group
Meeting, Atlanta, GA, USA, November.
36.X.-K.Zhu, 2012. Improved incremental J-integral
equation for determining crack growth resistance
curves. J. Pressure Vessel Technology, 134, Paper ID:
051404.
37.R.Herrera and J.D.Landes, 1988. Direct J-R curve
analysis of fracture toughness test. J. Testing and
Evaluation, 16, pp 427-449.
38.J.D.Landes, Z.Zhou, K.Lee, and R.Herrera, 1991.
Normalization method for developing J-R curves
with the LMN function. Idem, 19, pp 305-311.
39.J.A.Joyce, 2001. Analysis of a high rate round
robin based on proposed annexes to ASTM E
1820. Idem, 29, pp 329-351.
40.X.-K.Zhu and J.A.Joyce, 2007. J-resistance curve
testing of HY80 steel using SE(B) specimens and
normalization method. Eng Fract. Mech., 74, pp
2263-2281.
41.X.-K.Zhu and B.N.Leis, 2008. Fracture resistance
curve testing of X80 pipeline steel using SENB
specimen and normalization method. Journal of
Pipeline Engineering, 7, pp126-136.
42. X.-K.Zhu, P.S.Lam, and Y.J.Chao, 2009. Applications
of normalization method to experimental
measurement of fracture toughness for A285
carbon steel. Int. J. Pressure Vessels and Piping, 86, pp
599-603.
43.K.Wallin and A.Laukkanen, 2004. Improved crack
growth corrections for J-R curve testing. Eng Fract.
Mech., 71, pp 1601-1614.
44. X.-K.Zhu, B.N.Leis, and J.A.Joyce, 2008. Experimental
evaluation of J-R curves from load-CMOD record
for SE(B) specimens. Journal of ASTM International,
5, Paper ID: JAI101532.
45. S.Craveroand C.Ruggieri, 2007. Further developments
in J evaluation procedure for growing cracks based
on LLD and CMOD data. Int. J. of Fracture, 148,
pp 387-400.
46.J.A.Joyce E.M.Hackett and C.Roe, 1993. Effect of
crack depth and mode of loading on the J-R curve
behavior of a high-strength steel. In: Constraint
effects in fracture, ASTM STP 1171, American
Society for Testing and Materials, pp 239-263.
47.J.A.Joyce and R.E.Link, 1995. Effects of constraint
on upper shelf fracture toughness. In: Fatigue and
fracture mechanics: 26th Volume, ASTM STP 1256,
American Society for Testing and Materials, pp
142-177.
48.Ibid., 1997. Application of two parameter elasticplastic fracture mechanics to analysis of structures.
Eng Fract. Mech., 57, pp 431-446.
Sa
m
pl
e
co
py
-n
ot
f
18.J.A.Begley and J.D.Landes, 1972. The J-integral as a
fracture criterion. Fracture mechanics, ASTM STP
515, pp 1-23.
19.J.D.Landes and J.A.Begley, 1972. The effect of
specimen geometry on JIC. Idem, pp 24-39.
20. J.R.Rice, P.C.Paris, and J.G.Merkle, 1973. Some further
results of J-integral analysis and estimates. Progress in
flaw growth and fracture toughness testing, ASTM
STP 536, pp 231-245.
21.
J.D.Landes, H.Walker, and G.A.Clarke, 1979.
Evaluation of estimation procedures used in J-integral
testing. Elastic-plastic fracture, ASTM STP 668, pp
266-287.
22. J.G.Merkle and H.T.Corten, 1974. A J integral analysis
for the compact specimen, considering axial force
as well as bending effects. J. Pressure Vessel Technology,
96, pp 286-292.
23.G.A.Clarke and J.D.Landes, 1979. Evaluation of the
J-integral for the compact specimen. J. Testing and
Evaluation, 7, pp 264-269.
24.J.D.G.Sumpter, 1987. JC determination for shallow
notch welded bend specimens. Fatigue and Fracture of
Engineering Materials and Structures, 10, pp 479-493.
25.ASTM, 1982. ASTM E813-81: Standard test method
for JIc, a measure of fracture toughness. American
Society for Testing and Materials, West Conshohocken,
PA, USA.
26. G.A.Clarke, W.R.Andrews, P.C.Paris, and D.W.Schmidt,
1976. Single specimen tests for JIC determination.
Mechanics of crack growth, ASTM STP 590, pp
27-42.
27.S.J.Garwood, J.N.Robinson, and C.E.Turner, 1975.
The measurement of crack growth resistance curves
(R-curves) using the J integral. Int. J. Fracture, 11, pp
528-530.
28.M.R.Etemad, S.J.John, and C.E.Turner, 1988. Elasticplastic R-curves for large amounts of crack growth.
Fracture Mechanics: 18th Symposium, ASTM STP
945, pp 986-1004.
29. H.A.Ernst, P.C.Paris, and J.D.Landes, 1981. Estimations
on J-integral and tearing modulus T from a single
specimen test record. Fracture Mechanics: 13th
Conference, ASTM STP 743, pp 476-502.
30.ASTM, 1987. ASTM E1152-87: Standard test method
for determining J-R curves. American Society for
Testing and Materials, West Conshohocken, PA,
USA.
31.M.T.Kirk and R.H.Dodds, 1993. J and CTOD
estimation equations for shallow cracks in single
edge notch bend specimens. J. Testing and Evaluation,
21, pp 228-238.
32.A.Neimitz, 2008. The jump-like crack growth model,
the estimation of fracture energy and JR curve. Eng
Fract. Mech., 75, pp 236-252.
33.
M.Kroon, J.Faleskog, and H.Oberg, 2008. A
probabilistic model for cleavage fracture with a length
scale – parameter estimation and predictions of
growing crack experiments. Idem, 75, pp 2398-2417.
3rd Quarter, 2013
155
or
d
is
t
rib
ut
io
n
55.X.-K.Zhu, 2009. J-integral resistance curve testing
and evaluation. Journal of Zhejiang University Science A,
10, pp 1541-1560.
56. ASTM, 2013. ASTM E399-12e1: Standard test method
for linear-elastic plane-strain fracture toughness KIc
of metallic materials. American Society for Testing
and Materials, West Conshohocken, PA, USA.
57.Ibid., 1996. ASTM E1737-96: Standard test method
for J-integral characterization of fracture toughness.
Idem.
58.Ibid., 1996. ASTM E1820-96: Standard test method
for measurement of fracture toughness. Idem.
59.J.A.Joyce, 1996. Manual on elastic-plastic fracture:
laboratory test procedures. ASTM Manual Series:
MNL27.
60.Ibid., 2012. Background of KIc evaluation issues.
ASTM E08 Workshop on KIc testing, May, Phoenix,
Arizona, USA.
Sa
m
pl
e
co
py
-n
ot
f
49.Y.J.Chao and X.-K.Zhu, 2000. Constraint-modified
J-R curves and its applications to ductile crack
growth. Int. J. Fracture, 106, pp 135-160.
50.
X.-K.Zhu and B.N.Leis, 2006. Application of
constraint-corrected J-R curve to fracture analysis
of pipelines. J. Pressure Vessel Technology, 128, pp
581-589.
51.H.A.Ernst, 1983. Material resistance and instability
beyond J controlled crack growth. In: Elastic plastic
fracture: Second Symposium, Vol. I – Nonlinear
crack analysis, ASTM STP 803, American Society
for Materials and Testing, pp I191-I213.
52.H.A.Ernst and J.D.Landes, 1986. Elastic-plastic
fracture mechanics methodology using the modified
J, JM, resistance curve approach. J. Pressure Vessel
Technology, 108, pp 50-56.
53.X.-K.Zhu and P.S.Lam, 2012. Deformation versus
modified J-integral resistance curves for ductile
materials. Proc. ASME Pressure Vessels and Piping
Conference (PVP2012), July, Toronto, Canada.
54.J.D.Landes, 2000. Elastic-plastic fracture mechanics:
Where has it been? Where is it going? In: Fatigue
and fracture mechanics: 30th Volume, ASTM STP
1360, pp 3-19.
e
pl
m
Sa
py
co
io
n
rib
ut
is
t
or
d
ot
f
-n
3rd Quarter, 2013
157
Low-constraint toughness testing
1 CanmetMATERIALS, Natural Resources Canada, Ottawa, ON, Canada
2 CanmetMATERIALS, Natural Resources Canada, Hamilton, ON, Canada
3 CanmetMATERIALS, Natural Resources Canada, Calgary, AB, Canada
rib
ut
P
io
n
by Dr William R Tyson1*, Dr Guowu Shen2, Dr Dong-Yeob Park3, and
James Gianetto2
E
This factor, along with the increasing importance of
strain-based design allowing applied strains above yield,
has led to the development of low-constraint-toughness
tests that much better represent the actual loading
situation for girth-weld flaws.
Sa
m
pl
e
co
py
-n
ot
f
NSURING THE INTEGRITY of pipeline girth
welds is a vital aspect of fracture control. Defects
are difficult to avoid, especially when welding in harsh
environments with short construction seasons, and it
is important to develop repair criteria that are safe
but not overly conservative. Workmanship standards
are still important to ensure that the best procedures
are followed, but development of criteria based on
fracture mechanics is gaining increasing acceptance. This
involves comparison of measured imperfection size with
a critical size (or �acceptable’ size including suitable
safety factors) to determine whether the imperfection
could impair the integrity of a pipeline under worstcase loading situations and therefore must be rejected
or repaired. The determination is done by comparing
the worst-case crack-driving force (the crack-tip-opening
displacement CTOD, or J integral) with the fracture
toughness (CTOD, or J integral resistance) of the region
of the weld where the imperfection is located. In the
pipeline industry, toughness is commonly measured
using three-point-bend specimens which provide high
constraint (high triaxial stresses at the crack tip) and
therefore conservative toughness estimates. Constraint
has a substantial effect on toughness. Girth-weld flaws
are subjected to primarily tensile loading in service
which is a low-constraint situation, and the use of highconstraint toughnesses could unnecessarily penalize the
materials by dictating a need for unnecessary repairs.
or
d
is
t
IPELINES IN UNSTABLE terrain are sometimes subjected to large-scale bending and tensile deformation,
which places the girth welds in tension.These welds may contain small weld flaws, and it is important to
evaluate the resistance to growth (toughness) of these flaws under the relevant stresses. Current standards
require that toughness be evaluated in �high-constraint’ bending tests, and this can significantly underestimate
the crack-growth resistance in tension.To provide a more meaningful measure of toughness,�low-constraint’
tests are being developed at a number of laboratories around the world using single-edge-notched tensile
specimens (SENT, or SE(T)). The intent of this paper is to describe these developments and to indicate
the state-of-the-art in measuring crack-growth-resistance curves (R-curves) that can be used to assess the
tolerance of weld flaws to tensile loads.
* Corresponding author’s contact details:
tel: +1 613 992 9573
e-mail: [email protected]
Background
It has been known for some time that shallow cracks
and tensile loading decrease constraint from that of
deeply cracked bend specimens and thereby increase the
resistance (toughness) of steel (see, for example, [1]).
A practice for testing single-edge-notched specimens in
tension (SENT, or equivalently SE(T); the latter term
will be used here) was introduced by Det Norske Veritas
(DNV) in 2006 [2]. However, this practice required
the testing of multiple specimens to generate data of
toughness as a function of crack growth (Resistance,
or R-curve) which is costly in terms of testing time
and material. Tests requiring only a single specimen
have been developed in several laboratories around the
world, and those of CANMET [3] and ExxonMobil [4]
are nearing standardization.
This paper is meant to provide an introduction to
the state-of-the-art in low-constraint toughness testing
suitable for linepipe steel rather than an in-depth review
of the evolution of low-constraint toughness tests. It
is heavily weighted by the authors’ experience, and
apologies are extended to the many scientists around the
world who are involved in developing this technology
but whose work is not referenced for lack of space in
this brief article.
158
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
selected to simulate the constraint experienced by a
circumferential surface flaw in a girth weld1 as closely
as possible. It has been shown by FEA that the
crack-tip stress state in such a specimen is similar to
that in a surface-cracked pipe for cracks of the same
height a/W, where a is the crack size and W is the
specimen width [5]. In order to ensure transferability of
R-curves between test specimen and pipe, it is important
that W be the same for both geometries, i.e. that
W (= B for B x B specimens) be as close as possible to
the pipe wall thickness. The specimen is instrumented
with a single clip gauge to measure CMOD in the
CANMET procedure, and with two clip gauges mounted
at different heights above the specimen surface in the
ExxonMobil procedure. A specimen instrumented with
a single gauge is shown being tested in Fig.1; the
formation of a plastic hinge in the ligament in the
plane of the crack is easily visible.
Fig.1. SE(T) specimen under load.
-n
ot
f
or
d
In both procedures, unloading compliance (UC) is
used to measure crack size, although other methods are
allowed if they can be shown to give good estimates
of crack size. Because of the significant rotation of the
specimen evident in Fig.1, it is important to correct
the compliance for rotation; that is, the unloading
compliance of a deformed specimen differs from the
compliance of a straight specimen with the same crack
size because of the different geometries. The correction
can be made as a function of the ratio of the applied
load to the ligament limit load [6].
py
SE(T) test methods
Sa
m
pl
e
co
In addition to the DNV multi-specimen procedure, two
single-specimen procedures are in circulation. They are
very similar in most respects, and the major difference
between them is in the clip-gauge instrumentation and
the choice of toughness parameter. The ExxonMobil
procedure [4] uses a double-clip-gauge arrangement
directly to measure the CTOD at the original cracktip location, while the CANMET (Canada Centre for
Materials and Energy Technology) procedure uses a
single clip gauge to measure the crack-mouth-opening
displacement (CMOD) and thereby calculate the J integral
from load-CMOD data. In the CANMET procedure,
the CTOD is calculated from the J integral using the
relationship between the two parameters derived from
finite-element analysis (FEA).
The specimen design is identical in the two procedures.
This is a square-cross-section bar (B x B, where the
specimen width W = B, and B is the specimen thickness)
clamped between fixed grips with a �daylight’ H between
the grips of ten times the specimen width (H = 10B).
This specimen design and fixed-grip configuration was
In both procedures, side grooves are recommended to
generate a stress state close to that of plane strain,
which simulates the stress state along most of the
crack front of a circumferential surface flaw in a pipe.
Side-groove depths of 5% and 7.5% on each side are
recommended by ExxonMobil and CANMET respectively.
In the ExxonMobil procedure, the CTOD at the position
of the original crack tip is measured by triangulation
from the openings of the two clip gauges, and a plot
of the CTOD as a function of the crack growth О”a
then yields the CTOD R-curve. In the CANMET
procedure, the area under the load vs CMOD curve
is used in conjunction with equations developed using
FEA to calculate the J integral; a plot of the J integral
vs crack growth then gives the J-R curve. The CTOD
can be calculated from J, following general procedures
that have been adopted by ASTM for single-edge bend
SE(B) tests [7], using relations between J and CTOD
again developed for SE(T) specimens using FEA. Details
1. Extensive FEA calculations have been done to confirm similarity of
constraint. The alternative gripping arrangement – by pin loading – is not
only impractical because of the geometry (i.e. surface rather than throughthickness notches) but, owing to unrestrained rotation of the specimen, pin
loading is accompanied by a large amount of bending which increases the
constraint beyond that experienced in the field.
159
rib
ut
io
n
3rd Quarter, 2013
co
py
-n
ot
f
)
or
d
is
t
Fig.2. J-R curves for an X-100
pipe steel, with rotation
correction. PS: plain-sided;
SG: side-grooved (from [6]).
pl
e
Fig.3. Crack size for clamped
SE(T) specimens: ao, af : initial and
final crack size (SG: side-grooved,
PS: plain-sided) (from [11]).
Sa
m
of the procedures (equations for calculation of crack
size from compliance and for calculation of CTOD
and J) can be found in the published recommended
practices [3, 4].
A typical set of J-R curves is shown in Fig.2 [6]. The
sided-grooved specimens show, on average, slightly lower
R-curves than the plain-sided specimens, but the effect
is not large. The main advantage of the side grooves
is to promote straightness of the growing crack.
Determination of crack size by optical measurements on
the fracture surface to ensure correspondence between
the size estimated by unloading compliance and the
actual size follows well-established methods such as
found in ASTM standard E1820 [7]. Good agreement
can be found with careful experimentation as shown
in Fig.3 [11].
Finally, it is important to appreciate the magnitude
of the difference in R-curves resulting from differences
in constraint, and Fig.4 [11] illustrates this for a highstrength pipe steel. All specimens were pre-cracked to
a/W = 0.5, and the curves with symbols in the figure
show data for tensile SE(T) tests, and the continuous
curves show results for bend SE(B) tests. Clearly, the
SE(T) tests exhibit higher toughnesses than the SE(B)
tests. Some of the specimens were plain-sided (PS),
others were side-grooved (SG). For the bend specimens
(solid-line curves), there is a clear difference between
The Journal of Pipeline Engineering
ot
f
or
d
is
t
rib
ut
io
n
160
-n
the two conditions, the side-grooved specimens having
lower R-curves. There is also a reduction in toughness
with side grooves for the tensile specimens, although
the difference is much less pronounced.
Sa
m
pl
e
co
py
The general conclusion is that tensile loading on SE(T)
specimens is reflected in higher R-curves than threepoint bending on SE(B) specimens (i.e. imposes less
constraint than bending), at least for deep cracks. For
shallow cracks, the situation is not so clear: experience
shows that SE(T) R-curves are significantly less sensitive
to a/W than is the case for SE(B) R-curves, so that
the difference between tensile and bending loading
diminishes as a/W decreases. It may be the case that
shallow-cracked-bend specimens yield R-curves that are
only slightly more conservative than tensile specimens at
the same crack size, but this remains to be established.
In any case, it is certain that the R-curves generated by
testing shallow-cracked specimens in tension (simulating
service loading of girth-weld flaws) are substantially
higher than R-curves from deep-cracked specimens
tested in bending (as now used to characterize weld
toughness). Some data from reference [5] shown in
Table 1 emphasize this point, showing that the toughness
for a deep crack in bending (SE(B)) is only approximately
half that of a shallow crack in tension (SE(T)).
The data in the figures above have been obtained using
homogeneous material (base metal). The reason for
this is that the tests were done to illustrate the effect
of constraint on R-curves and to demonstrate that it
Specimen type
Fig.4. J-R curves measured using
SE(T) and SE(B) specimens,
a/W = 0.5 (from [11]).
a/W
PS
SG
SE(B)
0.5
650
450
SE(T)
0.25
1100
900
Table 1. J values (kJ/m2) at О”a = 0.5 mm.
is feasible to perform low-constraint SE(T) tests. Of
course, the practical application of the test methods
will be to characterize welds. In fact, SE(T) testing has
been used to derive R-curves for weld metals (WMs)
and heat-affected zones (HAZs) as well as base metals
(BMs); for example, see [10] from which Fig.5 has
been taken. Special problems associated with toughness
testing of WMs and HAZs are location of the notch
and path taken by the crack during tearing propagation.
Pros and cons
Of the three current test procedures (DNV, CANMET,
ExxonMobil), the latter two offer economies in test
time and material consumption compared to the former
because they require testing of only one B x B specimen
rather than the minimum of six 2B x B specimens
required for the DNV procedure.
Regarding crack-size measurement, a major difficulty
with using the UC technique is the prevalence of
apparent negative crack growth on initial loading. This
161
is
t
rib
ut
io
n
3rd Quarter, 2013
ot
f
or
d
Fig.5. SE(T) J-R curves for
X-100 BM,WMs, and HAZs
for single-torch welds (R1)
and dual-torch welds (R2);
H/W = 10, a/W = 0.34
(from [10]).
The advantage of this procedure is that corrections
can be made to the J integral for crack growth, using
deformation plasticity [8], enabling the actual driving
force to be found. The disadvantage is that the value
of m must be estimated, and because m is a function
of mechanical properties (primarily the work-hardening
coefficient) and crack size, approximations must
be made in estimating CTOD from J. Additionally,
the CTOD in this procedure is somewhat artificial
since it is a calculated property rather than a physical
measurement on the actual specimen. For both of these
reasons it would be better to use J as the driving force
rather than CTOD. Also, for the tough steels used in
current pipeline construction there is extensive plasticity in
the ligament, making the definition of J and its calculation
using FEA questionable. Nevertheless, small-geometrychange FEA yields consistent values of far-field J that can
be used for both small-scale measurement procedures and
large-scale (for example, cracks in pipe) geometries.
With the definition of CTOD used in the ExxonMobil
procedure, the advantage is that an actual physical
quantity is measured that can be readily interpreted,
both in small-scale and large-scale situations.
co
py
-n
phenomenon is familiar in SE(B) testing as well, and
is difficult to deal with since it appears seemingly
randomly and is not consistently connected with any
procedural detail. Nevertheless, ASTM E1820 [7] contains
a methodology to correct for it. Both CANMET and
ExxonMobil procedures use a similar technique to
derive a corrected initial crack size a0q when apparent
negative growth is encountered. Results show that this
technique works well, with a0q being very close to the
optically-measured initial crack size.
Sa
m
pl
e
Both of the single-specimen procedures deliver similar
results for CTOD at small amounts of crack growth
(defined in this paper to include blunting). In this
region, both procedures apply a practically equivalent
definition of the CTOD. However, with larger amounts
of growth – roughly, larger than 0.5 mm – it is expected
(although this has not been demonstrated) that there
will be a difference in the R-curves. This is because
the ExxonMobil procedure measures the opening at
the original crack tip while the CANMET procedure
derives the CTOD that would be appropriate for a
crack of the current size at the same load. That is,
the CANMET procedure uses the J integral as the
crack driving force variable with the CTOD being
estimated from the relation J = mПѓYОґ where ПѓY is the
flow stress (average of yield and tensile strength),
Оґ is the CTOD, and m is a parameter obtained from
FEA calculations. The CTOD in this case is defined
as the opening between the intercepts of the crack
flanks with lines drawn at В±45Вє from the crack tip.
Demonstration and application
All candidate procedures have been used in a variety
of contexts, up to and including practical material
assessment in strain-controlled situations. To obtain
general acceptance of a test procedure, however, it
is first necessary to demonstrate the practicality and
162
The Journal of Pipeline Engineering
as possible the actual field conditions. This objective
has been addressed vigorously with the development of
toughness tests that have been described briefly in this
paper. A multi-specimen test has already been published
as a Recommended Practice, and more-efficient singlespecimen tests are moving steadily toward standardization.
The intent of the present paper is to draw attention
to these developments and to highlight the relevant
literature where details may be found.
To use the R-curves measured in SE(T) tests, it is
necessary to compare the material resistance (toughness
as a function of crack growth) with the crack-driving
force in the structure (CTOD or J as function of
crack size and load or strain). The advantage of the
CANMET procedure is that the driving force can
be calculated using FEA with the same definitions of
J and CTOD as in the SE(T) tests. However, using
the ExxonMobil definition of CTOD (the opening at
the original crack tip), there is no straightforward way
to derive the CTOD numerically for comparison with
the R-curve without simulating the crack growth. That
is, to estimate the driving force in terms of CTOD
as the crack grows it is necessary to model the crack
growth in the FEA calculations, or to directly measure
the CTOD on experimental test pipes. It is to be
expected, although this has not been demonstrated,
that the CTOD at the original crack tip would increase
more rapidly with applied strain in the presence of
crack growth than if the crack were stationary. In other
words, the definition of CTOD significantly affects the
procedure that is used to predict crack growth and
instability in service.
Acknowledgements
io
n
repeatability of the method. The DNV procedure is
already published as a Recommended Practice and has
been used by the pipeline industry. A round-robin
has been completed for the CANMET procedure
applied to pipe steel base metal [9], and another
round-robin that will amalgamate both the CANMET
and ExxonMobil procedures and demonstrate practical
application to welds has been launched with the
support of members of PRCI.
-n
ot
f
or
d
is
t
rib
ut
This work has been made possible by the support of the
Federal Program on Energy Research and Development
(PERD). The research upon which the present review
is based has been supported by PERD as well as a
large number of organizations, notably the Pipeline and
Hazardous Materials Safety Administration of the US
Department of Transportation (DOT) and the Pipeline
Research Council International, Inc. (PRCI). The authors
have benefited by discussions with leading researchers
from laboratories around the world including ExxonMobil
Upstream Research Co, the Center for Reliable Energy
Systems (CRES), the National Institute of Standards and
Technology (NIST), the Lincoln Electric Co (LECO),
SINTEF, TWI, and others. The views and conclusions
in this paper are those of the authors and should not
be interpreted as representing the official policies of
any of these organizations.
pl
e
co
py
In other words, methods to use the R-curves to predict
material performance (strain capacity) are still under
development. In particular, there is not yet consensus
on key measures to be taken from the R-curves. Leading
candidates are toughness values at fixed amounts of
crack growth (such as 0.5 or 1 mm) and parameters of
power-law curve fits to R-curves, but toughness at the
maximum load in the SE(T) test could also be considered.
m
Closing remarks
Sa
To ensure integrity of high-performance pipelines,
engineers are increasingly using strain-based design (SBD)
which enables the pipe to withstand strains substantially
beyond yield strain. For efficient utilization of the high
strength available in current materials, this requires
accurate and realistic assessment of both the crack-driving
force and the material resistance. In particular, reliable
measurement of the toughness (R-curve) of pipe steel and
welds is required for accurate prediction of the effect
of conceivable flaws on the limiting strain capacity of
the pipe and circumferential girth welds. To make the
most efficient use of available materials, it is important
that the small-scale test conditions represent as closely
We would especially like to thank the staff of
CanmetMATERIALS (CMAT) for their part in developing
SE(T) test procedures, and the participants of the
CANMET round-robin for helping to demonstrate the
practicality of SE(T) testing. As noted in the introduction,
apologies are extended to the many scientists around the
world who are involved in developing SE(T) technology
and whose work is not referenced for lack of space
in this brief article.
References
1. J.A.Joyce, E.M.Hackett, and C.Roe, 1993. Effects
of crack depth and mode of loading on the
J-R curve behaviour of a high strength steel.
In: J.H.Underwood, K.-H.Schwalbe, R.H.Dodds
(Eds), Constraint effects in fracture. ASTM STP
1171, American Society for Testing and Materials,
Philadelphia, pp 239–263.
2. DNV Recommended Practice DNV-RP-F108, 2006.
Fracture control for pipeline installation methods
introducing cyclic plastic strain, Det Norske Veritas,
Norway.
3. (i) G.Shen, J.A.Gianetto, and W.R.Tyson, 2008.
Development of procedure for low-constraint
toughness testing using a single-specimen technique.
MTL Report No. 2008-18(TR); (ii) W.R.Tyson,
G.Shen, J.A.Gianetto and D.-Y.Park, 2011.
3rd Quarter, 2013
163
or
d
is
t
rib
ut
io
n
6. G.Shen and W.R.Tyson, 2009. Crack size evaluation
using unloading compliance in single-specimen singleedge-notched tension fracture toughness testing, J.
Testing and Evaluation, 37, 4, paper ID JTE102368,
ASTM International.
7. ASTM, 2011. ASTM E1820-11: Standard test method
for measurement of fracture toughness. ASTM
International, West Conshohocken, PA, USA.
T.L.Anderson, 2005. Fracture mechanics:
8. fundamentals and applications. CRC Press.
9. W.R.Tyson and J.A.Gianetto, 2013. Low-constraint
toughness testing: results of a round robin on a
draft SE(T) test procedure. Proc. ASME Pressure
Vessels & Piping Division Conference (PVP2013),
July, Paris, France, paper PVP2013-97299.
10.D.-Y.Park, W.R.Tyson, J.A.Gianetto, G.Shen, and
R.S.Eagleson, 2012. Fracture toughness of X100
pipe girth welds using SE(T) and SE(B) tests.
Proc. International Pipeline Conference (IPC2012),
September, Calgary, Canada, paper IPC2012-90289.
11.
G.Shen, J.A.Gianetto, and W.R.Tyson, 2009.
Measurement of J-R curves using single-specimen
technique on clamped SE(T) specimens. Proc.
International Offshore and Polar Engineering
Conference, Osaka, Japan, June.
Sa
m
pl
e
co
py
-n
ot
f
Development of a low-constraint SE(T) toughness
Test. 10th International Conference on Fracture
and Damage Mechanics, September, Dubrovnik,
Croatia; (iii) Y.-Y.Wang, H.Zhou, M.Liu, B.Tyson,
J.Gianetto, T.Weeks, M.Richards, J.D.McColskey,
M.Quintana, and V.B.Rajan, 2011. Weld design,
testing, and assessment procedures for high strength
pipelines. PRCI contract [277-PR-348-074512],
Summary Report 277-S-01, Section 4, Low-constraint
SE(T) test protocol.
4. (i) H.Tang, M.Macia, K.Minaar, P.Gioielli, S.Kibey,
and D.Fairchild, 2010. Development of the SENT
test for strain-based design of welded pipelines.
Proc. International Pipeline Conference (IPC2010),
September, Calgary, Canada, paper IPC2010-31590;
(ii) D.P.Fairchild, S.A.Kibey, H.Tang, V.R.Krishnan,
X.Wang, M.L.Macia, and W.Cheng, 2012. Continued
advancements regarding capacity prediction of
strain-based pipelines. Proc. International Pipeline
Conference (IPC2012), September, Calgary, Canada,
paper IPC2012-90471.
5. G.Shen, R.Bouchard, J.A.Gianetto, and W.R.Tyson,
2008. Fracture toughness evaluation of high-strength
steel pipe. Proc. ASME PVP2008 Conference,
Chicago, July, paper PVP2008-61100.
Held under the Patronage of His Excellency Shaikh Ahmed bin Mohamed Al Khalifa, Minister of Finance,
Minister in Charge of Oil and Gas Affairs, Chairman of National Oil & Gas Authority, Kingdom of Bahrain
rib
ut
20–23 October 2013, Bahrain
io
n
Keynote speaker: Mr Abdulrahman Al-Wuhaib, Senior Vice President for Downstream, Saudi Aramco
Opening address by Mr Abdulhakim Al-Gouhi, General Manager - Pipelines Department, Saudi Aramco
GULF CONVENTION CENTRE, BAHRAIN
ORGANIZERS
PLATINUM ELITE SPONSOR
is
t
or
d
W
В®
В®
Global
Webb
PLATINUM SPONSOR
-n
SUPPORTING ORGANIZATIONS
ot
f
SILVER SPONSORS
py
Join leaders in the international pipeline industry as they converge for the Best Practice in Pipeline
Operations and Integrity Management Conference and Exhibition in Bahrain.
co
CONFERENCE
Sa
m
pl
e
Technical streams presented by industry leaders covering a wide
range of subjects will run over the two and a half day event.
Some of the subjects to be discussed;
• Planning, design, construction and materials
• Operations and maintenance
• Asset integrity management
• Inspection and cathodic protection
• Repair and rehabilitation
• Automation and control
• Leak detection
Paper abstracts are now being accepted.
EXHIBITION
A comprehensive exhibition will be part of the event, allowing
companies from around the world to showcase their products and
services. Visit our website to book your space.
NETWORKING
Throughout the event there will be ample opportunities to network
with participants to further your business relationships. Meet with
industry leaders from around the world.
W
O
N
R
E
T
REGIS
www.pipelineconf.com
3rd Quarter, 2013
165
by Dr Su Xu*1, Dr William R Tyson2, and Dr C H M Simha1
rib
ut
1 CanmetMATERIALS, Natural Resources Canada, Hamilton, ON, Canada
2 CanmetMATERIALS, Natural Resources Canada, Ottawa, ON, Canada
io
n
Testing for resistance to fast
ductile fracture: measurement
of CTOA
E
ot
f
or
d
is
t
NSURING ARREST OF a fast ductile fracture (i.e. a running shear fracture) is an essential design
requirement for high-pressure natural gas and other (for example, CO2 and hydrogen) pipelines.
Fracture-arrest toughness has traditionally been measured and specified using standard Charpy absorbed
energy (CVN), such as used in the Battelle two-curve method (BTCM). But shortcomings of the Charpy test
have become evident when used to characterize modern high-strength and/or high-toughness steels. The
crack-tip-opening angle (CTOA) has been proposed as a better fracture-propagation toughness parameter.
In order to measure CTOA using a laboratory-scale specimen, a simplified single-specimen method (S-SSM)
has been developed. The S-SSM uses the familiar drop-weight tear test (DWTT) specimen, and CTOA is
calculated from instrumented outputs of force vs force-line displacement. This method is being evaluated
in an international round-robin project and has been proposed to ASTM for consideration of adoption.
This paper describes the test development and application with typical results, and step-by-step procedure.
On-going finite-element (FE) modelling work to support development of the CTOA procedure is discussed.
n: exponent in hardening
model
P: force
Pm: maximum force applied by
tup during test
rp: plastic rotation factor
S: specimen span between two
supports (S = 254 mm for
DWTT tests)
SE(B): single-edge-bend specimen
S-SSM: simplified single-specimen
method
t: thickness of specimen
T: temperature
TM: melting temperature
TR: reference temperature
Sa
m
pl
e
co
py
a: crack length
b: remaining specimen
ligament
B: specimen thickness
C1, C2: constants in flow stress
model
CTOA:crack-tip-opening angle
CTOAc:
crack-tip-opening angle in
the steady-state stage
CVN: Charpy absorbed energy
D: scalar damage variable
DWTT: drop-weight tear test
g: function modifying damage
evolution law
m: exponent in damage
evolution law
-n
Alphabetical listing of abbreviations and principal symbols used
P
REVENTION OF A FAST ductile fracture (i.e. a
long-running shear fracture) is the key element of
fracture control and design of high-pressure natural
gas and other (for example, CO2 and hydrogen)
transmission pipelines. The required toughness (i.e.
the arrest toughness) has traditionally been measured
and specified using standard Charpy V-notch impact
* Corresponding author’s contact details:
tel: +1 905 645 0815
e-mail: [email protected]
W: specimen width
О±: specimen rotation angle
Оґ: constant in function g
Оµp: plastic strain
О­: plastic strain rate
О­o: reference rate
Пѓf: average of the yield and
ultimate tensile strengths
О”: force-line displacement
(LLD)
О”m: force-line displacement
(LLD) at maximum force
О¶: negative slope of the Ln
(P/ Pm) vs (Δ – Δm)/S
curve
Оѕ: thermal softening exponent
absorbed energy in semi-empirical fracture-arrest models,
such as the Battelle two-curve method (BTCM) [1,2]. But
shortcomings of the Charpy test have become evident
when used to characterize modern high-strength and/or
high-toughness steels because much of CVN is related
to plastic (bending) deformation and crack initiation.
Also, the Charpy specimen is a small sub-pipe-thickness
specimen. The use of CVN in crack-arrest prediction
for high-toughness steels requires an empirical correction
factor [for example, Ref.3].
The Journal of Pipeline Engineering
rib
ut
io
n
166
co
py
-n
ot
f
or
d
is
t
Fig.1. Force vs displacement
curve of a DWTT.
Sa
m
pl
e
The DWTT specimen is full-thickness and the test
is performed under impact loading [4,5]. Following
experience of a number of running brittle fractures in
pipelines, the DWTT has been used to ensure ductile
rather than brittle fracture propagation. Historically, the
DWTT was developed to determine fracture appearance
and ductile-to-brittle transition temperature. Thus,
pipeline standards refer only to fracture appearance
rather than energy, and 85% of shear area indicating
ductile fracture at the operating temperature is the
usual requirement. In this case, fast fracture would be
ductile tearing. To obtain a more-reliable estimation
of ductile-fracture-arrest conditions, work (summarized
in review [6] and [7-10]) is under way to use data
from full-thickness dynamic DWTTs rather than CVN
to characterize ductile propagation including using (i)
DWTT total energy, (ii) DWTT propagation energy (i.e.
from post-maximum-force energy, or from dynamically
tested static-pre-cracked specimens), and (iii) CTOA.
Fig.2. Ln(P/Pm vs (О”-О”m)/S
for all data points beyond
the maximum force point
(Pm, О”m) and determination
of slope for CTOAc in the
central part of the specimen.
The CTOA has been proposed as the most promising
of the fracture parameters used in ductile-fracturepropagation control. In a CTOA-based fracture-arrest
methodology, the applied CTOA, usually derived
from a numerical method as a function of pipe
geometry and operating conditions, is compared with
the material CTOA (CTOAc) that must be measured
experimentally. The fracture-arrest criterion is met if
the applied CTOA is equal or less than CTOAc (i.e.
CTOAapplied ≤ CTOAc) [8,9]. A major challenge in
applying the CTOA approach to fast-ductile-fracture
control is to simplify and standardize the measurement
of CTOAc in a procedure suitable for a mill test.
Although CTOAc may be measured directly on lab.
specimens, the optical method requires high-speed
camera monitoring during DWTT, is time-consuming
to analyse the images, and does not reflect the effects
of constraint (i.e., variation of the CTOA through
the thickness); therefore the optical CTOA method is
3rd Quarter, 2013
167
Velocity
CTOA from Optical Measurement (В°)
(m/s)
Surface
Mid-thickness
CTOA from S-SSM (В°)
X100
5Г—10-5
19.0 (20.0,18.0) *
11.0 (10.0,12.0)
10.4
5.1
13.0 (12.0, 14.0)
9.5 (9.0, 10.0)
9.7
5Г—10-5
12.0
5.5 (6.0, 5.0)
4.6
5.1
10.5 (10.0, 11.0)
6.5 (6.5, 6.5)
7.0
5.1
11.8 (11.0, 12.5)
7.3 (7.0, 7.5)
rib
ut
X52
io
n
Steel
7.5
or
d
It has been found that the CTOA and crack velocity
remain reasonably constant (steady-state stage) while the
crack propagates through the central part of the ligament,
and the slope of the corresponding linear part of a
Ln(P/Pm vs (О”-О”m)/S plot yields the critical CTOA value:
Оѕ=
4r *
tanОі (4)
ot
f
not suitable for a pipe-mill test [10-14]. To overcome
these shortcomings, a simplified single-specimen method
(S-SSM) has been developed at CANMET [10-14].
In the S-SSM, CTOA is calculated using outputs
from instrumented DWTTs of force vs force-line
displacement. This indirect method is suitable for a
practical mill test, and has been proposed to ASTM
for consideration of adoption. The procedure is being
evaluated in an international round-robin. This paper
describes the test development and application with
typical results, the basic step-by-step test procedure,
and the on-going modelling effort.
is
t
Table 1. CTOA of interrupted DWT/CTOA tests.Values in brackets are measurements from opposite side surfaces or sections.
-n
or
Simplified single-specimen method
co
py
The CTOA is related to specimen geometry via a hinge
model in which rotation occurs about a centre in the
ligament. It may be shown [10] that:
e
dО±
пЈ« CTOA пЈ¶
tan пЈ¬
пЈ· = rp b
2
da (1)
пЈ­
пЈё
pl
The crack length is deduced from the ligament size b
(b = W-a) which is in turn obtained from the force
and specimen geometry as follows. Assuming that the
flow strength (Пѓf) is constant in the steady-state stage,
the limit force can be written as:
CTOA=
c
8rp 180
в‹…
Оѕ ПЂ
(В°) (5)
Figures 1 and 2 show typical force vs displacement and
Ln(P/Pm vs (О”-О”m)/S plots of an X-100 pipe steel tested
using a DWTT specimen of thickness 8 mm tested at
the usual impact rate (tup velocity 5.1 m/s). To apply
the S-SSM, a reference point (Pm and О”m) is chosen,
taken for convenience to be the maximum force.
The S-SSM has been applied to typical pipe steels [13,
14]. The main conclusions are:
4rp
пЈ« P пЈ¶
(∆ − ∆ m )
ln пЈ¬ пЈ· = в€’
(3)
tan(CTOA/2)
S
пЈ­ Pm пЈё
Table 1 shows a comparison of CTOA values from the
optical method and from the S-SSM method of typical
high-strength and low-strength pipe steels [14].
Sa
m
where A* is a constraint factor. It follows from these
equations that P, О”, and CTOA after the peak force
(Pm) are related as:
crack-front tunnelling exists in DWTT specimens.
The extent of tunnelling varies with steel and
loading rate (Fig.3 [14]);
CTOA values measured optically at the surface are
higher than those measured at mid-thickness. The
difference between surface and interior CTOAs
increases with the extent of tunnelling; and
CTOAs derived from force vs force-line
displacement (i.e. using the S-SSM) are in good
agreement with values measured at mid-thickness
(i.e. giving conservative values for fracture control).
P=
4A*Пѓ f Bb 2 (2)
S
i.
ii.
iii.
168
The Journal of Pipeline Engineering
S-SSM procedure
io
n
The development of the S-SSM using DWTT specimens
has been driven by the need to design for fast-ductilefracture arrest of axial-running cracks in steel highpressure gas pipelines [11]. The purpose has been to
develop a better test to characterize fracture-propagation
resistance than the traditional Charpy test in a form
suitable for use as a pipe-mill test. The recommended
practice measures fracture-propagation resistance in
terms of CTOA, and can be used to characterize
structural steels.
rib
ut
(a) X-100, quasi-static load
The S-SSM test procedure has been summarized in
a step-by-step format [15], and is being evaluated in
an international round-robin with seven participants.
This step-by-step procedure is described below.
is
t
Scope and summary
co
py
-n
(b) X-100, impact load
ot
f
or
d
The S-SSM procedure below describes a method to
determine fracture-propagation toughness in terms
of the critical crack-tip-opening angle (CTOAc) using
the drop-weight tear test (DWTT). The method is
intended to be used for structural steels, and draws
for apparatus, specimen geometry, and test procedures
from ASTM E436: Standard test method for drop-weight
tear tests of ferritic steels [5] and API 5L3: Recommended
practice for conducting drop-weight tear tests on line pipe [4].
Sa
m
pl
e
(c) X-52, quasi-static load
(d) X-52, impact load
Fig. 3. Crack tunnelling of DWTT specimens [14].
In addition to data (P vs О”) from the tests, calculation
of CTOA requires only an empirically determined
parameter, the plastic-rotation factor (rp). For typical
structural steels, plastic-rotation factor values have been
estimated experimentally, and will be discussed below.
The CTOA values derived according to this
recommended practice are representative of the average
through-thickness CTOA values, dominated by the
high-constraint middle-thickness region, and are usually
lower than the surface CTOA values measured optically
[13,14]. This reflects the effects of through-thickness
constraint and the resulting crack-tip tunnelling.
Crack velocities in the steady-state stage of crack
propagation through DWTT specimens for a tup
velocity of 5 m/s usually range from 12-20 m/s.
DWTT specimens of modern steels tested at room
temperature usually exhibit shear fracture under impact
loading. For pipe steels, this mimics the fracture mode
observed in full-scale pipe burst tests.
The apparatus, specimens, and test procedures are
consistent with those described in API RP 5L3
or ASTM E436-03, with the intent to employ
standardized DWTT procedures and machines (such
as hammer, anvil, and support span) to the fullest
extent possible.
169
is
t
rib
ut
io
n
3rd Quarter, 2013
or
d
Fig.5. An example of
anti-buckling guide for pipe
steel specimens of wall
thickness less than 10 mm.
Fig.4. DWTT/CTOA machine-notched specimen and dimensions.
The specimens are flattened unless a straightening
technique [17] is desired to leave the centre part of
the specimen unflattened, which may be desirable for
small-diameter pipes. The DWTT/CTOA specimen is
schematically shown in Fig.4. The recommended notches
are machined to a depth of 10 mm or pressed to a
depth of 5 mm as required by API RP 5L3 or ASTM
E436. It has been shown for typical high-strength
pipe steels that the pressed-notch and machined-notch
DWTT specimens produce similar CTOAc values [18].
Because the data required to derive fracture-propagation
toughness (CTOA) are obtained after peak force, other
notch types, sometimes used to promote easy fracture
initiation (such as Chevron notch or static pre-crack),
are acceptable but have little or no effect on CTOAc.
ot
f
Apparatus
co
py
-n
The test is normally conducted using either a vertical
drop-tower type or a pendulum-type-impact test machine
as described in API RP 5L3 or ASTM E436-03; the
key dimensions are support span S = 254 mm (10 in),
impact tup radius = 25.4 mm (1 in), and fixed support
anvils radius = 19.05 mm (0.75 in). The machine
must provide sufficient energy to completely fracture
a specimen in one impact. The absorbed energy of
typical pipe steels with thickness less 20 mm is usually
less than 10,000 J.
pl
Specimen
e
The initial velocity of the hammer at impact must be
at least 4.88 m/s1.
Sa
m
For pipe fracture-arrest applications, specimens are
removed from the pipe such that the length of the
DWTT/CTOA specimen is in the circumferential
direction. For straight seam-welded pipes, the specimens
are taken from the 90В° position with respect to the
pipe seam weld [16].
No maximum impact speed is specified in API RP 5L3 or ASTM E436.
The CTOA is not sensitive to velocity in the range of 4 to 20 m/s, which is
the range easily accessible in lab testing. With a drop tower, from Newton’s
second law we have v = в€љ(2ax), where v is velocity, a is acceleration due
to gravity (9.8 m/s2), and x is drop height; v = 4.88 m/s corresponds to a
drop height of 1.22 m (= 4 ft).
1
The specimens from steel pipe are of full thickness to
ensure that, because CTOA is dependent on thickness,
the CTOA measured in the test reflects the value that
would be obtained for fast ductile fracture in the fullthickness pipe.
Tests are considered to be invalid if the specimen buckles
during impact. For specimens of thickness less than 10
mm, anti-buckling guides may be needed to prevent
buckling and sliding. Simple guides acting at the ends
of the DWTT specimens mounted on the anvils have
proven to be effective. Figure 5 shows an example of
an anti-bucking fixture used at CanmetMATERIALS.
For very thin specimens, the buckling forces may be
less than the crack-initiation forces and it may be
necessary to reduce the width to avoid buckling. For
170
The Journal of Pipeline Engineering
rib
ut
io
n
The plastic-rotation factor (rp) stems from the �plastic
hinge’ model [20,21] that assumes the two arms of
a single-edge-notched bend (SE(B)) specimen rotate
symmetrically about an axis of rotation centred in the
uncracked ligament. The ratio of the distance between
the crack tip and the hinge point to the length of the
remaining ligament is defined as the plastic rotation
factor; its value is usually determined experimentally
or numerically. Values of rp for structural steels in the
recommended practice are:
пЈ±0.57 (CVN > 100 J)
rp = пЈІ
0.54 (CVN ≤ 100 J)
(6)
These values are empirical, and reflect the change in
geometry of the ligament from a rectangular to a neartrapezoidal shape. The change in geometry (thickening at
the loading point and thinning at the notch) depends
on the deflection of specimens before crack initiation
and therefore on initiation toughness. For pipe steels,
it has been shown that fracture initiation scales with
notch toughness and the step change at 100 J (Eqn 6)
has been chosen as suitable based on evidence in the
literature [3].
ot
f
The CTOA test requires measurement of initial hammer
velocity and of force vs time during the impact test.
The force vs deflection relation can be deduced from
the force vs time relation and the initial velocity of the
hammer. During an impact test, the energy absorbed
is supplied by the kinetic energy of the hammer, and
the instantaneous velocity and displacement can be
calculated based on the energy-conservation principle.
High-speed data-acquisition systems and instrumented
hammer tups are required. Systems commercially
available for instrumented Charpy tests are acceptable
for the DWTTs. Force-time data acquisition at a rate
of 5 x 106/s has proven sufficient. High-speed video
equipment (10,000 frames/s) can be used to monitor
surface-crack propagation during impact tests but is not
necessary for the determination of CTOAc according
to this method. The data-acquisition system and the
high-speed camera can be triggered via a photo-diode
that can also be used to record initial impact velocity.
is
t
Procedure
(i.e. Ln(P/Pm)) corresponding to this range of a/W was
calculated from the limit force equation (i.e. P proportional
to b2) [10,12]. The best-fitting slope corresponds to
the average CTOA in the steady-state region and this
technique has been shown to smooth CTOA scatter in
the steady-state region very effectively.
or
d
a 3-mm thickness, the DWTT specimen width had
to be reduced from 76 mm to 31 mm to prevent
buckling [13]; critical CTOA values were determined
in the same a/W range as the standard 76-mm-wide
DWTT specimens as described below.
Calculation
py
-n
Data from an instrumented DWTT can be imported
to and processed in commonly available office software
(such as MS Excel). The only information required for
CTOAc calculation are force vs deflection; an example
of such data obtained from a test on a high-strength
pipe steel is shown in Fig.1.
co
The steps in the recommended practice for calculating
CTOA are:
Sa
m
pl
e
1. Find the maximum force value (Pm) and the
corresponding force-line displacement value
(О”m) (Fig.1).
2. Calculate Ln(P/Pm) and plot vs (Δ – Δm )/S for
all data points beyond the maximum force
point (Pm, О”m) (i.e. the data relevant to fracture
propagation, see Fig.2).
3. Find the slope of the Ln(P/Pm) vs (Δ – Δm )/S
curve corresponding to Ln(P/Pm) values between
-0.51 and -1.21 (i.e. between the dashed lines
in Fig.2).
4. Calculate CTOAc according to Eqn 5.
The specified force range (i.e. corresponding to
Ln(P/Pm) values between -0.51 and -1.21) spans the
steady-state region ([12]; see Fig.2). Over this range,
the crack traverses the central part of the specimen,
i.e. 0.32 ≤ a/W ≤ 0.53 [19]; the force ratio
Validation
The CANMET recommended practice is intended for
ductile-fracture propagation and is not valid for cleavage
fracture. Cleavage fracture manifests itself as a sudden
drop of force in the force vs displacement curve; CTOA
values in cleavage fracture are very small.
Report
The values of CTOAc are reported to one decimal place.
The report must contain a summary including, as a
minimum: material and specimen ID, wall thickness,
specimen orientation, test temperature, initial tup velocity,
and fracture appearance (i.e. percent ductile fracture).
A graph of the force vs displacement curves (see Fig.1)
must be included in the report.
Finite-element (FE) modelling
FE modelling provides insight into the specimen
response (stress distribution, through-thickness effects,
temperature rise, etc.) and enables extrapolation into
regions not accessible to the DWTT test (notably,
171
rib
ut
io
n
3rd Quarter, 2013
higher strain rates and larger specimens). It is integral
to the development of the procedure.
or
d
is
t
Fig.6. Force-displacement curve
and contour plots of damage
for X-70 DWTT specimen.
The flow stress of the matrix material, accounting
for strain rate and temperature effects, is assumed
to be given by:
Sa
m
pl
e
co
py
-n
ot
f
Recent progress using the modified Xue-Wierzbicki
damage model [22] to simulate DWTTs is described
Оѕ
below. The material is assumed to be an elastic-plastic
пЈ«
пЈ® ОµпЂ¦ p пЈ№ пЈ¶ пЈ« пЈ® T в€’ TR пЈ№ пЈ¶
n
пЈ¬
Пѓ Пѓ o (1 + C1Оµ p ) пЈ¬пЈ¬1 + C2 пЈЇ пЈє пЈ·пЈ· 1 в€’ пЈЇ
(9)
пЈє пЈ· material and the strength is governed by work hardening, =
пЈ° ОµпЂ¦o пЈ» пЈё пЈ¬пЈ­ пЈ° TM в€’ TR пЈ» пЈ·пЈё
пЈ­
strain rate, and temperature. Damage in the material is
modelled using a damage-mechanics’ approach. Damage
is assumed to be a scalar isotropic damage variable that where the first terms on the right model the
is used to degrade the strength and elastic modulus quasi-static hardening curve, which is described by
of the material. Accordingly, damage, is assumed to the yield strength Пѓo, hardening constant C1, and
evolve by a phenomenological evolution law given in the hardening exponent n. The second set of terms
incremental form as:
in parentheses model a logarithmic dependence of
strength on strain rate, which is described by a constant
C2 and the ratio of plastic strain rate to a reference
(m в€’1)
1 пЈ«Оµp пЈ¶
1
rate ОµпЂ¦ p / ОµпЂ¦o . During plastic straining, 90% of the plastic
dD = m
dОµ p (7)
пЈ¬
пЈ·
g(ОµпЂ¦p ) пЈ­ Оµ f пЈё
Оµf
work is dissipated as heat, which leads to softening;
the terms in the final parentheses in Eqn 9 relate
where is our modification for strain-rate effect, Оµp is the to model softening owing to the corresponding
equivalent plastic strain, Оµf is the failure strain which temperature increase, which is assumed to depend on
depends on the Lode angle and mean stress, and m temperature, T, reference temperature TR, and melting
is an exponent. Function g is taken to be
temperature TM. Linear thermal softening is assumed
in our calculations, and the softening exponent is
set to unity.
ОµпЂ¦p
g(ОµпЂ¦p )= 1 + Оґ log( )
ОµпЂ¦o (8) The model was implemented via a user-subroutine
in Abaqus/Explicit finite-element software and
computations to simulate the DWTT were performed.
where ОµпЂ¦ p / ОµпЂ¦o is the ratio of the plastic strain rate to Element removal, when D = 1, in the element was
a reference rate, and Оґ is a constant. The reference used to model crack propagation. Strength properties
rate is usually chosen as the rate of the quasi-static of a typical X-70 pipe steel were used for the
tensile test. Notice that for positive Оґ, the introduction matrix. By trial and error, the model parameters were
of function g is equivalent to assuming a decrease in calibrated and some preliminary results are displayed
in Fig.6.
damage rate with increase in strain rate.
172
The Journal of Pipeline Engineering
Summary
References
is
t
rib
ut
io
n
1. W.A. Maxey, 1974. Fracture initiation, propagation,
and arrest. Proc. 5th Symposium on Line Pipe
Research, American Gas Association, Paper J,
Arlington, AGA Catalog Number L301175.
2. R.J.Eiber, T.A.Bubenik, and W.A.Maxey, 1993.
Fracture control technology for natural gas pipelines.
NG-18 Report #208, PRC of AGA, Catalogue No.
L51691.
3. B.N.Leis, R.J.Eiber, L.Carlson, and A.Gilroy-Scott,
1998. Relationship between apparent (total) Charpy
vee-notch toughness and the corresponding dynamic
crack-propagation resistance. Proc. 3th International
Pipeline Conference (IPC 1998), ASME, pp 723731.
4. API Recommended Practice 5L3, 1996. Recommended
practice for conducting drop-weight tear tests on
line pipe, 3rd Edn. American Petroleum Institute,
Washington, DC, USA
5. ASTM, 2008. ASTM E436-03: Standard test method
for drop-weight tear tests of ferritic steels. ASTM
International, West Conshohocken, PA, USA.
6. A.B.Rothwell, 2000. Fracture propagation control
for gas pipelines – past, present and future. Proc.
3rd International Pipeline Conference, pp 387-405.
7. H.Makino, T.Kubo, T.Shiwaku, S.Endo, T.Inoue,
Y.Kawaguchi, Y.Matsumoto, and S.Machida, 2001.
Prediction for crack propagation and arrest of shear
fracture in ultra-high pressure natural gas pipelines.
ISIJ International, 41, pp 381-388.
8. G.Demofonti, G.Buzzichelli, S.Venzi, and
M.Kanninen, 1995. Step by step procedure for
the two specimen CTOA test. Proc. 2nd Pipeline
Technology Conference: Pipeline Technology, Vol.
II, Ed. R.Denys, pp 503-512.
9. M.Di Biagio, A.Fonzo, G.Mannucci, A.Meleddu,
M.Murri, M.Tavassi,and L.N.Pussegoda, 2005.
Ductile fracture propagation resistance for advanced
pipeline designs. GRI Report No. GRI-04/0129,
Gas Research Institute, March.
10. S.Xu, R.Bouchard, and W.R.Tyson, 2007. Simplified
single-specimen method for evaluating CTOA. Eng.
Fract. Mech., 74, pp 2459-2464.
11.S.Xu and W.R.Tyson, 2008. CTOA measurement
of pipe steels using DWTT specimen. Proc.7th
International Pipeline Conference (IPC 2008),
ASME, IPC2008-64060.
S.Xu, W.R.Tyson, and R.Bouchard, 2009.
12.
Experimental validation of simplified single-specimen
CTOA method for DWTT specimens. Proc. 12th
International Conference on Fracture (ICF12), Paper
T35.018.
co
py
-n
ot
f
A test procedure for measurement of CTOA for
resistance to fast ductile fracture has been developed and
demonstrated. It is being evaluated in an international
round-robin project organized by CanmetMATERIALS
and results to date (i.e. the results from three of the
seven participants) showed good repeatability (within
В±10% of the mean). The recommended CTOA test
procedure is being studied by the relevant ASTM
committee. On-going FE modelling has shown good
agreement with experiment and excellent promise to
provide further insight into the test and to extend
investigations into regions of specimen size and loading
conditions not accessible by laboratory tests.
editor, Dr Xian-Kui Zhu, for helpful suggestions to
improve the manuscript. We would also like to thank
the CTOA round-robin participants and ASTM E08.07
for their contributions.
or
d
In Fig.6, open symbols indicate the force and displacement
corresponding to contour maps of damage. Post-peak,
at a force of approximately 150 kN, the propagating
fracture changes from initially flat (tunnelling) to slant,
and subsequently, the slant fracture becomes fully
developed. The CTOAc value and fracture front shape
are in agreement with experimental observations of the
DWTT specimen. Following the article by Xue and
Wierzbicki [22], the adoption of Lode-angle dependence
in the failure strain is the key to modelling slant
fracture. To estimate CTOA in the computation, the
mesh is sectioned at the mid-plane: the plane is such
that the long dimension of the sample lies in it. Two
vectors are drawn along the flanks of the crack and the
angle between them estimated and taken as the CTOA.
Preliminary estimates of the CTOA yield 15o and this
compares favourably with the value of 12.5o obtained by
carrying out the analysis described in the previous sections
on the force-displacement curve, for this X-70 steel, by
Xu and Tyson [11]. Details of our continuing modelling
effort will be reported in a forthcoming publication.
Sa
m
pl
e
The long-term objective of the work is to validate a
CTOA-based method to predict the toughness required
for crack arrest which has become problematic for highstrength steels, and to incorporate this approach in
pipeline standards. This will be done by: standardization
of crack-tip-opening angle (CTOA) measurement using
the drop-weight tear test (DWTT) specimen; and review
and development of crack-arrest toughness specification,
by analysis of transferability of CTOA from small-scale
specimen to full-scale pipe.
Acknowledgement
This work forms part of a CanmetMATERIALS
project on fracture arrest toughness measurement and
specification supported by the Federal Program on Energy
Research and Development (PERD). The authors would
like to thank their colleagues R.Bouchard, R.Eagleson,
D.Y.Park, J.Liang, J.Sollen, and R.Guilbeault for their
contribution to experimental work, and to the guest
3rd Quarter, 2013
173
is
t
rib
ut
io
n
18.S.Xu, J.Sollen, J.Liang, R.Zavadil, and W.R.Tyson,
2012. Effects of notch type and loading rate on
CTOA of an X65 pipe steel for CO2 pipeline.
CanmetMATERIALS Report, CMAT-2012-01(TR).
19.S.Xu, R.Bouchard, and W.R.Tyson, 2004. Flow
behaviour and ductile fracture toughness of a high
toughness steel. Proc. 5th International Pipeline
Conference (IPC 2004), IPC04-0192.
20. R.W.Nichols, F.M.Burdekin, A.Cowan, D.Elliot, and
T.Ingham, 1969. Practical fracture mechanics for
structural steel. Ed.: M.O.Dobson, United Kingdom
Atomic Energy Authority.
21.British Standard Draft for Development D19, 1972.
Methods for crack opening displacement (COD)
testing.
22. L.Xue and T.Wierzbicki, 2009. Numerical simulation
of fracture mode transition in ductile plates. Int.
J. Solids and Structures, 46, pp 1423–1435.
Sa
m
pl
e
co
py
-n
ot
f
or
d
13.
S.Xu, W.R.Tyson, R.Eagleson, C.N.McCowan,
E.D.Drexler, J.D.McColskey, and Ph.P.Darcis, 2010.
Measurement of CTOA of pipe steels using MDCB
and DWTT specimens. Proc.8th International
Pipeline Conference (IPC 2010), ASME, IPC201031076.
14.S.Xu, R.Eagleson, W.R.Tyson, and D.-Y.Park, 2011.
Crack tunnelling and crack tip opening angle in
drop-weight tear test specimens. Int. J. Fracture, 172,
pp 105-112.
15.S.Xu and W.R.Tyson, 2011. Recommended practice
for determination of crack-tip opening angle of
structural steels using DWTT specimens. CANMETMTL Report, 2011-03(TRR).
16. CSA Z245.1-07, 2007. Steel pipe. Canadian Standards
Association.
17.BS 7448, 1997. Fracture mechanics toughness tests,
part 2, method for determination of KIC, critical
CTOD and critical J values of welds in metallic
materials. British Standards Institution, London.
e
pl
m
Sa
py
co
io
n
rib
ut
is
t
or
d
ot
f
-n
3rd Quarter, 2013
175
Drop-weight tear test application
to natural gas pipeline fracture
control
io
n
by Dr Robert Eiber
rib
ut
Robert J Eiber Consultant Inc., Columbus, OH, USA
T
ot
f
or
d
is
t
HE DROP-WEIGHT tear test (DWTT) has had a significant positive impact on the fracture properties
of linepipe steels. This review summarizes the incidents that started the research leading to the
development of the DWTT from 1960 to present. The initial driver for the development of the test was
an incident that involved 8.3 miles (13.3 km) of brittle fracture during pre-service testing of a natural gas
pipeline with gas. The initial goal of the DWTT was to accurately define the ductile-to-brittle transition
temperature of pipeline steels to facilitate the specification of transition temperatures below the operating
temperature range for linepipe. As the pipeline industry used the low-transition-temperature steels, the
need for a measure of the steel toughness emerged to control ductile-fracture propagation arrest leading to
examination of the DWTT energy as a substitute for the Charpy V-notch energy which had been identified
as the way to define the steel fracture toughness.
T
Control of brittle fracture
propagation
Sa
m
pl
e
co
py
-n
HE GOAL OF this paper is to review the need for,
development of, and application of the drop-weight
tear test (DWTT) for controlling fracture propagation
in natural gas transmission pipelines. In the 1950
to 1960 time period, gas transmission pipelines were
being installed that used 16 to 30-in (406 to 917-mm)
diameter pipes. As the pipe producers were able to
produce larger-diameter pipes and higher-strength steels,
a problem occurred during pre-service testing of the
pipes using natural gas as the pressurizing medium.
The problem was the unstable propagation of brittle
fractures that could propagate for hundreds of feet
before arresting. In 1960, a new 36-in (914-mm) diameter
line involving 56,000-psi (386-MPa) yield-strength steel
was being constructed and, in moving gas from one
section to another for gas testing, a fracture initiated
at a shipping fatigue crack alongside the doublesubmerged-arc seam weld at a pressure of 63% SMYS
(specified minimum yield strength). One to six fractures
occurred running a total of 8.3 miles (13.3 km) before
arresting. This test failure stimulated a need to solve
the brittle-fracture problem and research was initiated to
define the problem and develop a solution to achieve
fracture control.
Author’s contact details:
tel: +1 614 538 0347
email: [email protected]
Research was performed by the AGA (American Gas
Association) Pipeline Research Committee at Battelle
which indicated that – depending on the brittleness of
the fracture surface – low-shear-area fractures could run
at speeds of 1100 to 2300 fps (335 to 701 mps), and
since the natural gas can only escape from the pipe at
its acoustic velocity of about 1300 fps (396 mps), the
pressure driving the fracture could not decay and was
the initial line pressure. The fracture appearance of the
high-speed fractures was 10-20% of the shear area. The
initial experimental results indicated that if the minimum
pipe temperature was at or above a shear area of 85%
in the Charpy V-notch (CVN) impact test, the fracture
would be assured of arresting in a pipeline incident. This
indicated that the way to ensure brittle-fracture-propagation
control in the event of an incident was to require the
linepipe steels to have a transition temperature, as defined
by the CVN impact test shear area of 85%, below the
minimum operating temperature of the pipeline.
The use of increasing wall thicknesses in pipelines lead to
discovery that the 2/3 thickness Charpy V-notch impacttest specimen did not accurately predict the transition
temperature of pipe thicknesses greater than 0.375 in
(9.5 mm). This led to the development of a test specimen
that used the full thickness of the pipe wall in order
to get a 1:1 correlation of the laboratory test shear-area-
The Journal of Pipeline Engineering
io
n
176
is
t
The DWTT test was standardized in ASTM E436 [1]
and API 5L RP5L3 [2]. The full-scale test research
indicated that this should solve the problem of fracture
propagation control in natural gas transmission pipelines
by making sure the pipe steel was ductile.
Sa
m
pl
e
co
py
-n
ot
f
• The fatigue pre-cracked Charpy specimen was
examined but did not show an improved
correlation over the standard mill-notched
specimen, mainly because the specimen was a
constant thickness for all pipe wall thicknesses.
• The US Navy tear test was modified by pressing
a sharp notch in place of the drilled hole. The
test was slow and expensive and the correlation
was poor because of the static loading.
• The Robertson thermal-gradient test was found
to exhibit crack-arrest temperatures approximately
30oF (17oC) below the full-scale crack-propagation
transition temperature because there was not
enough energy in the test specimen to drive
the fracture in contrast to a fracture in a
gas pipeline.
• The US Naval Research Laboratory (NRL) dropweight test (ASTM E 2081) aimed at defining
a nil-ductility transition (NDT) temperature was
examined, but the NDT temperature was found
to be approximately 60oF (34oC) below a pipe’s
ductile-to-brittle transition temperature. The main
concern was that the test did not measure
the ductile-to-brittle transition but rather the
temperature where ductility started to appear in
a steel sample under a low strain rate.
• The dynamic tear test (ASTM E 604) was
examined as it was being developed by the
NRL. One of its problems was that the
specimen was a constant thickness; in
discussions, NRL staff suggested the concept of the
drop-weight tear test which was explored
and developed.
drop-weight tear test (DWTT) because falling weighted
tups were used to break the specimen over a range of
temperatures. Figure 1 shows a comparison of the CVN
specimen with a DWTT specimen size and orientation
in the pipe: the test specimens are selected to measure
the properties in the longitudinal direction of the pipe
as this is the direction that a fracture propagates.
or
d
transition temperature with the pipe-transition temperature.
There were a number of specimens examined in the
development of the new test:
rib
ut
Fig.1. Comparison of DWTT
with a 2/3 thickness CVN
specimen (photo courtesy
of PRC International).
A 3 x 12-in (76 x 305-mm) full-wall-thickness specimen was
found to accurately predict the pipe-transition temperature
regardless of the wall thickness. The test is called the
Figure 2 presents the dimensions of the DWTT specimen
and support for the impact test. The notch is pressed
into the test sample using tool steel machined with a
45o angle and sharpened to a fine point. The resulting
notch radius has some variability but generally is
0.001 in (0.025 mm), and the velocity of the hammer
at impact is to be greater than 16 ft/sec (4.88 mps).
The energy available in the hammer must be adequate
to break the specimen into two pieces. As the toughness
of steels has increased, the available energy of pendulum
testing machines has had to keep increasing. The energy
available to break a specimen should be at least 25%
greater than the energy absorbed by the specimen, but
this has not been examined in great detail.
Figure 3 shows a series of DWTT specimens tested
over a range of temperatures (shown in Fahrenheit)
in the figure. The significant feature of the fracture
surfaces is that even at the highest temperatures the
fracture was initiated as a brittle fracture shown by
the light grey fracture surface. The pressed notch cold
works the steel at the notch, resulting in an embrittled
region where the fracture initiates as a brittle fracture.
Figure 4 compares the results of a 2/3-thickness CVN
specimen and a full-thickness DWTT specimen from a
30in (762-mm) diameter 0.375-in (9.5-mm) wall thickness
177
io
n
3rd Quarter, 2013
ot
f
or
d
is
t
rib
ut
Fig.2. DWTT specimen
and support dimensions
(figure courtesy of PRC
International [3]).
py
-n
Fig.3. DWTT specimens
tested over a range of
temperatures (Note:
pressed notch is at the
bottom of the fracture)
(courtesy of PRC
International [3]).
co
X-52 pipe. The DWTT results exhibit an abruptness in
the transition from brittle to ductile behaviour over a
narrow temperature range as contrasted to the CVN results.
Sa
m
pl
e
The question was which test represented the fracturesurface appearance in pipes pressurized with natural
gas. To resolve this, four full-scale fracture tests were
conducted over a range of temperatures on a single
length of 30-in diameter by 0.375-in wall thickness
X-52 pipe as indicated by the four solid data points
in Fig.5. The fractures were initiated in the pipes at
temperatures from 27o to 107oF (-2.7o to 42oC), and
the calculated stress levels varied from 73 to 78%
SMYS. It can be observed that the DWTT reproduces
the abrupt ductile-to-brittle transition region displayed
by the fractures in the full-scale pipe fracture tests.
Figure 6 presents the appearance of fractures in pipe
tests (right side photos) that were approximately 15-ft
(4.5-m) long pipe specimens pressurized with 6 to 10%
nitrogen. The sections through the fractures (left side
photos) show the difference in thickness reduction with
increasing ductility of the fracture.
To evaluate the effect of pipe thickness on the
transition temperature, a 0.750-inh (19-mm) steel
plate had DWTT specimens prepared from it in a
0.750-in (19-mm) thickness as well as 0.5-in (12.5-mm)
and 0.375-in (9.5-mm) thicknesses. The reduced-thickness
specimens were all machined from the mid-thickness
of the plate to keep the microstructure as constant as
possible. In addition, 2/3 thickness CVN specimens
were also prepared from the mid-thickness of the plate
for comparison. The results, presented in Fig.7, show
that as the DWTT thickness decreased the transition
temperature identified by the DWTT specimen also
decreased. The CVN curve was the same for all three
DWTT thicknesses since the specimen was a constant
thickness. The transition temperature shift from 0.375
in (9.5 mm) to 0.50 in (12.5mm) was 12oF (7oC), and
from 0.50 in (12.5 mm) to 0.75 in (19 mm) was 16oF
(9oC), indicating that the transition temperature shift
is not linear with thickness increases.
The temperature in the DWTT at 85% shear area was
selected as the indicator of a pipe’s transition from
ductile to brittle behaviour. The question that occurred
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
178
e
co
py
-n
ot
f
or
d
Fig.4. Comparison of DWTT and CVN test results on
0.375-in wall thickness X-52 pipe (figure courtesy of
PRC International [3]).
m
pl
was “Is this all that is necessary for the specification of
pipe that would control fracture propagation and avoid
long fractures in service?” The concern was whether it
was also necessary to specify a toughness level.
Sa
To evaluate this concern, several full-scale fracture tests
were conducted which involved 673 ft (205 m) of
pipe with a 150-ft (46-m) test section in the centre in
which a fracture was initiated with an explosive linear
cutter. The tests were conducted at stress levels from
70 to 72% SMYS on X-52 and X-60 pipe, and in
the tests the fractures were initiated above the DWTT
transition temperature. The fractures all arrested quickly
in each direction from the origin, and this appeared
to confirm that all that was necessary was to order
pipe with a DWTT transition temperature below the
lowest operating temperature.
Fig.5. Comparison of full-scale fracture
appearance versus DWTT and 2/3 CVN
(figure courtesy of PRC International [3]).
At this time, the pipe manufacturers started to produce
pipe with low transition temperatures regardless of the
CVN energy level.
Control of ductile fracture
propagation
In December, 1968, there was a service incident
involving an 850-ft long fracture in a new 36-in (914mm) diameter by 0.375-in (9.5-mm) wall thickness
X-65 pipe pressurized with natural gas. The fractures
in all pipe lengths were 100% shear. The obvious
question was why the fractures had not arrested as
they had in prior full-scale burst tests. The one piece
of evidence that pointed to the problem was that
the pipes through which the fractures propagated had
relatively low 2/3 CVN plateau energy levels of 11 to
179
is
t
rib
ut
io
n
3rd Quarter, 2013
py
-n
ot
f
or
d
Fig.6. Photos of fractures in the pipe-fracture
tests in Fig.5 (photos courtesy of PRC
International [3]).
e
co
Fig.7. Effect of thickness on transition
temperature based on the DWTT
(figure courtesy of PRC International [3]).
Sa
m
pl
13 ft-lbs1 (15 to 18J), and the two end or arrest pipes
had 14 and 22 ft-lb (19 to 30J) 2/3 CVN plateau
energy levels. The 2/3 CVN plateau energy of the
origin pipe length was 18 ft-lbs (24J) which may have
been high enough to have arrested the fracture, but
the pressure driving the fracture in the origin length
was higher than after steady-state fracture propagation
was established which is the reason for the propagation
of the fracture through the origin pipe length.
Research was again initiated by the AGA Pipeline
Research Committee at Battelle to develop a means
of controlling long-ductile-fracture propagation.
The CVN energies at the failure temperature were on the plateau and thus
the same as the values quoted.
1
Full-scale burst tests were conducted using 36-in
(914-mm) diameter pipe similar to that involved in
the service failure, which lead to the development
of a model that has been called the two-curve
model (TCM). Figure 8 presents the results
of the TCM for 36-in (914-mm) diameter by
0.375-in (9.5-mm) wall thickness X-70 pipe at
67% SMYS. The predicted CVN arrest toughness is
43 ft-lbs (58J) determined when the two curves are
tangent, which is the borderline condition for
arrest. Higher-toughness pipes (i.e. 53 and 63 ft-lbs
[72 and 85J]) do not intersect the gas-decompression
curve, indicating arrest, while lower-toughness curves
(i.e. 33 ft-lbs [45J]) intersect the gas-decompression
curve, indicating the fracture would propagate at
the intersection speed (i.e. 575 fps [175 mps]) for a
considerable distance.
The Journal of Pipeline Engineering
Fig.8.Two-curve model
prediction of arrest
toughness.
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
rib
ut
io
n
180
With the knowledge that a given CVN energy level was
necessary for the arrest of a ductile fracture, it became
apparent that not only was it necessary for the pipe to
be operating above its DWTT transition temperature,
but also that it needed to have sufficient CVN impact
energy to arrest a ductile fracture. Thus, the use of
both the DWTT transition temperature and the CVN
energy/toughness in a pipeline design can control brittle
and ductile fracture propagation, which provided fracturepropagation control for the vintage pipeline steels.
Fig.9. Comparison of the
three correction factors.
Recent observations with newer
high-toughness steels
The pipe manufacturers armed with the knowledge that
CVN plateau energy was important began to produce
high-toughness steels with low transition temperatures.
These high-toughness steels were achieved by removing
impurities of carbon and sulphur, with the result that
controlled-rolled or thermo-mechanically-processed
(TMP), steels can routinely achieve at least 200 ft-lb
3rd Quarter, 2013
181
io
n
rib
ut
CVNTCM-Leis = CVNTCM + 0.002 CVNTCM2.04 – 21.18(1)
where CVNTCM J is the Charpy energy predicted for
arrest using the TCM.
For X-80 steels, the Leis equation has been modified
as shown in Equn 2 [4]. The terms are the same as
in Equn 1, and the equation only applies to CVN
energies greater than 70 ft-lbs (95J).
ot
f
• Pre-compressed regions in the area of the notch
to embrittle the steel. Unfortunately, this did
not solve the problem.
• TIG (tungsten inert gas) spot welds which
embrittled the area under the notch, but which
did not solve the problem.
• Brittle weld beads (similar to the brittle weld
bead in a drop-weight test, ASTM E 208) across
the notch location with a machined notch
to initiate the fracture. This was effective in
initiating a brittle fracture most of the time
but the behaviour was erratic and occasionally
the complete weld would pop out of the
specimen during impact testing, and the idea
was therefore discarded.
• Fatigue-cracked pressed notch involved a pressednotch specimen being subjected to a cyclic
bending load to fatigue crack the notch. The
fatiguing, because of the residual stress field
from specimen flattening and the notch pressing,
did not result in a uniform depth fatigue crack
and therefore was not acceptable.
• Chevron notch was machined into the specimen
with a depth of 0.2 in (5 mm) at mid-thickness
with a notch included angle of 90o through the
thickness. The problem that was evident from
examination of the fracture surface was that
reducing the thickness of the notch tended to
introduce more ductility into the steel, which
was the wrong direction to go.
• Pre-cracked pressed notch involves statically
loading a standard PN (pressed-notch) specimen
in the normal testing fixture used to hold
the specimen for impact testing until the
static bending load develops a 0.2-in (5-mm)
crack depth below the tip of the pressed notch
which usually develops after a load decrease
of 1.5% after the maximum load has
been reached.
Another problem that has been encountered with the
new high-toughness high-strength linepipe steels is that
the TCM-predicted CVN energy level has been found
to under-predict the toughness of some of the newer
TMP steels (ranging from X-70 to X-100). As a result,
various factors to correct the TCM prediction of arrest
CVN energy have been proposed [4]. The correction
factor developed by Leis for X-704 is represented in
Equn 1. (Note: The Leis correction equation is only
applicable if the measured CVN toughness is greater
than 70 ft-lbs (95J).)
is
t
One of the problems was that it became more difficult
to initiate a brittle fracture from the notch that would
propagate all the way across the specimen. This led to
examination of other notch types to see if an improved
notch could be found that would reduce the energy
required to initiate a fracture and introduce a brittle
fracture. The notch types examined were:
a brittle crack, the PN-DWTT specimens could be
pre-cracked and then tested without having to make
new test specimens.
or
d
(271J) CVN energy levels on the upper plateau. Also,
the steels are generally thicker because of the trend
toward higher operating pressures in gas transmission
pipelines. However, the combination of the higher
toughness and increased thickness caused problems
in using the DWTT for their evaluation.
Sa
m
pl
e
co
py
-n
CVNTCM-Leis = CVNTCM + 0.003 CVNTCM2.04 – 21.18(2)
The most promising notch revision was the pre-cracked
pressed notch. One advantage was that if the initial
testing indicated a problem with the initiation of
The third correction factor is 1.7 times the CVN
TCM prediction as proposed by CSM [4] and shown
in Equn 3. Factors have also been developed for X-80
and above, as indicated in Ref. 4.
CVNTCM
CSM
= 1.7 CVNTCM(3)
Figure 9 shows the trends of the three correction
factors. When compared to the burst-test results, the
Leis X-70 and X-80 correction factors generally split the
non-conservative propagate points and the conservative
arrest data points from full-scale burst tests. The CSM
X-70 correction is somewhat more conservative than
the Leis X-70. The trend is that as the pipe grade
(yield strength) increases, the magnitude of the factor
also needs to increase.
The reader should examine the individual grade
correlations developed in Ref.4, which appear to be
grade-dependent as well as possibly mill-dependent.
It would be nice if a single correlation could be
developed for a test such as the DWTT that could
be used to predict the transition temperature and the
ductile-fracture arrest energy. Additional development
182
The Journal of Pipeline Engineering
The DWTT has been a major step forward in solving
pipeline fracture-control problems considering where
the industry was in the 1950s and 1960s. The DWTT
is used by the pipeline industry internationally for
specification of transition temperatures.
io
n
1. American Society for Testing and Materials. Book
of Standards, Part 10.
2. American Petroleum Institute. Specification for Line
Pipe, Recommended Practice 5L3.
3. R.J.Eiber, T.A.Bubenik, and W.A.Maxey, 1993.
Fracture control technology for natural gas pipelines.
AGA NG-18 Report 208, Catalog L51691, December.
4. R.J.Eiber, 2008. Fracture propagation 1, Oil & Gas
Journal, Fracture Propagation, 20 Oct; and Fracture
Propagation – Conclusion. Idem, 27 Oct.
Sa
m
pl
e
co
py
-n
ot
f
There are still problems occurring with the use of the
test on high-toughness heavy-wall-thickness steels. The
potential usage of the DWTT test energy as a substitute
for the CVN impact energy would be ideal as this would
eliminate the need for two tests to be conducted in the
pipe mill. At the moment, the industry needs to conduct
full-scale burst tests to validate fracture-control plans for
the newer high-strength and high-toughness grades of steel.
References
rib
ut
Conclusions
Which of the parameters CVN or DWTT should
be used, and how to define an arrest toughness for
describing the ductile-fracture propagation behaviour for
extremely high-toughness steels in Grade X-100 or above,
is still a challenge. The main conclusion is that both
a transition temperature and a toughness level (impactenergy level) are required to achieve fracture control.
is
t
As far as is known, no company has tried to order
pipe to a DWTT energy criterion as of this date.
One potential problem is that the measurement of
the DWTT energy has not been standardized. In order
to implement a DWTT energy requirement, this will
be a necessity.
One of the developments in the recent years is that
with increasing fracture toughness of the pipe steel,
there has been a tendency to specify only the CVN
energy necessary for ductile fracture arrest. This is a
mistake, and can lead to undesirable consequences, as
the transition temperatures of some of the newer steels
have tended to increase if a DWTT-based transition
temperature is not specified.
or
d
is needed before we can reach this point. One issue
is that the DWTT energy measurement has not
been correlated with ductile-fracture arrest, nor has
a machine design been proposed that will produce
reproducible energy values.
3rd Quarter, 2013
183
The Charpy impact test and its
applications
by Dr Brian N Leis
io
n
B N Leis, Consultant, Inc, Worthington OH, USA
T
ot
f
or
d
is
t
rib
ut
HIS PAPER REVIEWS the Charpy V-notch (CVN) impact test and assesses its utility to characterize
fracture resistance in applications to modern tough materials in contrast to those encountered prior
to the availability of such materials.The origin of the CVN test and its development into a standard for use
with metallic materials is discussed, with brief reference also made to application-based standards for use
with other engineering materials. Thereafter, the evolution of mechanical and other properties motivated
by industry demands is illustrated in regard to strength and toughness.The interpretation of the CVN test
in regard to (1) the force-displacement and compliance response that develops during the test, and (2)
factors affecting the energy measured and controlling failure of the CVN specimen, are discussed, including
the tup design and the use of sub-size specimens. The utility of CVN testing is illustrated and discussed in
the context of pipeline and other applications involving tough steels. Finally, the implications of evolution
in material properties is assessed for impact-test practices including ASTM E23 and ISO 148-1, which are
specific to the CVN practice, and the drop-weight tear test. It is concluded that where tough materials are
involved, alternative testing practices are needed that are better adapted to the specific loading and failure
response of the structure of interest.
A
more-resistant ones would only bend, such that a notch
was introduced to facilitate initiation circa the 1890s
[3]. The use of a pendulum-based impact test – whose
use permitted determination of the energy absorbed in
failure – is first evident in Russell’s work circa the mid
1890s [4]. Like the earlier drop-weight machines, his
practice made use of un-notched samples. Soon after
the turn of the century, Charpy [5] improved Russell’s
method by introducing a redesigned pendulum, a notched
sample, and more-precise specifications, thereby moving
the impact test toward a more-standardized practice.
Charpy’s name appears to be associated with impact
testing more because of his efforts to standardize the
practice over the ensuing years than because of his
role in the development of the hardware.
Sa
m
pl
e
co
py
-n
WEBCRAWL ON the topic of impact testing
and related standards develops more than 850,000
hits which, after allowing for duplicated hits and/or
otherwise similar publications, makes clear that this
topic has a broad following. Consideration of the
dates of the citations found in such a search range
from current times, with the history uncovered in this
process indicated [1] to trace back to the min-1800s.
Accordingly, impact testing also reflects an enduring
concern that addresses the integrity of structures.
Because the origins of impact testing have recently
marked a century of development, reviews have been
written that document this history (Ref.1, for example),
which note the early work was motivated by a need
to qualify materials resistant to impulse loadings. It is
apparent that the early work centred on military and
railroad applications.
The need to screen materials for their resistance to
impulse loadings led to the development of the first dropweight machine, which emerged circa 1860 [2], although
it was used then in testing smooth rectangular bars (i.e.
without a notch or starter crack). While that practice
was found adequate for the less-resistant materials, the
Author’s contact details:
email: [email protected]
Recognizing that impact testing was first developed to
screen ferrous structural materials in the context of what
has since been termed brittle-fracture resistance, a host
of variations on the initial impact-test practice evolved
to quantify resistance, as did other practices designed
to understand the failure mechanism. The documents
evaluated indicate that variations developed in the test
practices related to the nature of the imposed loading
and test-specimen geometry, as well as in the notch
shapes and the use of localized treatments to affect
desired changes in the nature of the failure process. As
the significance of strain rate became evident, practices
were introduced to vary the speed of the impact, with
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
184
(a)
(b)
(for example, Refs 10 and 11). Some factors known to
affect variability include the dynamic stiffness of the
system, differences in the tup geometry (radius and
overall width and angle to the striking edge), and the
use of sub-size (reduced-thickness) specimens, with the
nature of the material being characterized also being a
major consideration in regard to the energy required to
fail the specimen, and steel cleanliness. Suffice it here
to note that the significance of some of these factors
is considered later in more detail.
pl
e
co
py
-n
ot
f
slow-bend practices introduced in complement to the
impact machines. In a related vein, it became evident that
sufficient energy was required to continue to drive the
failure, such that �excess energy’ became a consideration.
In this context, it has been noted [6] that “more than
50 types or variations of tests can be clearly identified
for evaluating the susceptibility of materials to brittle
fracture” that include well-known through now-obscure
practices. Examples of some well-known tests aside from
the Charpy V-notch (CVN) practice include the Izod test,
the explosion bulge test, the Pellini crack-starter test, the
NRL-NDT drop-weight test, and the Battelle drop-weight
tear test (DWTT), while some lesser-known practices
include the Kahn tear test, the Bagsar cleavage-tear test,
and the Charpy keyhole and U-notch tests. These and
other practices can be found with more details in the
citations listed in Reference 6, beginning at page 316.
While many such impact tests exist, the focus hereafter
is on the CVN test.
or
d
Fig.1.Trends in CVN energy over time, adapted from Refs 18 and 19: (a) energy trends with grade, circa 1969; (b) historical
view of plateau energy and grade.
Sa
m
CVN test development and
applications-based standards
The CVN test is a simple, low-cost, and reliable test
method that today is commonly required by regulatory
and other codes for fracture-critical structures such as
pipelines, bridges, and pressure vessels. Today, three
standards are prominent in the general use of the CVN
test [7-9] that are relevant for applications to �metallic
materials’, and provide a viable basis for the consistent
use of this practice. In contrast, the history of the CVN
practice from about 1900 through to 1960 is marked in
regard to data scatter and variability, more generally in
regard to the factors that adversely affect its outcome
Realizing that it took until the 1960s before the CVN
test transitioned from a qualitative to a practically
useful quantitative method, issues due to factors such
as those just cited persisted long after the CVN
practice became a tentative standard of the American
Society for Testing and Materials (ASTM) in 1933.
Then designated E 23-33T1, this tentative practice
somewhat loosely covered two notched-bar impact
practices that differed significantly in regard to the
methods to support the test specimen (i.e. the CVN
test involving three-point beam-bending and the Izod
involving cantilevered beam-bending). The tentative
standard also did not adequately control the striking
edge or �tup’ for either test practice, and it allowed
flexibility in regard to the notch geometry. Adequately
managing such differences was a key to move the
CVN practice from one viewed as qualitative, in part
due to its scattered results, to one that is quantitative
and so can be practically useful aside from screening
materials for fracture mode.
For those less familiar with the ASTM process, the tentative designation is
no longer part of its standards’-development process.
1
3rd Quarter, 2013
185
Evolution in properties: implications
for ASTM E23 or ISO 148-1
io
n
Evolution of materials and properties
is
t
rib
ut
While much has been done to �standardize’ and �specify’
the essential details of the CVN test practice over the
years, the metallic materials being developed for use
today differ significantly from those in the era that
the CVN test transitioned from qualitative into what
is considered today a quantitative test. Figure 1, which
adapts trends from Refs 18 and 19, serves to illustrate
and emphasize this point.
Figure 1a summarizes trends in CVN energy for data
developed up through the late 1960s. The y-axis in the
figure is CVN energy, which is shown as a function of
test temperature on the x-axis. To facilitate comparing
trends for which differences exist in the transition
temperature, this figure centres them at close to the
50%SA temperature, such that this axis is labelled
�relative temperature’. As is evident from the figure,
these trends are presented only in the units reported,
because the format of the plot precludes presenting
dual units. Results are shown for several steels whose
yield stresses, denoted sy, range from 40 to 200 ksi
(276 to 1378 MPa). These trends are shown as the
dash-double-dot lines. Results also are presented for two
aluminium alloys with yield stresses of 38 and 75 ksi
(261 to 517 MPa), which are shown as dashed trends
toward the lower edge of this figure.
py
-n
ot
f
Thus, although today the ASTM E23 standard still covers
these quite different bend-bar impact tests, and also still
allows different notching practices, it now makes clear
the differences between these tests and the notching
practices, and addressed those differences through the
requirements of the standard. Unfortunately, differences
remain today between the geometry of the �tup’ in what
might be considered a North American [7] and European
standards [8], which can open to differing outcomes
particularly when evaluating more resistant materials.
because the force versus load-point displacement record
from such set-ups helps understand the extent to which
the substituted material behaves differently from the
material it is replacing.
or
d
Over the years, aspects unique to the specimen
geometries and loadings have been identified and
understood such that today the standard embeds clear
requirements for the test specimens, as well as for the
testing procedures and the reporting. Significantly the
reporting requirements prescribe clear identification
of the type of test and specimen, as well as other
details that could promote what otherwise lead to
what could be misconstrued as data scatter. Annexes
to the standard have been developed that address the
test machines, including verification of the Charpy
impact machine. These annexes also outline optional
specimen configurations, and discuss CVN pre-cracking,
specimen orientation, determination of the percent of
shear fracture area (%SA), and methods to measure the
centre of strike. Finally, the significance of notched-bar
impact testing is considered in regard to the correlation
between the test practice and its outcomes with the
service situation. There it notes that the Charpy or
Izod tests may not directly predict the ductile or brittle
behaviour of steel – an aspect that is elaborated later
in the section that illustrates the application of the
CVN test to specifying linepipe steels.
Sa
m
pl
e
co
In addition to the above-noted standards that focus on
metallic materials, such as steel, other standards have
evolved based on the basic tenets of ASTM E23 that
focus on miniaturized testing and address materials other
than the types of materials that were targeted by the
original standard (for example, 12-16). Such standards
have evolved to address the unique concerns of polymers
and cast iron, for example, and are developed to meet
the needs specific to the application involved. Needs
for an application-specific alternative to E23 can become
apparent as new materials emerge that will find use under
impulse-loading conditions, or under circumstances that
otherwise could promote a change in fracture resistance
due, for example, to cold-temperature service, or irradiation.
Materials substitution also can create applications wherein
an existing material is replaced by an emerging material,
which needs to be screened for changes in its fracture
resistance over the range of operational circumstances
that are unique to its adapted use. Note that, in this
context, the use of an instrumented testing system,
consistent with standards [17], can facilitate the use of
impact testing in cases involving material substitution or
new applications concerned with impulse loads. This is
It is evident from Fig.1a in regard to the data shown
for the steels that these CVN trends up through the
plateau (upper-shelf) energy, denoted CVP, indicate that
the resistance to fracture (or toughness as characterized
by CVP) decreases as the yield stress increases, and
that this decrease is significant. This same trend is also
evident for the aluminium alloys. While the trends are
similar for the steels with sy = 200 ksi (1378 MPa), it
is apparent that the energy dissipated in a CVN test is
not uniquely quantified relative to the grade of steel.
Rather, experience indicates that the energy dissipated
reflects the chemistry and processing history of the steel
on a case-by-case basis. Finally, these data indicate that
for steels produced up through the late 1960s a five-fold
increase in yield stress (grade) could affect more than a
ten-fold reduction in toughness. This tendency reflects
the well known tradeoff between strength and ductility
that was characteristic of the carbon-manganese (C-Mn)
steels produced prior to the advent of the high-strength
low-alloy (HSLA) grades that emerged commercially in
the 1960s.
The Journal of Pipeline Engineering
(b) X-100 specimen @ 118 ft-lb (160J)
py
-n
ot
f
or
d
is
t
(a) X-70 specimen @ 250 ft-lb (339 J)
rib
ut
io
n
186
co
(c) fracture features @ 250 ft-lb (339 J)
(d) fracture features @ 18 ft-lb (~24 J)
Fig.2. Differences in CVN response with toughness, adapted from Refs 22 and 23.
Sa
m
pl
e
HSLA steels have evolved over time, coupling chemistry
and processing changes to achieve strength while limiting
the frequency and size of particles that nucleate voids
and trigger the onset of plastic collapse, through the
use of cleaner steels and sulphide shape control [20].
Rolling practices also evolved through use of controlrolled and thermal-mechanically controlled process
(TMCP) techniques that coupled controlled rolling with
accelerated cooling, which subsequently gave way to
high-temperature-processed (HTP) steels. Thus, since the
mid-1960s, it has become possible to couple strength
gains with improved toughness and ductility to enhance
fracture resistance. Such changes have developed in
the context of automotive steels and linepipe steels,
for which weldability also was a major consideration.
Thus, in contrast to Fig.1a, the trends in part (b) of
this figure show a reversal of the decrease in ductility
with increasing grade, with this shift apparent in Fig.1b
circa more or less the mid 1960s. Note that in Fig.1b
that the data points flagged as Q&T (quenched and
tempered) that lie toward the bottom of the interval
for the mid-1960s lie at quite low toughness values,
which is as expected for higher-strength steels based on
the results evident in Fig.1a.
Figure 1b, which presents data for linepipe steels produced
up through circa 2000, indicates that CVP energies
on the order of 250 ft-lb (339 J) were being achieved
toward the turn of the millennium. But, in comparison
with such gains over about a 30-year period, it is not
unusual currently to see energy values approaching
double that level in grades up to X-80 (551 MPa) – all
within the last 10 or so years. It is also apparent that
for grades above X-80, as for example X-120, which are
produced by approaches that differ from those used to
make modern X-80, it is much more difficult to achieve
3rd Quarter, 2013
187
io
n
rib
ut
Figures 2c and 2d contrast the fracture features for results
such as those shown in Figs 2a and 2b, respectively.
While the view in Fig.2d shows evidence of crack
initiation and propagation, as has typically been observed
and expected in the context of the interpretation of
the CVN test as evident for example in ASTM E23,
the image in Fig.2c for the much-tougher steel does not
show the same traits. Rather, it shows significant local
stretching at the notch, very large lateral contraction
below the notch, with thickening occurring toward
the back face, and evidence of some stable tearing
and stretching occurring in lieu of what might be
called cracking.
ot
f
It follows from the significant increase in toughness evident
in Fig.1b, for steels produced recently in contrast to
those of late 1960s, that for materials where high CVN
energies have been achieved the failure response of a
CVN specimen could differ radically from that typically
seen in the late 1960s. The results in Fig.2, which are
adapted in part from Ref.21 and from unpublished work
[22], clearly support this expectation. The images in Figs
2a and 2b are post-test perspective views of full-size CVN
specimens that both showed 100% SA, following testing
in a 512 ft-lb capacity pendulum machine equipped with
unmodified anvils and a Dynatup striker. The machine used
had been certified by the National Institute of Standards
and Technology (NIST) practice [23], and maintained in
calibration. While clearly apparent from the back-face
deformation for the specimen shown as Fig.2a, but less
evident for that shown in Fig.2b, during the test both
specimens had been wedged onto and around the tup
as they were pushed through the machine. Thus, these
specimens were intact after testing – although to a much
different extent for what was the higher toughness case
as compared to that requiring much less energy to push
it through the anvils – as becomes clear, as follows.
is
t
Implications for standardized impact testing
face and flanks of the tup, with the remainder of the
specimen otherwise largely un-deformed, although the
ligament remains intact. Nevertheless, close examination
of the image reveals the imprint of the tup shape on
the back face. As such, this specimen rode the tup
through the anvils, in like manner to that for the
higher-toughness steel. And while a crack did develop,
which was quite deep, manually closing the angle
between the back faces after the test did not break
the specimen into two pieces. A later section further
considers that and other aspects that appear to limit
the utility of the CVN test in applications to highertoughness materials.
or
d
the high toughness levels necessary to fully capitalize on
the strength gained for the applications for which such
grades were designed.
Sa
m
pl
e
co
py
-n
The image in Fig.2a reflects a measured energy of about
250 ft-lb (339 J) for a CVN specimen made of X-70
steel produced circa the mid-1990s. This image focuses
on the area of a specimen in the vicinity of the notch
and fracture, to illustrate the complex nature of tearing
and the significant lateral and local back-face deformation.
As noted in the above discussion, this image clearly
shows that the back-face of the specimen has deformed
around and about the face and flanks of a tup whose
size and shape conformed to ASTM E23, whereby the
striker radius is 8 mm. Had the narrower, smaller-radius
(2-mm) tup of ISO 148-1 been used, less wedge force
would have developed, such that less energy would have
been required to push the specimen through the anvils.
In this context, the total measured energy involves the
energy to deform the specimen, to initiate cracking or
tearing/stretching, and to propagate that cracking or
tearing/stretching through the depth of the specimen,
along with the energy required to push the specimen
through the anvils. Further on this aspect follows later in
the section that deals with factors affecting the measured
energy and controlling failure of a CVN specimen.
Suffice it here to conclude that such observations
indicate that the test in applications to higher-toughness
materials no longer develops features typical of those
observed throughout its historic applications, with a
subsequent section further considering factors affecting
the measured energy and controlling failure of CVN
specimen, which closes considering the implications in
regard to the utility of the CVN test as standardized
by ASTM E23 [7] or ISO 148 [8].
Figure 2b reflects a measured CVN energy of about
118 ft-lb (160 J) for a specimen made of X100 steel,
which was produced just into the new millennium.
This image is in strong contrast to that in Fig.2a, as
only limited local distortion has occurred around the
The CVN test: its interpretation and
utility
The CVN test can be implemented using either a
drop-weight system, which requires an instrumented
tup, or a pendulum set-up than can make use of an
instrumented tup, but does not require it to determine
the dissipated energy. While an instrumented set-up is not
required using the a pendulum, the results of measured
force as a function of the load-point displacement do
provide much insight into the dissipative processes that
contribute to the energy required to fail the specimen.
Output from an instrumented test is shown in Fig.3
as the basis for further discussion on the test and its
interpretation, based on results reported in Ref.24. The
results in this figure reflect transverse CVN samples
cut transverse to the longitudinal axis of a joint of
linepipe, with the notch oriented through the thickness
of the pipe. This orientation corresponds to testing in
the y-x direction as identified in ASTM E23.
The Journal of Pipeline Engineering
rib
ut
io
n
188
(b)
is
t
(a)
Force-displacement and compliance response
bending and tearing coupled with stretching in tougher
steels, or cracking in less-tough steels – the sample
moves on through un-modified anvils. While less-tough
steels fail as two pieces that can be forcefully ejected
from the machine, as indicated in Fig.2a, tougher
steels can be found still wedged around the tup. In
extreme cases, specimens can bend around the tup to
an extent that unloads the face of the striker – which
is also discussed in a later section. Because the format
of this figure precludes the use of dual units, the
results are presented using in the units used to quantify
this response.
Sa
m
pl
e
co
py
-n
ot
f
Force as presented on the y-axis of Fig.3 is as measured
by a Dynatup striker (as made circa the early 1990s) as
a function of the load-point displacement (determined
from pendulum arm rotation) that is shown on the x-axis.
The secondary y-axis is a measure of the specimen’s
compliance, as it presents the x-axis displacement value
divided by the corresponding load. Force – as shown
on the y-axis of Fig.3 – has been divided by the
initial net-section area of the CVN specimen, with that
three-point bending net-section stress then divided by
the average value of the ultimate tensile stress (UTS)
for the steel involved. The value of the UTS used
is based on companion round-bar tension tests made
from flattened blanks, which also were cut in the
transverse (hoop) direction, and tested in accordance
with ASTM E8 [25]. On this basis, the value of the
y-axis should normalize the results for this steel and
others up through either the limit load or fracture,
with the outcome shown on the y-axis being a multiple
of the ratio of the instantaneous load relative to the
limit-load that depends on the bending stiffness of
the CVN sample.
or
d
Fig.3.Typical response trends developed in instrumented CVN testing: (a) moderate to higher toughness; (b) a range from low
to high toughness.
Figures 3a and 3b present results that show the specimen’s
initial elastic response where, after yielding occurs, plastic
bending ensues as the load increases through the limit
load, beyond which the load falls off as it would in
a tensile test as necking develops. The analogue to
necking for tough steels leads to localized thinning
across the notch face of the CVN specimen, with the
response under quasi-static loading similar to that under
impact loading evident – as presented in a later section.
Lateral thinning along the notch is accompanied by
back-surface expansion, during which – after sufficient
The plot shown in Fig.3a presents results for a controlrolled X-80 steel (labelled #3) cut from mill-expanded
line pipe with 0.61-in (15.5-mm) thick wall that was
produced in 1992. Figure 3b supplements this result
with data for an accelerated-cooled X-70 steel from
mill-expanded linepipe with 0.75in (19.1-mm) thick
wall that also was produced in 1992 (labelled #2), a
control-rolled X-60 steel from mill-expanded linepipe
with 0.546-in (13.9-mm) thick wall that was produced
in 1970 (labelled #7), and a conventionally-rolled X-52
steel from mill-expanded linepipe with 0.375-in (9.5-mm)
thick wall that was produced in 1960 (labelled #8).
All involved testing of full-size samples except for the
X-52, which made use of a 2/3 thickness specimens.
All results reflect tests that developed 100 %SA, such
that these trends reflect fully ductile response.
In reference to Fig.3a, a series of dotted vertical lines
have been added to the test record, which have been
located to tie key changes in compliance to events
evident in the force-displacement response. Note in this
context that if the stress-strain response is mapped onto
this coordinate system, the locus traced (assuming the
3rd Quarter, 2013
189
is
t
rib
ut
io
n
In contrast, the force-displacement response developed
by steel #2 is comparable to that shown by steel #3 in
regard to Fig.3a, except that this steel does not show
evidence of a significant reduction in the force required
to move the specimen through the anvils over the course
of the testing. As for the other steels, analysis indicates
that the peak load for this case correspond closely with
what is inferred as the limit load for simple bending
relative to this steel’s UTS. And as for the other steels,
it is evident that the force-displacement response for this
steel is normalized as expected by the use of the ordinate
noted above, with the difference in response up to the
limit load reflecting its unique flow response and strainhardening rate. However, in contrast to the results for
steel #3, which shows the force-displacement response in
the CVN test drops below the locus of the stress-strain
response of the steel when mapped on to this figure,
that trend for steel #2 remains at or very close to the
locus of the stress-strain curve. As such, what in 1997 was
partitioned as largely deformation energy with a smaller
initiation component (for example, see Ref.28) would
more correctly be represented as all deformation energy.
ot
f
The end of the regime associated with �initiation’
is marked by the second vertical line from the left
side of the plot, which lies at the onset of what is
identified in Fig.3a by the label �propagation’ – that
for this steel occurred ductile tearing. Beyond the area
labelled propagation, the response for this specimen
becomes complex in regard to the force-displacement
response and the change in compliance. Fractography
and observations made using a quasi-static set-up infer
that the tearing quickly transitions into rotation and
stretching. While this transition is ill-defined by the
dynamic force-displacement behaviour and related change
in compliance, the quasi-static results indicate the
transition first involves the back-face of the specimen
wrapping around the tup, after which the associated
wedging load causes failure while the specimen passes
through the anvils, and eventually breaks in two.
Accordingly, the response shown in Fig.3a beyond that
labelled propagation is divided into regions labelled
deformation (as the specimen wraps around the tup)
and propagation (as it fails between the anvils).
ordinate discussed earlier, with the differences in trends
up to the limit load reflecting the inherent differences
in their flow response and hardening rates.
or
d
specimen geometry remained as it began) would begin
to drop at the same displacement that marks the peak
load, which is marked by the first of the vertical lines.
The locus tracked by the stress-strain response depends
on the actual strain-hardening behaviour of the steel,
and for this test tracks at a higher loads beyond the
peak load. It follows that, for this test, the peak load
evident in Fig.3a corresponds to the bending limit
load, and that the response up through the limit load
reflects energy dissipated in deforming the specimen. As
such, this initial regime has been labelled �deformation’.
Because of the difference in load drop for this test
as compared to the locus tracked by the stress-strain
response, it also follows that this loss in stiffness (or
increase in compliance) evident beyond the limit load
is associated with local stretching and thinning and the
initiation of tearing, such that this regime has been
labelled �initiation’. This has been done in spite of the
observation that a major component of that energy is
due to continued plastic flow.
pl
e
co
py
-n
Not surprisingly, steels #7 and #8 both broke cleanly in
two pieces. In contrast, steel #2 simply deformed and
showed local stretching and lateral local thinning at the
notch, and expansion toward the back face sufficiently so
to pass between the anvils – and showed no evidence of
the potential to break as a result of the testing. In all
three cases, the compliance trends are as anticipated. For
the pair of steels with the lowest toughness the compliance
increasing sharply once the peak load is reached for
each of these steels. In contrast, the compliance for steel
#2 increases only slightly from its initial trend, which
signifies that little change in bending stiffness occurred
over the course of the test.
Sa
m
As the trends for the other steels shown in Fig.3b indicate,
none shows the complexity evident for the steel just
discussed. The energies measured relative to these trends
for steels #2, #3, #7, and #8 are respectively in ft-lb
250, 192, 17.5, and 36 (in J, 339, 259, 23.7, and 49) –
the last of which is a linearly scaled full-size equivalent
(FSE) energy. The lowest energy pair of results (steels #7
and #8) developed force-displacement response that is
characteristic of steels and other metallic materials that
fail at relatively low energies, which was commonly the
case through the 1960s. Analysis indicates that these peak
loads correspond closely with what is inferred as the limit
load for simple bending relative to their respective values
of the UTS. It is apparent that their force-displacement
response is normalized as expected by the use of the
It follows from the pair of results that developed for the
very tough steels that the CVN test is tending toward
the limits of its utility, whereas as history has shown it
is a viable basis to quantify the response at much lower
toughness levels.
Utility of CVN testing in pipeline applications
A webcrawl also was done on the topic of Charpy impact
testing, but for the purposes of this section it was filtered
with the terms of pipeline and applications. This search
led to well in excess of 600,000 hits, many of which
are duplicated or repeated in some manner. However,
the number of results, the dates of the citations, and
content and scope of the documents make clear that this
testing practice is broadly still used and that the CVN
test remains relevant in the pipeline industry.
The Journal of Pipeline Engineering
Fig.4. Energy
components from
instrumented CVN
testing partitioned
as in Fig.3.
ot
f
or
d
is
t
rib
ut
io
n
190
e
co
py
-n
It is apparent in this context that the CVN test remains
the basis for fracture-control plans for the industry. The
American Petroleum Institute (API) and others adopt
the CVN test as the basis to quantify fracture resistance
(for example, Ref.26), whereas the Battelle drop-weight
tear test [27] (B-DWTT) – that subsequently simply is
identified by the acronym DWTT – is used to characterize
fracture mode. At the same time, literature based on
full-scale testing has been developing that is consistent
with the above noted results (such as Ref.28), indicating
that the CVN test has reached the limits of its utility
in applications to the tough and weldable high-strength
steels being adopted by the pipeline industry.
Sa
m
pl
Figure 4 (adapted from Ref.28) was derived from the
components of dissipated energy as partitioned in the
context of Fig.3 to shed light on the future utility of
the CVN test in applications to high-toughness steels.
The y-axis in this figure is the relative fraction of the
total energy dissipated in each of several tests involving
a total of nine steels whose toughness as quantified by
CVP ranged up to 260 ft-lb (352 J), values for which
are shown on the x-axis. For each test, these fractions
must sum to unity.
As often occurs for CVN tests, the results in Fig.4 show
scatter, which is made worse for the present analysis
by the gradual changes in compliance that make it
difficult to identify the breakpoints between the energy
components. Although there is scatter, the key point
in regard to Fig.4 is not clouded by such concerns: it
is clear from the figure that the trend for propagation
energy decreases to zero a value of CVP at about
250 ft-lb (339 J) and above, as indicated by the data
points circled in the figure. This outcome is anticipated
in light of Fig.3b, and the related discussion, because
the force-displacement response for steel #2 (which had
a CVP of 350 ft-lb (339 J)) involved only bending and
stretching, with lateral contraction across the notch
and expansion on the back face – but no propagation.
In contrast, the models developed in the 1970s to
characterize the fracture-arrest process, which remain in
use today to quantify the toughness required to arrest
the phenomenon termed running fracture, were calibrated
in a framework that involved only lower-toughness steels
that Fig.4 shows involved a significant component of
crack-propagation energy.
Clearly one can question the use of the above-noted
fracture-arrest models that assume fracture controls
failure in applications where the CVN test does not
produce cracking. One could also question the initial
use of the CVN test – which involves a notched beam
in simple bending – for use in applications for which
little or no bending occurs. Justification for that choice
was as easy then as it is today – as the models were
empirically calibrated such that the value of CVP could
have been a surrogate for some other related metric.
In addition, the basic concepts and tools of fracture
mechanics were in the early stages of development – but
more critically the phenomenon that was then termed
�propagating fracture’ is complex and since has defied a
3rd Quarter, 2013
191
rib
ut
io
n
(a)
is
t
(b)
the use of full-thickness samples, such as the DWTT. It
is, however, clear that as time has passed the utility of
the DWTT even to quantify the fracture mode, which
is its purpose, is open to question. Issues with so-called
inverse fracture have become widely evident since about
2000, and as the steels have become increasingly tougher
the DWTT practice is also developing gross stretching
and evidence of overall distortion – as similarly has
occurred for the CVN practice.
-n
ot
f
first-principles’ formulation. Today it is understood that
this phenomenon first involves axial straining ahead of
the crack in a plastic zone that propagates along the
pipeline, with symmetric thinning leading to �propagating
shear’ as shown in Fig.5, with in-plane stretching occurring
until shear failure (plastic collapse) occurs through-wall.
or
d
Fig.5. Cross sections through propagating shear contrasted to tearing shear (from Ref.27):
(a) symmetric shear (as polished, observed during steady-state propagation): t = 0.560 in (14 mm);
(b) asymmetric tearing shear (rough grind, observed during ring-off arrest): t = 0.560 in (14 mm).
m
pl
e
co
py
The significant in-plane stretching evident in Fig.5a
results in symmetric through-wall thinning and eventual
collapse via shear – much as could occur in a tensile
test. This symmetric process develops during steady-state
propagation, which corresponds to the scenario the
CVN test has used to characterize the arrest toughness.
Figure 5b shows the transition that occurs in this failure
behaviour as steady-state transitions to arrest, which
often results in the ring-off of the failure path. While,
as Fig.5b indicates, this transition leads to asymmetric
(so-called tearing) shear, the failure process during arrest
remains shear dominated – with much less thinning
occurring, as now the driving force for the process is
greatly diminished.
Sa
It follows that at least for pipelines made of highertoughness steels the usual CVN test practice appears to
have limited utility. Whereas the steady-state pipelinefailure process primarily involves stretching and thinning
that is largely free of bending, the CVN test relies
on bending with thickening developing as the failure
process in the specimen continues to the back face.
In this context, any bending-impact test falls short of
emulating the structural response. In spite of this fact,
it has been asserted that the issues with the CVN test
in pipeline applications evident in the stretching, tearing,
and lateral flow parallel to the notch could be offset by
Figure 6 (courtesy of Bernard Hoh) provides graphic
evidence of the gross distortion that can develop through
the use of the DWTT in applications involving tough
steels. While these images reflect heavy-wall linepipe made
with a thickness of 1.5 in (38 mm) in X-70 steel, the
results reflect testing done in a machine with adequate
excess energy. Thus, while considered by some a potential
alternative to the CVN practice in anticipation that its
increased depth and larger overall size would offset such
concerns, the DWTT practice does not provide a path
forward for such applications, even as a stopgap solution.
On this basis, work is needed to better understand
the factors controlling the failure process such that an
appropriate mill test can be identified in lieu of the
bending impact tests in use today.
Utility of CVN testing in other applications involving
tough steels
Central to the discussion of the CVN test in its application
to pipelines was the thesis that the issues discussed can
be traced to the toughness of the steel. This section
further considers this hypothesis in regard to the classes
of steel in use in the ground-vehicle industry.
The Journal of Pipeline Engineering
or
d
is
t
rib
ut
io
n
192
py
-n
ot
f
(a)
co
(b)
Sa
m
pl
e
Like the pipeline industry, the automotive / groundvehicle industry has pushed the development of tough,
strong, weldable steels. As such, it is not surprising that
about the same time that issues with the CVN test
were emerging in applications to the pipeline industry,
similar concerns with the CVN test became evident in
that industry. For example, a 1987 paper [29] authored
from within the Structures and Dynamics Division of
Caterpillar Inc. notes that “recently, two new families of
steels …… have become available” that were of interest
to that industry. After considering those steels in light
of several basic materials’ tests, the paper concludes by
noting that some tests are “reliable indicators of material
performance in components while Charpy V-notch energy
is of little value for quantitative engineering analysis”.
The paper goes on to state that an “alternative impact
test which more closely represents material behaviour
in components” is needed.
Fig.6. Unusual fracture and
gross distortion in a DWTT of
high-toughness steel:
(a) �fracture’ features (notch is up);
(b) perspective view of the tested
specimen.
Another paper associated with the evaluation of highpower beam welds (such as electron-beam or laser
welds) noted that the failure of such welds might
result in a plastic constraint loss around both the
notch and crack tip, making it difficult to evaluate
fracture performance of girth-welded pipe joints. This
paper determined that intrinsic fracture toughness was
lower than the results of standard Charpy specimens
and fatigue pre-cracked three-point bend specimens.
It was concluded that differences in plastic constraint
between the structure and a three-point bend specimen
underlie this situation.
Thus, in addition to applications where plasticity acts
to limit the utility of the CVN test, as occurred for
the pipeline and the ground-vehicle cases, issues can
develop that limit the direct transferability of the
CVN test results due to constraint – in the absence
193
rib
ut
io
n
3rd Quarter, 2013
py
-n
ot
f
or
d
is
t
(a)
co
Fig.7. Energy measured in a
quasi-static evaluation of forcedisplacement response [21]:
(a) photographs of the set-up;
(b) results of the comparison.
pl
e
(b)
Sa
m
of the significant effects of plasticity. It follows that
care should be exercised in adapting what otherwise
begins as screening tests to more general applications
Factors affecting the energy
measured in a CVN specimen
Earlier discussion alluded to the use of subsize specimens
and differences in tup design as drivers for variability
in the measured energy and differences in the failure
response in a CVN specimen. As both of these
aspects have been the subject of related research, this
section briefly elaborates on such effects in regard to
measurements directed at assessing their significance.
Tup design issues
Consider first the role of tup geometry, which for the
present is evaluated specifically in regard to the energy
measured via ASTM E23 (8-mm radius striker) rather
than contrasting such results relative to the outcome for
ISO 148-1 – with the results shown in Fig.7 developed
in the late 1990s being instructive in this context.
Figure 7a shows a view of the specimen after loading
well beyond the limit load, whereas part (b) of this figure
shows the force-displacement response as measured by a
Dynatup striker mounted in series with a load cell and
a linear variable differential transformer. It is evident
The Journal of Pipeline Engineering
ot
f
or
d
is
t
rib
ut
io
n
194
Sa
m
pl
e
co
py
-n
from images synced to specific loads that over the course
of the loading the specimen begins to wrap around
the tup, which occurs as the rotation increases just the
limit load. As this happens the load, as measured by
the tup, falls off in comparison with the load measured
by the load cell. In view of the inset in Fig.7a, this
decrease could reflect the observation that the face of
the tup is unloaded as the specimen rotation about the
tup increases. It is further evident from the inset that,
as the rotation increases, the load is increasingly being
transferred to the back face of the CVN specimen along
the outer extremities of the radiused tup face. Such
behaviour indicates that as the rotation increases the
CVN specimen is subjected to a wedge-opening load
that is not quantified in simple terms by the measured
load. Accordingly, while the response evident in Fig.7b
indicates that the measured load underestimates the
vertical component of the loading by about 7%, the
wedge-opening load is not quantified.
It follows that aspects such as tup wear could lead
to variability, particularly if this wear occurs local to
the shoulders of the ASTM E23 tup design. Likewise,
it follows that the use of a smaller-radius tup that
makes contact over a narrower zone, as occurs for the
ISO 148-1 tup design in comparison to that for E23,
would lead to a reduction in the wedge-opening load.
Finally, it follows that the observation of a threshold
toughness below which tup geometry has no apparent
Fig.8. Comparison of
subsize and full-size
CVN energy for one
X-70 steel.
effect (for example, Ref.31) reflects the observation
that less-tough steels do not survive in the test to
the point beyond which rotation and contact with
the flanks of the tup occur to a significant extent.
Thus, care must be taken when interpreting the energy
measured in a CVN test for higher-toughness steels,
as it can underestimate the actual force involved,
while at the same time fail to reflect the effects of
wedge-opening loading.
Subsize specimen issues
Results have been reported since the 1950s (such as
Ref.32) concerning the effects of subsize specimens,
with references cited in that work that date to the
late 1940s. As such, the effect is not new. While
the topic has been investigated for many decades,
the Appendices of ASTM E23 have not provided a
conclusive view on the role of thickness in the form of
a correlation, suggesting that some uncertainty remains
as to the circumstances that control the effect, and
whether there is a consistent pattern. Issues in the
context of reproducibility have been reported for the
CVN test longer than the effects of subsize specimens.
For example, work also reported in the 1950s (such
as Ref.33) that cite “poor steel, poor heat treatment,
or both” and note “variations….caused by poor testing
techniques, and the poor condition of the Charpy
machine, or both” as underlying causes for those issue,
195
or
d
is
t
rib
ut
io
n
3rd Quarter, 2013
(a)
(b)
ot
f
Fig.9. Force-displacement response for CVN specimens made of X-70 steel:
(a) full-size specimen; (b) 2/3-thick specimen.
separated into two pieces – which, for the samples
tested in the 516-ft-lb (700-J) Charpy machine (most of
them), was well below 80% of that machine’s capacity.
co
py
-n
all of which could contribute to the lack of clarity in
regard to thickness effects. Also possibly contributing
to this circumstance is the host of factors that have
more recently been considered in the context of
ASTM STP 1248 [34], which considered the specimen,
the anvil and striker, the test procedures, and still
other topics. Accordingly, this section adds only to
the knowledge base – particularly given it relies on
the ASTM E23 tup and considers subsize effects in
reference to tough steel.
Sa
m
pl
e
Figure 8 presents results for 1990s’ vintage X-70 steel
cut from large-diameter pipe that had a wall thickness
of 0.560 in (14.2 mm), comparing the energy to fail
for sub-thickness and full-size thickness CVN specimens.
In this figure, the x-axis represents the energy from
testing full-size specimens whereas the y-axis presents
the FSE energy for the 2/3-thick specimens, where
FSE energy is linearly scaled energy density per unit
area. While some of the specimens were tested in a
264-ft-lb (358-J) Charpy impact machine, the majority
were evaluated in a 516-ft-lb (700-J) Charpy machine.
For the higher-toughness cases, some of the specimens
did not separate into two pieces. In particular, for
cases where the FSE energy was about 200 ft-lb
(270 J), the 2/3-thick specimens did not separate into
two pieces, whereas those that were made full thickness
did. However, at FSE energy levels above 250 ft-lb (339
J), neither the full-thickness nor the 2/3-thick specimens
This same comparison after excising results where
the measured energy exceeded 80% of the machine’s
capacity (tests in the 264-ft-lb (358-J) machine) suggest
that for energies less than 180 ft-lb (244 J) linearly
scaling the 2/3-size Charpy data to an FSE value slightly
underestimates the result measured on full-thickness
specimens. In contrast, for energies greater than
180 ft-lb (244 J), linearly scaling the 2/3-thick specimen
data overestimates the result measured on full-thickness
specimens, which could be due to constraint developing
in the specimens at higher-toughness levels.
As the degree of constraint decreases, the extent of the
through-thickness thinning along the notch increases,
which adds to the fraction of the deformation component
of energy. Fig.9 illustrates this point, and makes clear
the earlier assertion that instrumented testing is essential
for understanding the outcomes of such testing.
Figure 9a presents the force-displacement response of
a full-size CVN specimen with toughness greater than
170 ft-lb (230 J), while Fig.9b presents the comparable
result for a 2/3-thick specimen made of the same
steel. When these results are plotted with the y-axis
used in Fig.3 (here the UTS is the same for both),
196
The Journal of Pipeline Engineering
io
n
is
t
rib
ut
This paper also considered the utility of standardized
impact testing such as ASTM E23 and ISO 148-1
in the context of applications involving tough steels,
with the potential utility of the DWTT practice also
considered. It was apparent in that discussion that the
response of bending-impact specimens is, as expected,
a consequence of the loading and the structural
geometry involved, and thus a poor analogue for the
circumstances that significantly differ. This aspect was
discussed in regard to pipelines and the ground-vehicle
industry, with the conclusion that bending geometries
are a poor analogue for the fracture-propagation
process that develops in pipelines, and also for some
other applications. Given that structural geometry
and the loadings, in conjunction with the material’s
properties, dictate the failure response, where critical
differences exist in that context, alternative testing
practices should be developed.
Sa
m
pl
e
co
py
-n
ot
f
Inlaid into the force-displacement for these figures
is a cross-hatched estimate of the energy dissipated
in propagation – which for the subsize specimen
excludes the component of energy due to deformation
induced by the decreased constraint. Each part of
Fig.9 also indicates the ratio of propagation to total
energy decreases based on that estimate. This ratio
indicates that as the specimen thickness is reduced, the
deformation component for the 2/3-thick specimen has
increased as compared to the full-size specimen. While
the scope of the effort for the client that supported
this work precluded more comprehensive analysis, it
was expected that if the energies are separated into
initiation, deformation, and propagation components as
just noted, a direct correlation would emerge between
the propagation energies for 2/3-thick and full-size
specimens. In that context, comparing the outcomes
for the results shown in Fig.9, the propagation energy
density for the standard thickness specimen is 717
ft-lb/in2 (150 J/cm2), while that for the 2/3-thick
specimen is 704 ft-lb/in2 (148 J/cm2). While not an
exact match, these results are well within material
scatter limits, and suggested further consideration of
this plausible correlation – which unfortunately was
beyond the scope of that effort.
materials. The origin of the CVN test and its development
into a standard was discussed, and the evolution of
mechanical and other properties motivated by industry
demands was evaluated. Interpretation of the CVN test
was discussed in regard to the force-displacement and
compliance response that develops during the test, and
factors affecting the energy measured and controlling
failure of the CVN specimen were assessed, including
the tup design and the use of subsize specimens.
or
d
the trends for both tests are comparable up through
the peak load and beyond the compliance change that
indicates �propagation’ has begun. However, during the
course of the propagation phase, the trends deviate.
It is apparent that for the full-size specimen the
propagation trend in Fig.9a continues in the same
manner evident for the data shown in Fig.3. As in
those trends, the energy dissipated in propagation
tracks along the slope shown until the change in
compliance indicates a transition to deformation has
occurred, which (as in Fig.3) is associated with final
rotation and eventual separation of the halves. In
contrast, the trend for the subsize specimen shown in
Fig.9b indicates that after some propagation occurs as
for the full-size specimen, the energy in what was the
propagation phase begins to increase. As suggested above,
this reflects the decreased constraint, which admits
greater through-thickness thinning along the notch,
and so increases the relative fraction of deformation
energy during what was otherwise steady propagation
in the thicker specimen. This deformation component
continues to increase through the transition to overall
deformation energy, flagged by the vertical trend, after
which rotation and separation occur as for the fullthickness specimen.
Summary and discussion
This paper has reviewed the Charpy V-notch impact test
and assessed its utility to characterize fracture resistance
in applications to modern tough materials in contrast
to those encountered prior to the availability of such
Conclusions
While a number of conclusions have been drawn as
the paper developed, two primary conclusions bear
repeating here:
• because the behaviour in an impact test can
be complex, data interpretation and assessment
of their practical implications is best based
on data developed using a well-instrumented
machine; and
• because structural geometry and the loadings
act in conjunction with the material’s properties
to control the failure response, where critical
differences exist in that context, alternative
testing practices are needed that are adapted
to the specific loading and failure response of
the structure of interest – which is now the
case for fracture propagation in pipelines where
tough materials are involved.
Acknowledgments
The data reported herein in Figs 7, 8, and 9 were
generated under contract to Alliance Pipeline, as part
of developing its fracture-control programme in the
late 1990s. Permission to release that data is gratefully
acknowledged, as is the author’s related collaboration
with David Rudland, then with Battelle.
3rd Quarter, 2013
197
Sa
m
pl
e
co
py
-n
io
n
rib
ut
is
t
ot
f
1. T.A.Siewert, M.P.Manahan, C.N.McCowan, J.M.Holt,
F.J.Marsh, and E.A.Ruth, 1999. The history and
importance of impact testing. In: Pendulum impact
testing: a century of progress. ASTM STP 1380,
American Society for Testing and Materials, pp 3-16.
2. A.E.White and C.L.Clark, 1925. Bibliography of
impact testing. Dept of Engineering Research,
University of Michigan.
3. A.LeChatalier, 1892. On the fragility after immersion
in a cold fluid. French Testing Commission, 3.
4. S.B. Russell, 1898. Experiments with a new machine
for testing materials by impact. Trans ASCE, 39, p
237, June.
5. M.G.Charpy, 1901. Note sur lГ©ssai des mГ©taux Г la flexion par choc de barreaux entaillГ©s. Soc. Ing.
Français, pp 848–877. Reprinted in ASTM STP
1380, American Society for Testing and Materials,
2000.
6. W.J.Hall, H.Kihara, W.Soete, and A.A.Wells, 1967.
Brittle fracture of welded plate. Prentice Hall.
7. ASTM E23 - 12c. Standard test methods for notched
bar impact testing of metallic materials. American
Society for Testing and Materials, 03.01.
8. BS EN ISO 148-1: 2009. Metallic materials - Charpy
pendulum impact test - Part 1: Test method; Part
2: Verification of testing machines; and Part 3:
Preparation and characterization of Charpy V-notch
test pieces for indirect verification of pendulum impact
machines, International Standards Organization.
9. EN 10045-1: 1990. Charpy impact test on metallic
materials. Test method (V- and U-notches) – now
withdrawn and superseded by BS EN ISO 148-1.
10.ASTM, 1938. Symposium on impact testing. Proc.
41st Annual Meeting of the American Society for
Testing and Materials, Atlantic City, 28 June.
11.ASTM, 1955. Symposium on impact testing. Proc.
58th Annual Meeting, Atlantic City, 27 June.
ASTM STP 176, American Society for Testing and
Materials, 1956.
12.ASTM E2248 – 13. Standard test method for impact
testing of miniaturized Charpy V-notch specimens.
American Society for Testing and Materials, 03.01.
13.ASTM D6110 – 10. Standard test method for
determining the Charpy impact resistance of notched
specimens of plastics. American Society for Testing
and Materials, 08.03.
14. ASTM E1253 – 13. Standard guide for reconstitution
of irradiated Charpy-sized specimens. American
Society for Testing and Materials, 12.02.
15.ASTM F2231 - 02(2008). Standard test method for
Charpy impact test on thin specimens of polyethylene
used in pressurized pipes. American Society for
Testing and Materials, 08.04.
16.ASTM A327 / A327M – 11. Standard test methods
for impact testing of cast irons. American Society
for Testing and Materials, 01.02.
17.ASTM E2298 – 13. Standard test method for
instrumented impact testing of metallic materials.
American Society for Testing and Materials, 03.01.
18.W.T.Matthews, 1969. The role of impact testing in
characterizing the toughness of materials. ASTM ST
466, American Society for Testing and Materials,
pp 3-20.
19. B.N.Leis and T.F.Forte, 2005. Managing the integrity
of early pipelines – crack growth analysis and
revalidation intervals. Battelle Final Report to the
Research and Special Projects Agency, Contract
DTRS56-03-T-0003, February.
20. W.J.McGregor Tegart, 1966. Elements of mechanical
metallurgy. Macmillan.
21.D.L.Rudland and B.N.Leis, 2007. Pipeline dynamic
fracture test program. Battelle’s Final Report to
R.J.Eiber, Consultant, and Alliance Pipeline, 7
February: filed in the Archives of the Canadian
National Energy Board (NEB) Hearing GH 3-97.
22.R.D.Galliher and B.N.Leis, 2003. Evaluation of
fracture resistance via CVN and DWTT practices
for recently produced X80 and X100 line pipe
steels. Battelle’s Final (Proprietary) Report to a
commercial client, 23 October.
23.NIST impact verification program: http://www.
nist.gov/mml/acmd/structural_materials/charpyverification-program.cfm
24.B.N.Leis, 1993. Instrumented CVN and tensile
testing of nine line-pipe steels in grades from X42
to X80. Battelle IR&D Report, April.
25.ASTM E8 / E8M – 11. Standard test methods
for tension testing of metallic materials. American
Society for Testing and Materials, 03.01.
26. B.N.Leis, 1977. Relationship between apparent (total)
Charpy Vee-notch toughness and the corresponding
dynamic crack-propagation resistance. Battelle’s Final
Report to R.J.Eiber, Consultant, and Alliance
Pipeline, 7 February: filed in the Archives of the
Canadian NEB Hearing GH 3-97.
27. B.N.Leis, R.J.Eiber, L.E.Carlson, and A. Gilroy-Scott,
1998. Relationship between apparent Charpy Veenotch toughness and the corresponding dynamic
crack-propagation resistance. International Pipeline
Conference, ASME, Calgary, pp 723-732.
28.B.N.Leis, 1999. Characterize fracture features for
samples from the first and second burst tests.
Battelle’s Final Report on Task 6, to R.J.Eiber,
Consultant, and Alliance Pipeline, 7 May: filed
in the Archives of the Canadian NEB Hearing
GH 3-97.
29.D.J.Lingenfelser, 1987. Application of basic material
tests for evaluating new engineering materials. SAE
Technical Paper 870801.
30.M.Ohata, M.Toyoda, N.Ishikawa, and T.Shinmiya,
2007. Significance of fracture toughness test results
of beam welds in evaluation of brittle fracture
performance of girth welded pipe joints. J. Pressure
V. Tech., 129, 4, pp 609-618.
or
d
References
198
The Journal of Pipeline Engineering
33.D.E.Driscoll, 1956. Reproducibility of Charpy
impact test. ASTM Special Technical Publication
176, American Society for Testing and Materials,
pp 70-74.
34.
T.A.Siewert and A.K.Schmieder, 1995. Eds:
Pendulum impact machines: procedures and
specimens for verification. ASTM STP 1248,
American Society for Testing and Materials.
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
rib
ut
io
n
31.E.A.Ruth, 1995. Striker geometry and its effect
on absorbed energy. ASTM STP 1248, American
Society for Testing and Materials, pp 101-110.
32.R.S.Zeno, 1956. Effects of specimen width on
the notched bar impact properties of quenched
and tempered and normalized steels. ASTM STP
176, American Society for Testing and Materials,
pp 59-69.
3rd Quarter, 2013
199
CTOA testing of pipeline steels
using MDCB specimens1
io
n
by Dr Robert L Amaro, Dr Jeffrey W Sowards, Elizabeth S Drexler,
J David McColskey, and Christopher McCowan*
rib
ut
NIST, Applied Chemical and Materials Division, Boulder, CO, USA
T
ot
f
or
d
is
t
HE CRACK-TIP-OPENING angle (CTOA) is used to rank the relative resistance to crack extension
of various pipeline steels. In general, the smaller the CTOA value, the lower the resistance to crack
extension. It is unclear, however, whether CTOA is a material property that is valid for all thicknesses and
rates of crack growth. Historically, drop-weight tear tests (DWTT) and modified double-cantilever beam
(MDCB) specimens have been used for measuring CTOA. Tests using either specimen may be conducted
at quasi-static and dynamic rates. The fastest displacement rates achieved in our laboratory were near
14 m/s, resulting in crack extension rates near 30 m/s for high-toughness linepipe steels. In-service crack
extensions for ductile-crack fracture can be more than 100 m/s. The failure mode at this rate is plastic
collapse, and it is uncertain if correlations can be drawn between in-service failures and laboratory tests
conducted on thinner material tested at slower rates.We describe the evolution of our test method using
MDCB specimens from 2006 to 2012 and the direction we anticipate for future CTOA research.
T
worked on developing new measures of fracture control.
Among these, crack-tip-opening angle (CTOA) is one
alternative for characterizing fully plastic fracture [6-7],
especially for running ductile cracks in pipes [2,7-13].
In cases where the fracture process is characterized by a
large degree of stable tearing, CTOA has been recognized
as a measure of the resistance of a material to fracture
[6,9]. The main advantages of CTOA are that it can
be directly measured from the crack-opening profile and
can be related to the geometry of the fracturing pipe.
However, there are difficulties in determining CTOA
with a simple measurement technique that would be
widely available to many material-test laboratories. In
addition, the CTOA criterion can be implemented in
finite-element models of the propagating-fracture process
[6,9,13,14].
co
py
-n
HE INCREASING DEMAND for natural gas as an
alternative energy source implies continued growth
of gas pipeline installations and the qualification of
materials in the actual pipeline network. A difficult
problem to be solved for the economic and safe
operation of high-pressure gas pipelines is the control
of ductile-fracture propagation [1]. As a result, the
accurate prediction of the resistance to fracture and
ductile-fracture arrest in pressurized gas pipelines are
currently important issues.
Sa
m
pl
e
Initially, the measure of a material’s fracture resistance was
determined on the basis of Charpy V-notch (CVN) shelf
energy, such as that used in the Battelle two-curve model
[1]. Later correlations were developed between Charpy
and dynamic drop-weight tear test (DWTT) data. The
Battelle two-curve model worked well for many years, but
when applied to modern higher-strength pipeline steels,
significant errors are apparent [2–5]. Correction factors have
been developed [1,4,5] for high-strength steels; however,
use of these correction factors adds further uncertainty
to the estimates. Thus, in parallel with the CVN- and
DWTT-based fracture strategies, pipeline designers have
1. Contribution of NIST, an agency of the US government: not subject to
copyright in the United States.
*Corresponding author’s contact details:
tel: +1 303 497 3699
email: [email protected]
The literature contains a number of different specimen
geometries for studying ductile-fracture propagation with
the CTOA criterion, such as middle-tension specimens,
M(T) [6,15,16], compact-tension specimens, C(T) [6,15,16],
DWTT specimens (with methodologies based on one
specimen [1,8,9,10] or two specimens [17]), three-point
bend specimens, 3-PB [7,12], and modified doublecantilever beam specimens, MDCB [3,13]. Our efforts
have focused on test methods using the MDCB specimen
[3,13,18-20] that is promising for CTOA measurement
in pipeline steel, because this specimen design allows
an extended region for steady-state crack growth and
for larger plastic deformation at the crack tip. This may
200
The Journal of Pipeline Engineering
API Designation
SMYS MPa (ksi)
O.D.
mm(inch)
Thickness mm
1
N/A (~X70)
517 (75)
0.51 (20)
9.7
2
X52
359 (52)
0.51 (20)
8.0
3
Grade B
244 (35)
0.56 (22)
7.4
4
N/A (~X52)
335 (48)
0.51 (20)
7.9
5
N/A
281 (40)
0.56 (22)
7.8
6
X65
448 (65)
0.61 (24)
7
X65
448 (65)
0.51 (20)
8
X65
448 (65)
0.76 (30)
17.0
9
X100
689 (100)
1.32 (52)
20.6
10
X100
689 (100)
1.22 (48)
20.0
11
X100
689 (100)
1.22 (48)
20.0
12
X70 spiral
483 (70)
0.91 (36)
13.7
is
t
rib
ut
25.0
• Steel #2 is an API X-52 characterized by a
ferrite-pearlite structure, with a significantly larger
ferrite grain size than steel #1. This steel has
the most pronounced banding (of pearlite) of
the steels evaluated here.
• Steel #3 is an API Grade B ferrite-pearlite steel
without banding.
• Steel #4 is a ferrite-pearlite steel with low
banding.
• Steel #5 is a ferrite-pearlite steel without banding.
• Steel #6 is an API X-65 grade of ferrite-pearlite
steel, which might be better described as ferritecarbide, because there is very little pearlite in
the microstructure. The grain size of this steel
was not measured, but the ferrite grain size is
similar in size to that of steel #1.
• Steel #7 is an API X-65 grade with no pearlite
and a fine non-equiaxial microstructure.
• Steel #8 is an API X-65 grade with a ferritepearlite microstructure and heavy bands of
pearlite.
• Steel #9 is an API X-100 grade: this is an
experimental alloy that was used for full-scale
testing.
• Steels #10 and #11 are two modern API X-100
bainitic steels.
• Steel #12 is an API X-70 spiral pipe steel, the
microstructure of which is not known.
-n
ot
f
simulate the conditions surrounding running cracks on
pipelines as they exhibit plastic regions on the order
of 2.5 pipe diameters ahead of the crack tip and
0.3 diameters on each side of the crack line [10].
Moreover, the MDCB specimen can be cut directly
from pipe with no subsequent flattening required, which
avoids potential load-history effects due to pre-straining
the material.
31.5
or
d
Table 1. Information on pipeline steels tested.
io
n
ID Number
co
py
NIST has a history of conducting CTOA tests with the
MDCB specimen. The gripping mechanism has evolved
to increase constraint, methods to mark grids on the
specimens have improved, and loading and recording systems
were developed to conduct tests at high rates [21-31].
Sa
m
pl
e
In this summary of our CTOA testing, data from 12
pipeline steels are presented. They are described here
and referenced by ID number in the sections that follow.
Table 1 summarizes the dimensions of the pipes from
which all samples were extracted. The specified minimum
yield strength (SMYS) and the API designations are
also provided in the table. In Table 2, the chemical
compositions of the steels are given, while in Table 3,
the grain size and pearlite volume fraction are given for
the ferrite/pearlite steels (#1 – #5).
Microstructures
The microstructures of the 12 pipeline steels tested
are briefly described as follows:
• Steel #1 is a ferrite-pearlite steel with low carbon
(low pearlite) content and a fine ferrite grain
size. This steel represents a modern, fine-grained
ferrite pipeline steel.
Tensile properties
The tensile properties of steels #1 to #5 were measured
with flat tensile specimens (due to plate thickness),
while round tensile specimens (6-mm diameter) were
tested for steels #6 to #11. The flat specimens were
3rd Quarter, 2013
201
#1
#2
#3
#4
#5
#6
Al
0.031
B
<0.0002
0.06
0.24
0.27
0.18
0.25
0.07
Co
0.006
0.025
0.007
0.014
0.025
0.003
Cr
0.02
0.024
0.029
0.021
0.019
0.12
Cu
0.11
0.038
0.015
0.054
0.046
Mn
1.46
1.03
0.36
0.52
0.97
Mo
0.025
0.016
0.007
0.009
0.017
Nb
0.054
0.007
0.005
0.005
Ni
0.10
0.064
0.021
0.021
P
0.01
0.016
0.005
0.026
0.013
0.008
S
<0.01
0.013
0.015
0.010
0.012
0.004
Si
0.28
0.057
0.009
0.043
0.061
0.094
or
d
#7
#8
Al
0.030
0.039
B
<0.0002
0.0002
C
0.07
0.08
#9
Co
0.002
Cr
0.13
Cu
py
0.066
0.17
0.03
0.04
#10
#11
#12
0.025
0.012
0.039
<0.0001
0.0003
0.064
0.04
0.084
0.001
0.003
0.03
0.021
0.023
0.07
0.09
0.30
0.286
0.28
0.31
1.59
1.56
1.90
2.092
1.87
1.56
0.006
0.15
0.127
0.23
0.20
0.005
0.003
0.008
0.041
0.017
0.069
0.003
pl
Mo
0.04
0.10
e
Mn
0.07
0.003
0.002
ot
f
0.002
-n
0.045
co
V
1.48
rib
ut
0.003
Ti
0.12
0.007
is
t
N
io
n
C
0.03
0.04
Ni
0.14
0.21
0.50
0.501
0.47
0.11
P
0.009
0.011
0.008
0.10
0.009
0.010
S
0.004
0.003
0.0005
0.002
<0.001
0.009
Si
0.092
0.325
0.10
0.108
0.099
0.24
Ti
0.02
<0.01
0.007
0.17
0.013
V
0.04
0.04
0.006
0.002
0.003
m
Nb
Sa
N
Table 2. Chemical composition of the pipeline steels tested, by mass. Column numbers give identification number for the
steel, as defined in Table 1.
202
The Journal of Pipeline Engineering
Steel #
Ferritic grain size (Ојm)
Pearlite volume fraction (%)
1
2
3
4
5
6.5
11.8
10.8
N/A
22.2
5
37.1
25.3
37.9
17.1
Orientation
E (GPa)
Пѓ 0.2
(MPa)
Пѓ UTS
(MPa)
Пѓ0.2 / ПѓUTS
eu (%)
ef (%)
1
L
211*
517
611
0.846
6.7%
35.0%
0.19
T
N/A
543
606
0.896
8.0%
27.4%
0.29
L
211*
360
556
0.647
12.3%
32.7%
0.38
T
N/A
448
576
0.777
11.1%
25.6%
0.43
L
212*
244
451
0.541
19.6%
37.8%
0.52
T
N/A
255
459
0.555
18.8%
38.0%
0.49
L
210*
335
535
0.626
12.9%
34.9%
0.37
T
N/A
428
560
0.764
10.5%
22.0%
0.48
L
214
265
454
0.583
16.0%
38.0%
0.42
T
NA
248
453
0.547
19.5%
35.0%
0.56
L
201
460
534
0.870
8.2%
24.7%
0.33
T
218
497
560
0.890
7.7%
15.9%
0.48
L
NA
502
570
0.880
6.8%
25.7%
0.26
T
N/A
511
577
0.885
7.2%
20.9%
0.34
L
217
522
618
0.844
10.1%
27.3%
0.37
T
N/A
576
644
0.894
6.9%
24.8%
0.28
L
N/A
694
801
0.910
4.6%
20.3%
0.23
N/A
797
828
0.966
4.1%
19.3%
0.21
192
722
855
0.844
4.6%
17.8%
0.26
T
213
912
916
0.995
2.6%
18.0%
0.14
L
198
729
838
0.869
5.8%
20.5%
0.28
T
207
833
868
0.989
4.7%
17.5%
0.27
T
NA
576
650
0.940
NA
NA
NA
5
6
7
8
9
T
L
co
10
m
pl
12
e
11
rib
ut
is
t
or
d
4
ot
f
3
py
2
eu/ef
io
n
Steel #
-n
Table 3. Measurements of the grain size and ferrite fraction for the ferrite/pearlite steels.
*Average determined from dynamic elastic modulus test
Sa
Table 4.Tensile properties of the materials. (Note: * = average determined from dynamic-elastic-modulus test.)
6 mm wide. Full-thickness specimens (Table 1) were
tested for the longitudinal orientation, and typically
3-mm thick specimens were tested for the transverse
orientation. All specimens had a gauge length of 25.4
mm. Experiments were performed either in a screwdriven tensile testing machine of 100-kN capacity, or a
closed-loop servo-hydraulic machine of 100-kN capacity.
Tests were conducted in displacement control at rates
of 0.25 mm/min for the flat specimens and 0.1 mm/
min for the round specimens.
The measured mechanical properties of the steels are
given in Table 4, where E is the Young’s modulus,
Пѓ0.2 the yield stress, ПѓUTS the ultimate strength, eu the
uniform elongation, and ef the fracture elongation.
In addition to the standard properties, the ratios of
Пѓ0.2/ПѓUTS (stress ratio) and eu/ef (strain ratio) are also
given in Table 4. These two parameters indicate the
strain-hardening potential of the steel. As shown in Fig.1,
the stress ratio increases as the strain ratio decreases.
3rd Quarter, 2013
203
CTOA test matrix
rib
ut
io
n
CTOA tests were conducted on X-52, X-65, X-70, and
X-100 pipeline steels and other pipeline grades not
identified with an API designation (Table 1). The tests
were conducted by tensile loading MDCB specimens
at actuator rates ranging from 0.002 to 14,000 mm/s.
The 8,000 and 14,000-mm/s displacement rates were
attained with a disc spring set-up [22]. Early tests were
quasi-static, and later a series of tests were conducted
on X-65 (#6) and X-100 (#9) with changing actuator
rates. High-rate tests were also conducted on two
additional X-100 steels (#10 and #11), and also on an
X-70 steel (#12).
CTOA test specimen
py
is
t
-n
ot
f
• It can be cut directly from a pipe, without
flattening;
• The width and thickness are limited by pipe
curvature and wall thickness;
• The long ligament in the gauge section allows
for the CTOA to be measured multiple times
and averaged;
• High constraint in the test section is promoted
by two thicker loading arms;
• The test section does not restrain the transition
to slant mode shear fracture;
• The test section is flat near the crack tip for
ease of CTOA measurement.
Fig.1. Strain ratio versus stress ratio.
or
d
A modified double-cantilever beam (MDCB) specimen
(Fig.2) was used to conduct the CTOA test. The
specimen exhibits the following characteristics:
e
co
The test specimens were cut with the notch direction
along the axis of the pipe. The thickness of the curved
plate was reduced by machining to obtain a flat plate,
which eliminated the probable residual plastic strains
that would be caused by flatting the plate by use of
a straightening procedure.
Sa
m
pl
The specimens were fatigue pre-cracked following the
ASTM standard procedure for conducting crack-tipopening displacement (CTOD) tests [32]. The precracking loads were selected to ensure that the ratio
of stress intensity factor range to the Young’s modulus
(∆K/E) remained below 0.005 mm-2. All specimens
were fatigue pre-cracked at a ratio of R = 0.1 [13], to
a crack-to-width ratio of a0/W = 0.3 to 0.5 [with a
specimen width, W, equal to 182 mm, and a0 equal
to the machined notch length (60 mm) plus the initial
fatigue pre-crack length (approx. 10 mm)].
Methods and procedures
Two apparatuses were used for CTOA testing, a �quasistatic’ set-up and a dynamic set-up.
Fig.2. CTOA specimen, with dimensions in millimetres.
Quasi-static apparatus
For quasi-static testing (0.002 to 3 mm/s), a 250-kN
uniaxial servo-hydraulic test machine was used. Tests
were conducted in displacement control. As shown
in Fig.3a, the load line ran through the centreline
of the first pair of holes in the specimen. A digital
camera and frame-capture software/hardware were used
to capture images. The camera was mounted on an
XYZ stage, which provided a stable platform to follow
the crack tip. The image acquisition was controlled by
a personal computer with image-analysis software: the
captured images had a size of 2048 pixels Г— 1536 pixels,
which resulted in a resolution of about 32 pixel/mm.
Images were acquired and stored, along with time, load,
and displacement data as the crack propagated across
the specimen. Tests were stopped at 80 mm of crack
extension beyond the machined notch tip. Details of
the set-up have been reported previously [21].
Dynamic apparatus
Tests with actuator rates of 3, 30, and 300 mm/s were
performed on a 500-kN uniaxial servo-hydraulic test
machine shown in Fig.3b [22]. As with the quasi-static
204
The Journal of Pipeline Engineering
Data processing
rib
ut
io
n
tests, the load line was located at the centreline of the
first pair of holes in the specimen. The machine was
adapted with large-capacity servo-valves to accomplish
rapid loading. For actuator rates greater than
300 mm/s, the 500-kN uniaxial test frame was
configured with a set of disc springs; the potential
energies stored in these springs for the X-65 and
X-100 tests were 5.6 kJ and 7.5 kJ, respectively [22].
Higher crack velocities were obtained by further
increasing the stored energy with the use of sacrificial
links1, which were made of aluminium alloy 7075-T6.
In this configuration, grip-displacement rates up to
14 m/s were attained.
or
d
is
t
Once images were captured, the CTOAs were measured
using data within the distance from the crack-tip
ranges prescribed by the ISO draft standard [35] and
the ASTM standard [36]. Within these ranges on
the samples, we used the following four approaches
to measure the CTOA [23]:
-n
ot
f
• Method 1 used an algorithm that located the
crack tip in the data, and then selected pairs
of points along the crack profile at prescribed
distances from the tip to calculate CTOA. The
crack tip was always included in this calculation
of the CTOA.
• Method 2 used data-point pairs that were within
the range 0.1 mm to 0.2 mm behind the crack
tip to fit lines within this region (Fig.5a). This
method never included the crack tip.
• Method 3 used data points marking the upper
and lower grid lines to fit lines for CTOA
calculation. Each line was fitted with 2 to 10
points, located within the increment 0.5 to
1.5 mm from the crack tip (Fig.4b).
• Method 4 used all of the profile data in the
interval 0.5 to 1.5 mm to define the two bestfit lines associated with the upper and lower
crack-tip profiles to calculate the CTOA. In
this case, typically 100 to 200 data points were
used for each line fit (Fig.4b).
Highspeed
camera 2
Specimen
Sa
m
pl
e
co
py
Discspring
setup
Highspeed
camera 1
Fig.3. Set-ups for (a - top) �static’ set-up with camera on a
motion-control XYZ stage, and (b - bottom) dynamic set-up
with high-speed cameras and springs.
The software required the operator to trace the profile
of the crack tip, and mark data points along the
closest set of upper and lower grid lines, as shown
in Fig.4. The CTOA values for each method were
then calculated from the collected images from each
specimen, and an average CTOA for each method
was determined.
1 Sacrificial links were inserted into the load line and loaded to compress
the springs. The links were calculated to fail at the load required to full
compress the springs, which resulted in abruptly releasing the stored energy
onto the CTOA specimen.
205
io
n
3rd Quarter, 2013
or
d
is
t
rib
ut
Fig.4. Showing (a - left) crack edge traced by the operator and points marked on the gridlines adjacent to the crack, and
(b - right) two sets of lines fitted with grid points and crack trace respectively (grid = 1 mm x 0.5 mm).
ot
f
(a)
(b)
Results and discussion
py
Steady-state region
-n
Fig.5. (a) Method 1 and (b) Method 2 for determining the CTOA. For both Methods, n was set equal to
3, L1 = r1 + r0 = 0.5 mm, L2 = r2 + r0 = 1 mm, and L3 = r3 + r0 = 1.5 mm (r0 was set to 0.15 mm).
co
CTOA values reached a constant at crack lengths ranging
from about 3 to 5.2 mm, which is 1 to 1.8 times the
specimen thickness. This result is consistent with those
observed by Mannucci et al. [9], Shterenlikht et al. [18],
and Hashemi et al. [3,19,20].
pl
e
Comparison of CTOA algorithms
Sa
m
For methods that measure CTOA directly on the fracture
surface (Methods 1, 2, and 4) the scatter in the CTOA
decreased with an increased measurement basis (Fig.5).
For example, increase in the measurement basis r reduced
the standard deviation of the CTOA data from 2.11o
to 0.97o for steel #1, and from 6.57o to 2.30o for steel
#3. For method 1, increase in L decreased the standard
deviation of CTOA measurement from 1.52o to 0.90o
for steel #1, and from 1.35o to 0.76o for steel #3. This
result is attributed to factors such as: the difficulty of
identifying the exact location of crack edges and the
crack tip; the local deformations in regions adjacent to
the apparent crack tip; and the effect associated with the
longer line segments. Typically, the crack edges appear
to be irregular, which is a natural result of the ductile-
fracture mechanism, and when all the profile data (0.5
to 1.5 mm interval) are used to fit lines and calculate
CTOA for Method 4, the result compares well with the
average CTOA calculated for Method 2, as expected.
Method 1 had the highest scatter in CTOA. Of the five
ferrite-pearlite steels tested, only steel #1 had a standard
deviation less than 1В°; for the other steels, the standard
deviations were between 2.2В° and 4.1В°. Method 1 also
resulted in the highest average CTOA value for the
steels. For steel #2, for example, the average CTOA value
by Method 1 was 5.8В° higher than that by Method 2.
Method 1 depends on accurately locating the apparent
crack tip, and is the most sensitive method for local
deformations as well as operator judgment.
Method 2 had standard deviations in CTOA measurements
between 0.7В° and 0.81В°, and the CTOA values consistently
agreed with Method 3 more than with Method 1. Method
3 had the smallest standard deviation in CTOA values
(0.46В° and 0.64В°). Method 4 tends to track well with
Method 2 results.
More discussions of selecting a proper measurement
basis L or r were given by Heerens et al. [36]. It can
be expected that an increase of L or r may give rise to
size and geometric effects. In our analysis, L or r values
206
The Journal of Pipeline Engineering
Stable CTOA (o)*
Standard deviation
(o)
Cross head mm/s
Crack velocity
(mm/s)
1
11.7
2.04
0.05
0.22
2
9.1
1.71
0.05
0.26
3
9.8
1.39
0.05
0.20
4
10.0
2.00
0.05
0.28
5
9.51
0.05
6
11.4
0.02
7
9.9
NA
8
NA
0.002
9
8.6
1.42
NA
10
7.8
1.9
0.02
11
8.2
2.3
12
11.9
1.3
6-X65
11.7
6-X65
0.22
is
t
rib
ut
NA
NA
0.02
NA
1.2
0.002
0.004
11.4
1.2
0.02
0.044
6-X65
10.5
1.0
0.2
0.5
6-X65
11.6
2.2
5
9.2В±0.6
6-X65
11.0
2.4
30
45.5В±1.5
6-X65
11.2
1.1
300
594В±8
6-X65
11.3
1.7
8000
6500В±600
9-X100
8.6
1.1
0.002
0.008
8.3
1.8
0.02
0.088
9.3
1.1
0.2
0.66
9.4
1.0
3
6.7В±0.7
ot
f
-n
py
9-X100
co
9-X100
or
d
0.02
9-X100
8.1
1.0
300
762В±35
9-X100
8.8
1.6
30
118В±3
9-X100
8.6
1.1
8000
7250В±605
9-X100
9.8
7500
5500
10-X100
7.3
2.3
10000
13000
10-X100
10.6
5.3
20000
29000
11-X100
8.1
11-X100
8.9
2.5
8000
7000
11-X100
9.3
3.2
20000
20000
12-X70
10.2
1.8
NA
5467
m
pl
e
9-X100
Sa
io
n
Steel #
2900
Table 5.The CTOA values (Method 4) calculated for the steels at various testing rates.
207
is
t
rib
ut
io
n
3rd Quarter, 2013
co
py
-n
ot
f
or
d
Fig.6. Ranking of steels by
CTOA results.
e
Fig.7. CTOA versus crack
velocity for steel #6 (X-65)
and steel #9 (X-100).
m
pl
within the range from 0.5 to 1.5 mm were chosen to
calculate CTOA data. This is a reasonable compromise
between the demands for minimizing scatter and possible
size effects in calculating CTOA for pipeline steels.
Sa
CTOA ranking of the pipeline steels tested
The results for the 12 steels are summarized in Table
5 and plotted in Fig.6, where all CTOA values were
determined with Method 4 and averaged multiple
specimen results. In Fig.6 the baintic steels (#9, #10,
#11) have the lowest CTOA values and the highest
strengths. For the ferrite-pearlite steels, we see that
increasing strength by increasing the volume fraction
of pearlite did not result in lowering the CTOA (#3
and #5 compared with #2 and #4). Not unexpectedly,
reducing the grain size for strengthening (#1) resulted
in significant increase in CTOA, as compared with
steels with larger grain size but higher ferrite content
at similar strength levels (#1 and #6 compared with
#2 and #4). These data raise the question whether
fine-grained ferritic steels with higher pearlite contents
(or another strengthening addition) can provide an
improvement in the strength to CTOA ratio.
Influence of loading rate on CTOA
Actuator displacement rates covering nearly seven orders
of magnitude – from 0.002 mm/s to approximate 8,000
mm/s – are shown in Fig.7 for the X-65 (#6) and
X-100 (#9) steels [22 and 24]. The average crack-growth
velocities for the test matrix are given in Table 5.
The Journal of Pipeline Engineering
Fig.8. Crack velocity
vs displacement rate.
m
pl
e
co
py
-n
ot
f
or
d
is
t
rib
ut
io
n
208
Sa
As shown in Fig.7, the CTOA results indicate that the
X-65 steel (#6) consistently had a higher resistance to
cracking than the X-100 steel (#9). The CTOA for the
X-65 is typically more than 2В° higher than the CTOA
for the X-100 throughout the range of rates evaluated.
Neither steel showed a trend for CTOA with actuator
or crack velocity. Additional data for X-100 steels #10
and #11 provide data at rates up to 29 m/s (Table 5),
and the CTOA values range from 7.3В° to 8.2В° with
no clear trend with velocity. The X-70 steel at higher
velocity (#12) had a CTOA of 10.2, and an increased
crack velocity did not result in decreased CTOA.
Fig.9. Idealized slant and flat
fracture-mode morphologies
are illustrated on the
left, and traces of actual
fractures are shown on
the right: (A) full slant; (B)
double slant; and (C) flat.
Given that the measured CTOA values are independent
of the crack-tip velocities, material-specific correlations
were sought between test variables (load, displacement,
displacement rate) and the measured crack-tip velocity.
Not surprisingly, as shown in Fig8, the displacement
rate correlates extremely well with the measured cracktip velocity. However, for the same actuator rates, the
X-65 specimens typically had a slightly lower crack
velocity than the X-100 specimens, which indicates that
the X-65 exhibits higher resistance to crack growth
than does the X-100. It is surprising, however, that
the correlation was found to be so valid for all
209
rib
ut
io
n
3rd Quarter, 2013
fracture, the characteristic features of the CTOA specimens
have much in common with the morphology-associated
uniaxial tensile failures of ductile steels; however, in
the case of the CTOA specimens, the morphology is
cup-cup rather than cup-cone. The shear-oriented regions
associated with the cup-cup fractures are formed by
plastic flow that results in fractures with a knife-edge
morphology. An important point associated with this
observation is that the CTOA angle measured on the
outside surface of a specimen in a flat-fracture mode is
the angle formed between the two knife-edges as final
fracture occurs. This is not an angle formed by the
interior fracture planes; it is more like a plastic hinge.
This is not the case for the slant-fracture mode, for
which the fracture planes intersect the outside surface
of the specimen.
ot
f
materials tested. One may infer from Fig.8 that
the crack-tip velocity is primarily a function of the
far-field loading rate, for the materials and loading
rates tested here.
or
d
is
t
Fig.10. Cross sections of CTOD
specimens of X-100 (top)
and X-65 (bottom). From
left to right the crosshead
displacement rates applied to
these specimens range between
0.002 mm/s and 8,000 mm/s.
The thickness at the bottom
edge of the cross sections is
approximately 8 mm.
-n
Macroscopic fracture modes
e
co
py
The macroscopic failure mode for pipes and CTOA
specimens is often described as either a flat or a slant
fracture mode (Fig.9). However, mixed-mode (flat and
slant) fracture morphologies are observed for both field
fractures and laboratory fractures. The range of fracture
modes observed in our studies for 8-mm thick MDCB
specimens is shown Fig.10: the basic slant fracture
occurs on a single macroscopic shear plane through
the thickness of the sample, but double-slant fractures
and mixed-mode fractures are not uncommon.
Sa
m
pl
Flat and mixed modes were the typical fracture modes
for CTOA specimens tested at crosshead displacement
rates of 300 mm/s and less (Fig.10), while at rates near
8,000 mm/s, slant fractures were typically observed.
This is in agreement with fracture modes observed for
full-scale, high-rate tests of the X-100. However, the
fracture surface features can differ significantly, so it
is useful to look a bit closer to determine whether
the fracture mechanisms for laboratory fractures are
representative of field fractures.
Both slant and flat fracture modes have significant areas
of their fracture surfaces on angles near 45В° to the
applied loading. The fracture-surface features on these
two types of shear-oriented surface, however, indicate
that they are formed by different mechanisms. For flat
Details of the fracture modes
Considering details for the fracture modes (Figs 11
and 12), flat fracture initiates in the centre of the
specimen thickness on a plane perpendicular to the
applied tensile force and grows to form an internal
void. As this void grows, it effectively divides the
specimen thickness into two thinner thicknesses, with
lower constraint. These two thinner plates deform on
shear planes until they thin down and fail in a knifeedge morphology. The ductile dimples on these thinned
shear planes are characteristically elongated along the
primary loading direction. But, unlike the case for cupcone shear rupture, ductile dimples on the knife-edge
cup-cup fracture surfaces do not have �mating’ dimples
on the opposing fracture surface characteristic of shear
failure and void sheet coalescence.
The Journal of Pipeline Engineering
(a)
or
d
is
t
rib
ut
(b)
io
n
210
(d)
co
py
-n
ot
f
(c)
(e)
(f)
Sa
m
pl
e
Fig.11. Fracture-surface features associated with flat fracture.The overview (a) shows the �flat’ central portion of an X-65
fracture, bounded at both surfaces by shear regions.The central region (b and c) is a mixture of large ductile dimples, elongated
in the direction of crack growth (and plate rolling), surrounded by smaller equiaxial dimples.The knife-edge final-fracture region
has a shear-fracture region for a distance of about 100 Ојm into the specimen (d) on which shear dimples are apparent (e).
There is a gradient in texture on the surface of the final fracture, with a smoother shear dimple surface near the outside edge of
the specimen (e), and a more textured equiaxial dimple surface toward the centre of the specimen (f).
Due to the extensive plastic flow associated with
formation of the knife-edge regions for the flat-fracture
morphology, the surface roughness of these regions
is smooth. As shown in Fig.11d, the surface texture
near the outside surface is smooth, and this region
becomes smoother and extends into the specimen
further with increased testing rate. This trend is
evident for both the X-65 and the X-100 steels tested,
and is noticeable with the unaided eye (due to the
increased light reflection for the smoother surfaces). So,
flat-fracture morphologies like these like these might
be interpreted to some extent to determine whether
fracture occurred dynamically or not, and may give
some guidance on the relative rate of fracture. These
results also point out that a model of ductile fracture
in pipeline steels should take into consideration the
true failure mode of the steel.
A full-slant fracture mode results in the fracture surface
on a single plane, tilted at an angle of 45В° to the
primary stress on the CTOA specimen. Details of the
fracture features on these slant planes differ from those
211
(b) Centre
or
d
is
t
rib
ut
(a) Edge
io
n
3rd Quarter, 2013
(d) Centre
co
py
-n
ot
f
(c) Centre
(e) Centre
(f) Edge
m
pl
e
Fig.12. Details of a slant-fracture mode from an X-100 CTOA specimen showing ductile dimple morphologies: (a) region very
near the outside surface of specimen, (b, c, d, e) regions through the thickness, not near the final fracture regions, (f) higher
magnification of a region very close to the final fracture of the knife-edge showing shear dimples.
Sa
for both cup-cup planes, formed with the flat-fracture
morphology, and shear planes formed by cup-cone
failure modes (Fig.12). The ductile dimple morphology
over most of the slant surface is indicative of the
ductile rupture typically observed on the interior �flat’
portion of the fracture that is normal to the applied
load, rather than the rupture on the shear lips of the
tensile specimen. Elongated shear dimples are found
only very near the outside edge of the shear planes
on CTOA specimens. Across most of the slant failure,
dimples are typically equiaxial and have full rims. If
they are elongated, the elongation is in the direction
of crack growth, as is the case for the central region
of flat failure modes. This indicates that mode-I loading
is the primary influence. In general, evidence of shear
dimple failure (mode III) is limited to regions very
near the outside surface of the specimen.
Crack-front shape
The �crack tip’ measured at the outside surface of
the specimen in the CTOA test is not the tip of the
crack in the interior of the specimen. For example,
in Fig.13 the tip of the crack front is about 1.5 mm
ahead of the intersection of the crack front with the
surface of the specimen.
212
The Journal of Pipeline Engineering
The extent of crack-tip tunnelling varied on test specimens
from about 1.5 mm to 8 mm, and the shape of the crack
fronts varied from a gentle curve to an arrowhead-like
shape. Occasionally crack fronts with irregular features
were observed, and crack-front markings were not clear
on all fracture surfaces.
is
t
Intermittent crack growth
Intermittent crack growth was sometimes observed for
the X-100 specimens tested at quasi-static rates. Dark and
light bands on the fracture surface mark this behaviour,
as shown in Fig.14. In general, the leading edge of
the crack front is coincident with the centreline of the
plate (which is not always in the middle of the CTOA
specimen), and intermittent crack growth does not always
occur on both sides of the centreline.
py
-n
ot
f
or
d
Fig.13. Crack-front trace as observed by markings on the
fracture surface of a CTOA specimen.The regions of the
specimen are (1) the outside surface, (2) the �flat’ portion of
the fracture, (3) the �knife-edged’ portion of the surface, and
(4) the final fracture region.
rib
ut
io
n
Castings of CTOA specimens were made under loading
to evaluate the 3D shapes of voids formed by tunnelling
cracks in CTOA test specimens1. In the case of �flat’
fractures, the casting shapes indicate that the �flat’ fracture
surfaces in the centre of the specimen thickness formed
an interior CTOA of 9.2o (X-100), which compares
reasonably well with the CTOA measured on the surface
of the X-100 specimens. Castings also showed that the
final fracture planes (knife-edge planes) on the X-100
samples were typically 45В° to 50В°.
Sa
m
pl
e
co
Fig.14. Intermittent bands of fast and slow ductile fracture.
Details on the fracture surface in the banded regions
show regions that resemble stair steps of quasi-ductile
fast fracture followed by ductile re-blunting regions [27].
The appearance of the quasi-ductile regions is similar to
the details on the surfaces of secondary cracking (splits)
in the burst test fractures for this X-100 steel. Since
the �riser’ sections of the stair step would have the
same orientation as secondary cracking, this is not too
surprising. In the �tread’ orientation, ductile dimpling is
apparent and indicates more ductility for this orientation.
These banded regions are observed in the higher-strength
materials during CTOA testing for X-80 [37] and X-100
[27, 37]. The lower-strength grades (X-65 and X-70) did
not exhibit banding. Both reports suggested that the
higher-strength grades may exhibit the contrasting bands
due to alternating regions of quasi-cleavage and ductile
fracture. Quasi-cleavage regions are likely associated with
brittle microstructural constituents associated with rolling
and segregation during production of higher-strength pipes.
Comparison to full-scale, high-rate test
The example shown (Fig.15) for the X-100 steel that
failed in a full-scale burst test has two slant-fracture
regions separated by a region of flat fracture. This is
Fig.15. Cross section of an X-100 steel (#9) fractured in
a full-scale test. Profiles of the fracture vary with position
along the length of the fractured pipe. Some regions have
flat regions joined to the outside by a slant fracture. Other
regions show a full slant fracture mode.
1 Castings were
material used for
region when the
further extended,
made with a polysiloxane precision-impression material, a
dental casting. The material was injected into the crack-tip
CTOA specimen was under load and when the crack was
the casting was released.
213
or
d
is
t
rib
ut
io
n
3rd Quarter, 2013
ot
f
Fig.16. Changes in the
elongation of the X-100,
X-65, and X-70 steels with
test rate.
of around 25% while the X-100 has an elongation
around 20%. As testing rates increase, the percent
elongation decreases slightly for all three alloys (Fig.16).
The results show that the deformation of the grid is
in the loading direction and the grid rectangles rotate
or deform in the cracking direction. The deformation
of the grid lines shows non-uniform �elongation’ at
the �necked’ regions.
-n
something of a mixed mode, but is generally characterized
as a slant fracture, because failure is essentially on a
single shear plane, with no cup-cup or cup-cone shear
region associated with the fracture. In addition, some
regions along the full-scale tested pipe fractured in a
pure slant mode, with little or no flat-fracture regions.
co
py
For the CTOA specimens, the slant-mode failures tend
to have very little or no �flat’ fracture. The constraint
differences and higher rate of crack growth in the fullscale pipe may have influenced the observed difference
in fracture mode.
Sa
m
pl
e
Differences in the appearance of the fracture surfaces
from the full-scale test and laboratory CTOA tests
are also apparent. No laboratory tests have produced
fractures that reproduce all the features observed on
the full-scale, high-rate test fractures. Interestingly, some
of the quasi-static test fractures have more in common
with appearance of the burst-test fracture than do the
highest-speed CTOA fractures compared in this study.
This is because a number of the X-100 tests conducted
at 0.02 mm/s had intermittent crack growth, which has
regions of dynamic crack growth that might be similar
to the actual fracture conditions of the full-scale test.
Plastic deformation
Deformation through the gauge length of the reduced
section indicates that the X-65 and X-70 steels are more
ductile than the X-100 steel, with quasi-static values
Generally the X-100 and X-65 specimens show similar
profiles for thinning due to plastic flow. The shoulder
where the specimen thickness changes from 8 mm to
15 mm constrains the plastic flow, and thinning is
limited in the first 6 mm or 7 mm from the shoulder,
and then increases in a similar manner for all of the
alloys and test rates evaluated. Both fracture modes follow
the same basic trend, although the slant-shear fracture
mode has less thinning during final fracture than the
flat-fracture mode, which �necks’ during final fracture.
The grids show little rotation, indicating that most
of the plastic deformation is parallel to the load line.
Numerical modelling
The stable tearing behaviour of the CTOA test was
modelled. Finite-element analysis (FEA) was applied to
predict the applied load vs crack extension behaviour
of steels #1 – #5, and showed correlation coefficients
between the experimental and FEA results of between
0.92 and 0.993 [21]. This FEA model under-predicted
the initial crack extension, when the crack extension was
214
The Journal of Pipeline Engineering
io
n
rib
ut
ot
f
• The energy dissipation rate R reaches a minimum
value in the case of slant fracture for a final
tilt angle equal to 45В°. This result is consistently
obtained for different material hardening or
damage parameters. The energy dissipation rate
correlates well with CTOA values.
• Stress and strain states in the stable tearing
region hardly depend on the assumed tilt angle.
• The CTOA on the surface of the specimen is
close to the CTOA at the centre of the specimen
(steady-state propagation).
1. A.B.Rothwell, 2000. Fracture propagation control
for gas pipelines - past, present, and future.
Pipeline Technology, 1. Elsevier, Netherlands, pp
387-405.
2. G.M.Wilkowski, Y.-Y.Wang, and D.L.Rudland,
2000. Recent efforts on characterizing propagating
ductile fracture resistance of linepipe steels. Idem,
pp 359-386.
3. S.H.Hashemi, I.C.Howard, J.R.Yates, R.M.Andrews,
and A.M.Edwards, 2004. A single specimen CTOA
test method for evaluating the crack tip opening
angle in gas pipeline steels. Proc. International
Pipeline Conference, pp 0610.1-7.
4. G.Demofonti, G.Mannucci, C.M.Spinelli, L.Barsanti,
and H.-G.Hillenbrand, 2000. Large diameter X100
gas linepipes: fracture propagation evaluation by
full-scale burst test. Pipeline Technology, 1. Elsevier,
Netherlands, pp 509-520.
5. B.N.Leis, R.J.Eiber, L.Carlson, and A.Gilroy-Scott,
1998. Relationship between apparent (total) Charpy
V-notch toughness and the corresponding dynamic
crack propagation resistance. Proc. International
Pipeline Conference, 2, pp 723-731.
6. J.C.Newman, Jr, and M.A.James, 2001. A review
of the CTOA/CTOD fracture criterion – why it
works! Proc. 42nd AIAA/ASME/ASCE/AH/ASC
Structures, Structural Dynamics, and Materials
Conference and Exhibit, paper AIAA-200-1324,
Seattle, Washington, USA, pp 1042-1051.
N.Pussegoda, S.Verbit, A.Dinovitzer, W.Tyson,
7. A.Glover, L.Collins, L.Carlson, and J.Beattie,
2000. Review of CTOA as a measure of ductile
fracture toughness. Proc. International Pipeline
Conference, 1. pp 247-254.
8. D.J.Horsley, 2003. Background to the use of
CTOA for prediction of dynamic ductile fracture
arrest in pipelines. Eng Fract. Mech. 70, 3-4, pp
547-552.
9. G.Mannucci, G.Buzzichelli, P.Salvini, R.J.Eiber, and
L.Carlson, 2000. Ductile fracture arrest assessment
in a gas transmission pipeline using CTOA. Proc.
International Pipeline Conference, 1, pp 315-320.
10.R.Jones and A.B.Rothwell, 1997. Alternatives to
Charpy testing for specifying pipe toughness.
Fracture control in gas pipelines. WTIA/APIA/
CRC-MWJ International Seminar, pp 5-1-21.
O.E.O’Donoghue, M.F.Kanninen, C.P.Leung,
11.
G.Demofonti, and S.Venzi, 1997. The development
and validation of a dynamic fracture propagation
model for gas transmission pipelines. Int. J.Pressure
Vessels Piping, 70, 1, pp 11-25.
12.N.Pussegoda, L.Malik, A.Dinovitzer, B.A.Graville,
and A.B.Rothwell, 2000. An interim approach
to determine dynamic ductile fracture resistance
of modern high toughness pipeline steels. Proc.
International Pipeline Conference, 1, pp 239-45.
is
t
A recent model also focused on X-100 steel [26] via the
computational cell technique in simulation of slant-crack
advance. The dependence of crack-growth parameters
on the tilt angle was systematically investigated, and a
simple GTN model was used to simulate ductile damage
growth within the computational cells. The main results
are summarized as follows:
References
or
d
less than twice the specimen thickness, and accurately
described the crack extension behaviour beyond the
peak stress. In another model, the focus was on X-100
steel, and the failure parameters of the JohnsonCook and Hooputra et al. models were evaluated by
parametrical computation [28]. From this work, the
equivalent plastic-strain parameters of damage, for both
the Johnson-Cook and Hooputra et al. models, were
defined for X-100 steel.
-n
Future work
Sa
m
pl
e
co
py
The difficulty in producing accurate and reproducible direct
CTOA measurements on the surface of test specimens is
clear from our results, and suggests the need for robust,
automated measurement procedures and evaluation of
indirect estimations of CTOA from load-displacement
data. Furthermore, cross-head displacement has been
used as a proxy for crack-mouth-opening displacement
(CMOD), as the clip gauges typically used in these tests
are prone to slip and lag behind when loading rates are
increased. This indicates the need for improved dynamic
displacement measurement which will require a novel
test apparatus and specimen-preparation modifications,
so that more-accurate correlations of crack location,
velocity, crack-path, and crack morphology can lead to
the understanding of the fracture mechanisms and their
associated changes. Correlating crack-tip velocity to the
applied load and far-field deformation for various pipeline
steels, and methods of indirectly calculating CTOA, will
likely be of interest for future FE-modelling efforts.
Acknowledgments
The authors thank the many guest researchers who were
involved in developing and conducting CTOA tests over
the years: P.P.Darcis, G.Kohn, A.Bussiba, A.Shtechman,
R.Reuven, J.M.Treinen, and H.Windhoff.
3rd Quarter, 2013
215
or
d
is
t
rib
ut
io
n
25.
S.Xu, W.R.Tyson, R.Eagleson, C.N.McCowan,
E.S.Drexler, J.D.McColskey, and PhP Dacis, 2010.
Measurement of CTOA using MDCB and DWTT
specimens. Idem, IPC2010-31076.
26.J.Besson, C.N.McCowan, and E.S.Drexler, 2013.
Modeling flat to slant fracture transition using the
computational cell methodology. Eng Fract. Mech.,
104, pp 80-95.
27. J.W.Sowards, C.N.McCowan, and E.S.Drexler, 2012.
Interpretation and significance of reverse chevronshaped markings on fracture surfaces of API X100
pipeline steels. Mat. Sci. Eng A – Struct., 551, pp
140-148.
M.Szanto, C.N.McCowan, E.S.Drexler, and
28.
J.D.McColskey, 2011. Fracture of X100 pipeline
steel: combined experimental-numerical process.
NIST BERB publication (B2011-0116).
29. R.J.Fields, J.D.McColskey, C.N.McCowan, P.P.Darcis,
E.S.Drexler, S.P.Mates, and T.A.Siewert, 2012.
Mechanical properties and crack behavior in line
pipe steel. Final Report to the Department of
Transportation. http://primis.phmsa.dot.gov/matrix/
FilGet.rdm?fil=7883.
30. E.S.Drexler, PhP.Darcis, C.N.McCowan, J.W.Sowards,
J.D.McColskey, and T.A.Siewert ,2011. Ductilefracture resistance in X100 pipeline welds measured
with CTOA. Weld J., 90, 12, pp 241-s - 248-s.
31.PhP.Darcis, C.N.McCowan, J.D.McColskey, and
R.Fields, 2008. Crack tip opening angle measurement
through a girth weld in an X100 steel pipelines.
Fatigue Fract. Mater. Struct.. 31, pp 1065-1078.
32.ASTM, 1999. Standard E1290-99. Standard test
method for crack tip opening displacement (CTOD)
fracture toughness measurement. ASTM Book of
Standards, West Conshohocken, PA, USA.
33.K.-H.Schwalbe, J.C.Newman, Jr, and J.L.Shannon,
Jr, 2005. Fracture mechanics testing on specimens
with low constraint–standardisation activities within
ISO and ASTM. Eng Fract. Mech., 72, pp 557-576.
34.K.-H.Schwalbe, J.Heerens, U.Zerbst, H.Pisarski, and
M.Kocak, 2002. EFAM-GTP 02, The GKSS test
procedure for determining the fracture behaviour
of materials, 2nd issue. GKSS Report 2002/24,
GKSS-Forschungszentrum Geesthacht.
35.ISO/FDIS 22889, 2013. Metallic materials: method
of test for the determination of resistance to stable
crack extension using specimens of low constraint.
J.Heerens and M.Schodel, 2003. On the
36.
determination of crack tip opening angle, CTOA,
using light microscopy and delta-5 measurement
technique. Eng Fract. Mech., 70, 3-4, pp 417-426.
37.S.H.Hashemi, 2012. Comparative study of fracture
appearance in crack tip opening angle testing of
gas pipeline steels. Mat. Sci. Eng A – Struct., 558, pp
702-715.
Sa
m
pl
e
co
py
-n
ot
f
13.
R.M.Andrews, I.C.Howard, A.Shterenlikht,and
J.R.Yates, 2002. The effective resistance of pipeline
steels to running ductile fractures; modelling of
laboratory test data. In: ECF14, Fracture mechanics
beyond 2000. EMAS Publications, Sheffield, UK,
pp 65-72.
14.D.S.Dawicke, 1996. Residual strength predictions
using a CTOA criterion. Proc. FAA-NASA Symposium
on Continued Airworthiness of Aircraft Structures,
Atlanta, GA, USA.
15.K.-H.Schwalbe, J.C.Newman, Jr, and J.L.Shannon,
Jr, 2005. Fracture mechanics testing on specimens
with low constraint–standardisation activities within
ISO and ASTM. Eng Fract.Mech., 72, pp 557-576.
16.ASTM ,2006. Standard E2472-06. Standard test
method for determination of resistance to stable
crack extension under low-constraint conditions.
ASTM Book of Standards, West Conshohocken,
PA, USA.
17.
G.Demofonti, G. Buzzichelli, S.Venzi, and
M.Kanninen, 1995. Step by step procedure for the
two specimen CTOA test. Pipeline Technology, 2.
Elsevier, Netherlands, pp 503-512.
18. A.Shterenlikht, S.H.Hashemi, I.C. Howard, J.R.Yates,
and R.M.Andrews, 2004. A specimen for studying
the resistance to ductile crack propagation in pipes.
Eng Fract. Mech., 71, pp 1997-2013.
19.S.H.Hashemi, R.Gay, I.C.Howard, R.M.Andrews,
and J.R.Yates, 2004. Development of a laboratory
test technique for direct estimation of crack tip
opening angle. Proc. 15th European Conference
of Fracture, Stockholm, Sweden.
20.S.H.Hashemi, I.CHoward, J.R.Yates, R.M.Andrews,
and A.M.Edwards, 2004. Experimental study of
thickness and fatigue precracking influence on the
CTOA toughness values of high grade gas pipeline
steel. Proc. International Pipeline Conference, pp
0681.1-8.
P.P.Darcis, G.Kohn, A.Bussiba, J.D.McColskey,
21.
C.N.McCowan, R.Fields, R.Smith, and J.Merritt,
2006. Crack tip opening angle: measurement and
modeling of fracture resistance in low and high
strength pipeline steels. Idem, IPC2006-10172.
22. A. Shtechman, C.N.McCowan, R.Reuven, E.Drexler,
Ph.Darcis, J.M.Treinen, R.Smith, J.Merritt,
T.A.Siewert, and J.D.McColskey, 2008. Dynamic
apparatus for the CTOA measurement in pipeline
steels. Idem, IPC2008-64362.
P.P.Darcis, C.N.McCowan, H.Windhof f,
23.
J.D.McColskey, and T.A.Siewert, 2008. Crack tip
opening angle optical measurement methods in five
pipeline steels. Eng Fract. Mech., 75, pp 245-246.
24.R.Reuven, E.Drexler, C.McCowan, A.Shtechman,
P.Darcis, M.Treinen, R.Smith, J.Merritt, T.A.Siewert,
and J.D.McColskey, 2008. CTOA results for X65
and X100 pipeline steels: influence of displacement
rate. Proc. International Pipeline Conference,
IPC2008-64363.
io
n
rib
ut
-n
ot
f
or
d
is
t
THE
PROFESSIONAL
INSTITUTE OF
PIPEINE
ENGINEERS
Sa
m
pl
e
co
py
The global organization
for oil and gas
pipeline engineers
• Recognizing your skills and status
• Providing a professional network
• Promoting pipeline engineering
Sign up today!
www.pipeinst.org
3rd Quarter, 2013
217
by Prof. Claudio Ruggieri* and Leonardo L S Mathias
io
n
Fracture-resistance testing of
pipeline girth welds using bend
and tensile fracture specimens
rib
ut
Department of Naval Architecture and Ocean Engineering, University of SГЈo Paulo, SГЈo Paulo,
Brazil
S
ot
f
or
d
is
t
TRUCTURAL-INTEGRITY ASSESSMENTS of pipe girth welds play a key role in the design and safe
operation of pipeline systems. Current practices for structural-integrity assessments advocate the
use of geometry-dependent resistance curves so that crack-tip constraint in both the test specimen and
the structural component is similar. Thus, testing standards now under development to measure fracture
resistance of pipeline steels often employ single-edge-notched (SE(T)) specimens under tension.This paper
presents an investigation of the ductile-tearing properties for a girth weld of an API 5L X-80 pipeline steel
using experimentally measured crack-growth-resistance curves (also termed J-R curves).Testing of the girthweld pipeline steels employed clamped SE(T) specimens and three-point bend (SE(B)) specimens with weld
centreline cracks to determine the J-resistance curves. The experimental toughness data enables further
evaluation of crack growth resistance properties of pipeline girth welds.
T
high-toughness steels, often undergo significant stable
crack growth (a) prior to material failure. Under
sustained ductile tearing of a macroscopic crack, large
increases in the load-carrying capacity for the flawed
structural component are possible beyond the limits
given by the onset of crack-growth initiation. Simplified
engineering approaches for defect assessments, such
as BS7910 [4] and API579 [5] methodologies, among
others, incorporate the effects of ductile tearing on
crack-driving forces to evaluate the severity of crack-like
flaws in structural components in terms of crack-growth
resistance (J-a) curves (also often denoted J-R curves)
in which the J-integral fracture parameter characterizes
the significant increase in toughness over the first few
millimetres of stable crack extension (Da) [1,3].
Sa
m
pl
e
co
py
-n
HE ENGINEERING APPLICATION of fracture
mechanics remains invaluable in assessments of
macroscopic fracture behaviour and defect-analyses
procedures of safety-critical structural components,
including pressure vessels and pipeline girth welds.
Fitness-for-service (FFS) fracture-assessment approaches,
also referred to as engineering-critical assessment (ECA)
procedures, rely upon a single parameter to define the
crack-driving force and to characterize fracture resistance
of the material [1-3]. These approaches provide a means
for introducing acceptance criteria in cracked structural
components by relating the operating conditions to a
critical applied load or critical crack size. While a oneparameter description of applied loading in terms of the
J-integral or the crack-tip-opening displacement (CTOD)
and their corresponding macroscopic measures of fracture
toughness (Jc or Оґc) [3] usually suffices to characterize
the essentially stress-controlled failure by cleavage mode,
quantitative analyses of fracture preceded and accompanied
by extensive plastic deformation becomes more complex.
Low-constraint and structural components containing
defects and flaws, including circumferentially cracked
pipelines and their weldments made of high-grade,
*Corresponding author’s contact details:
tel: +55 11 3091 5184
email: [email protected]
Standardized techniques for crack-growth resistance
testing of structural steels, including ASTM 1820
[6], routinely employ three-point bend (SE(B)) and
compact-tension (C(T)) specimens containing deep,
through cracks (a/W ≥ 0.45-0.50). However, a variety
of crack-like defects are most often surface cracks
formed during in-service operation and exposure to
aggressive environment or during welding fabrication.
Structural components falling into this category include
girth welds made in field conditions for high-pressure
piping systems and steel catenary risers (SCRs). These
crack configurations generally develop low levels of
crack-tip stress triaxiality which contrast sharply to
The Journal of Pipeline Engineering
rib
ut
io
n
218
-n
ot
f
or
d
is
t
(a)
(b)
pl
e
co
py
conditions present in deeply cracked specimens [7].
Recent defect-assessment procedures advocate the use
of geometry-dependent fracture-toughness values so that
crack-tip constraint in both test specimen and structural
component is similar. In particular, fracture-toughness
values measured using single-edge-notch tension (SE(T))
specimens appear more applicable for characterizing the
fracture resistance of pressurized pipelines and cylindrical
vessels than standard, deep-notch, fracture specimens
under predominantly bend loading [8-10].
Sa
m
Recent applications of SE(T) fracture specimens to
characterize crack-growth-resistance properties in pipeline
steels [11] have been effective in providing larger flaw
tolerances while, at the same time, reducing the otherwise
excessive conservatism which arises when measuring the
material’s fracture toughness based on high constraint,
deeply-cracked SE(B) or C(T) specimens. However, some
difficulties associated with SE(T) testing procedures raise
concerns about the significance and qualification of
measured crack-growth-resistance curves. While slightly
more conservative, testing of shallow-crack-bend specimen
configurations may become more attractive due to its
simpler testing protocol and laboratory procedures,
and the much smaller loads required to propagate the
Fig.1. (a) Partial unloading during the
evolution of load with displacement; (b)
definition of the plastic area under the
load-displacement curve.
crack. Although deeply-cracked SE(B) specimens are the
preferred crack geometry often adopted in conventional
defect-assessment methods, recent revisions of ASTM
1820 [6] and ISO 15653 [12] have also included
J-estimation equations applicable to shallow-crack-bend
specimens. Consequently, use of smaller specimens which
yet guarantee adequate levels of crack-tip constraint to
measure the material’s fracture toughness becomes an
attractive alternative.
This work presents an experimental investigation of
the ductile-tearing properties for a girth weld made
of an API 5L X-80 pipeline steel in terms of crackgrowth-resistance curves. Use of these materials is
motivated by the increasing demand in the number
of applications for manufacturing high -strength
pipes for the oil and gas industry. Testing of the
pipeline girth welds employed side-grooved, clamped
SE(T) specimens and three-point-bend SE(B) specimens
with a weld-centreline notch and varying crack sizes to
determine the crack-growth-resistance curves using
the unloading compliance (UC) method and
a single specimen technique. Recently developed
compliance functions and О·-factors applicable for
SE(T) and SE(B) fracture specimens are introduced
3rd Quarter, 2013
219
to determine crack -growth -resistance curves from
laboratory measurements of load-displacement records.
Overall, these experimental results provide toughness
data which enable further evaluation of crack-growthresistance properties of pipeline girth welds.
standards such as ASTM E1820 [6]) to evaluate J with
crack extension follows from an incremental procedure
which updates Je and Jp at each partial unloading point,
denoted k, during the measurement of the load vs
displacement curve depicted in Fig.1(b) as:
Experimental evaluation of
J-resistance curves
k
J=
J ek + J pk (1)
io
n
where the current elastic term is simply given by:
rib
ut
пЈ« K2 пЈ¶
J ek = пЈ¬ I пЈ· (2)
 E ′ k
is
t
and the current plastic term follows an incremental
formulation which is applicable to CMOD data in the
form [14-16] in Equn 3 below, in which the geometry
factor ОіLLD is evaluated from:
or
d
пЈ®
пЈ« bk в€’1 dО· Jk в€’в€’1LLD пЈ¶ пЈ№
k в€’1
Оі LLD
= пЈЇ в€’1 + О· Jk в€’в€’1LLD в€’ пЈ¬пЈ¬
пЈ·пЈ· пЈє (4)
k в€’1
пЈЇпЈ°
пЈ­ WО· J в€’ LLD d ( a W ) пЈё пЈєпЈ»
In the above expressions, KI is the elastic stress intensity
factor for the cracked configuration, Ap represents the
plastic area under the load-displacement curve, BN is
the net specimen thickness at the side groove roots
(BN = B if the specimen has no side grooves, where
B is the specimen gross thickness), and b denotes
the uncracked ligament (b = W - a, where W is the
specimen width and a is the crack length). In the above
Eqn 2, plane-strain conditions are adopted such that
E’ = E/(1-ν)2, where E and ν are the (longitudinal) elastic
modulus and Poisson’s ratio, respectively. The factor
О·J in Eqns 3 and 4 represents a non-dimensional
parameter which relates the plastic-deformation
contribution to the strain energy for the cracked body
and J. Figure 1a illustrates the essential features of
the estimation procedure for the plastic component
Jp, where Ap (and consequently О·J) is defined in terms
of load-LLD (or О”) data or load-CMOD (or V) data.
These geometry factors are denoted О·J-LLD and О·J-CMOD,
respectively, when LLD or CMOD are used.
ot
f
Conventional testing programmes to measure crack-growth
resistance (J-О”a) curves in metallic materials routinely
employ the UC method based on a single specimen
test. The estimation procedure used in ASTM E1820
standard [6] predominantly employs load-line displacement
(LLD) records to measure fracture-toughness-resistance data
incorporating a crack-growth correction for the J-integral.
An alternative method which potentially simplifies the
test procedure involves the use of crack-mouth-opening
displacement (CMOD) to determine both the J-integral
and the amount of crack growth. This section provides
a brief overview of the analytical framework needed to
evaluate
data for common fracture specimens from
laboratory measurements of load-displacement records.
Attention is directed to an incremental procedure to
obtain estimates of J and crack length for the SE(T)
and SE(B) configurations based on CMOD data.
Incremental estimation procedure of the J-integral
m
pl
e
co
py
-n
The procedure to estimate crack-growth-resistance data
considers the elastic and plastic contributions to the strain
energy for a cracked body under Mode I deformation
[3] so that the J-integral conveniently derives from its
elastic component, Je, and plastic term, Jp, as J = Je + Jp.
Here, an estimation scheme for the plastic component
employs a plastic О·-factor introduced by Sumpter and
Turner [13] to relate the crack-driving force to the
plastic area under the load versus LLD (or CMOD) for
cracked configurations (see also Refs [14-16]) as illustrated
in Fig.1(a). The procedure to estimate Jp based on the
О·-methodology has proven highly effective in testing
protocols to measure fracture toughness in stationary
cracks while, at the same time, retaining strong contact
with the deformation plasticity definition of J.
Sa
However, the area under the actual load-displacement
curve for a growing crack differs significantly from
the corresponding area for a stationary crack (upon
which the deformation definition of J is based) [3].
Consequently, the measured load-displacement records
must be corrected for crack extension to obtain an
accurate estimate of J-values with increased crack growth.
A widely used approach (which forms the basis of current
The incremental expression for Jp defined by Eqn 3
coupled with Eqn 4 contains two contributions: one is
from the plastic work in terms of CMOD and, hence
О·J-CMOD, and the other due to crack growth correction
in terms of LLD by means of О·J-LLD; evaluation of Eqns
3 and 4 is relatively straightforward provided that these
two geometric factors are known. For the clamped
SE(T) specimens with H/W = 10 and the conventional
SE(B) specimen with S/W = 4 utilized in this study,
пЈ®
пЈ№ пЈ® Оі k в€’1
пЈ№
О· k в€’1
=
J pk пЈЇ J pk в€’1 + J в€’CMOD ( Apk в€’ Apk в€’1 ) пЈє пЈЇ1 в€’ LLD ( ak в€’ ak в€’1 ) пЈє (3)
bk в€’1 BN
пЈ°
пЈ» пЈ° bk в€’1
пЈ»
220
The Journal of Pipeline Engineering
в€’1
Be=
B
io
n
By measuring the instantaneous compliance during
unloading of the specimen (see Fig.1b), the current
crack length follows directly from solving the functional
dependence of crack length and specimen compliance
in terms of ОјCMOD. For the clamped SE(T) and SE(B)
specimens analysed here, the corresponding compliance
expressions are given by Cravero and Ruggieri [14] and
Appendix X.2 of ASTM E1820 [6] respectively as:
пЈ®aпЈ№
1.921 в€’ 13.219Вµ + 58.708Вµ 2 в€’ 155.282 Вµ 3 пЈЇпЈ°W =
пЈєпЈ»
SET
(12)
+ 207.399 Вµ 4 в€’ 107.917 Вµ 5
0.1 ≤ a / W ≤ 0.7
пЈ®aпЈ№
1.019 в€’ 4.537 Вµ + 9.01Вµ 2 в€’ 27.333Вµ 3
пЈЇпЈ°W =
пЈєпЈ»
(13)
SEB
ot
f
Current testing protocols to measure the crack-growthresistance response using a single-specimen test are
primarily based on the unloading compliance (UC)
technique to obtain accurate estimates of the current
crack length from the specimen compliance measured at
periodic unloadings with increased deformation. Figure
1b illustrates the essential features of the method. The
slope of the load-displacement curve during the k-th
unloading defines the current specimen compliance,
denoted Ck, which depends on specimen geometry
and crack length. For the clamped SE(T) specimen
with
H/W = 10 and the SE(B) specimen with
S/W = 4 analysed here, the specimen compliance
is often defined in terms of normalized quantities
expressed as [6, 14]:
2
rib
ut
Crack-length estimation
( B в€’ BN ) (11)
Bв€’
is
t
Equations 7 and 8 applicable to SE(B) specimens agree
very well with the О·-factors in the revised J-integral
expressions developed by Zhu et al. [19] which form
the basis of current ASTM E1820 [6] and ISO 15653
[12] standards using CMOD records.
CCMOD denotes the specimen compliance in terms of
CMOD (CCMOD = V/P) and the effective thickness, Be,
is defined by:
or
d
a convenient polynomial fitting of the results given by
Cravero and Ruggieri [14], Donato and Ruggieri [17],
and Ruggieri [18] provides the corresponding О·-factor
equations for homogeneous materials in the form shown
in Equns 5-8 below.
+ 74.4Вµ 4 в€’ 71.489Вµ 5
0.05 ≤ a/W ≤ 0.45
Effect of weld strength overmatch on plastic О·-factors
and
Current test standards employ J-estimation expressions
which are mainly applicable to fracture specimens made
of homogeneous materials. For a given specimen geometry,
mismatch between the weld metal and base-plate strength
affects the macroscopic mechanical behaviour of the
specimen in terms of its load-displacement response,
with a potentially strong impact on the coupling
relationship between J and the near-tip stress fields.
в€’1
(10)
py
пЈ®
EWBeCCMOD пЈ№
SEB
ВµCMOD
= пЈЇ1 +
пЈє
S 4
пЈ°
пЈ»
-n
SET
ВµCMOD
= пЈ°пЈ®1 + EBeCCMOD пЈ№пЈ» (9)
e
co
where ОјSETCMOD and ОјSEBCMOD define the normalized
compliances for the SE(T) and SE(B) specimens. In the
above expressions, E is the longitudinal elastic modulus,
2
3
4
a
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
+ 7.81пЈ¬ пЈ· в€’ 18.27 пЈ¬ пЈ· + 15.30 пЈ¬ пЈ· в€’ 3.08 пЈ¬ пЈ·
W
W
W
W
пЈ­ пЈё
пЈ­ пЈё
пЈ­ пЈё
пЈ­W пЈё
pl
m
О· JSET
= 1.07 в€’ 1.77
в€’CMOD
5
(5)
Sa
0.2 ≤ a / W ≤ 0.7
О· JSET
в€’ LLD = в€’0.62 + 9.34
2
3
4
a
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
в€’ 4.58 пЈ¬ пЈ· в€’ 47.96 пЈ¬ пЈ· + 87.70 пЈ¬ пЈ· в€’ 44.88 пЈ¬ пЈ·
W
пЈ­W пЈё
пЈ­W пЈё
пЈ­W пЈё
пЈ­W пЈё
5
(6)
0.2 ≤ a / W ≤ 0.7
2
a
пЈ«aпЈ¶
О· = 3.65 в€’ 2.11 + 0.34 пЈ¬ пЈ· (7)
W
пЈ­W пЈё
0.1 ≤ a / W ≤ 0.7
SEB
J в€’CMOD
2
3
4
5
a
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
пЈ«aпЈ¶
О· = 0.02 + 18.09 в€’ 73.26 пЈ¬ пЈ· + 152.22 пЈ¬ пЈ· в€’ 159.777 пЈ¬ пЈ· + 66.88 пЈ¬ пЈ· (8)
W
пЈ­W пЈё
пЈ­W пЈё
пЈ­W пЈё
пЈ­W пЈё
SEB
J в€’ LLD
0.1 ≤ a / W ≤ 0.7
221
rib
ut
io
n
3rd Quarter, 2013
where ПѓMBys and ПѓWMys denote the yield stresses for the
base-plate metal and the weld metal.
ot
f
Accurate estimation formulas for J more applicable to
welded fracture specimens may become important in
robust defect-assessment procedures capable of including
effects of weld-strength mismatch on fracture toughness.
or
d
is
t
Fig.2. Geometry of tested
fracture specimens with
weld-centreline notch and
B x B cross section:
(a) clamped SE(T) specimen
with a/W = 0.4 and
H/W = 10; (b) three-point SE
(B) specimen with
a/W = 0.25 and S/W = 4.
Material description and welding procedures
co
py
-n
Previous work by Donato et al. [20], and Paredes and
Ruggieri [21], introduced a functional dependence of
geometry factor О·J-CMOD on crack size and weld-strength
mismatch for weld centreline fracture specimens. The
expressions of factor О·J-CMOD for the weld cracked
SE(B) with S/W = 4 and SE(T) with H/W = 10 are
summarized as follows:
Experimental details
= 3.88 + 0.22
О· JSEB
в€’CMOD
3
a
пЈ«aпЈ¶
в€’ 5.01пЈ¬ пЈ·
W
пЈ­W пЈё
2
(14)
e
пЈ«aпЈ¶
+ 4.02 пЈ¬ пЈ· в€’ 0.41M y в€’ 0.05M y2
пЈ­W пЈё
m
pl
0.2 ≤ a/W ≤ 0.7, 1.0 ≤ My ≤ 1.5
Sa
О· JSET
в€’CMOD = в€’0.36 + 11.69
a
пЈ«aпЈ¶
в€’ 23.59 пЈ¬ пЈ·
W
пЈ­W пЈё
3
2
(15)
пЈ«aпЈ¶
+ 13.90 пЈ¬ пЈ· в€’ 0.28M y в€’ 0.03M y2
пЈ­W пЈё
0.2 ≤ a/W ≤ 0.7, 1.0 ≤ My ≤ 1.5
In the above expressions, the mismatch ratio, My, is
defined as:
My =
Пѓ WM
ys
Пѓ ysMB
(16)
The material used in this study was a high-strength,
low-alloy (HSLA), API grade X-80 pipeline steel
produced as a base plate using a control-rolled processing
route without accelerated cooling. The mechanical
properties and strength/toughness combination
for this material are mainly obtained by both
grain-size refinement and second-phase strengthening
due to the small-size precipitates in the matrix.
A 20-in diameter pipe with longitudinal seam
weld from which the girth weld SE(T) and SE(B)
specimens were extracted was fabricated using the
UOE process.
The tested weld joint was made from the API X-80
UOE pipe having wall thickness tw = 19 mm. Girth
welding of the pipe was performed using the FCAW
process in the 1G (flat) position with a single V-groove
configuration in which the root pass was made by
GMAW welding. The main weld parameters used
for preparation of the test weld using the FCAW
process are:
(i) n
umber of passes = 12 (including the root
pass made by the GMAW process);
(ii) welding current = 165 A;
(iii)welding voltage = 23 V;
(iv) average heat input = 1.5 kJ/mm.
The Journal of Pipeline Engineering
rib
ut
io
n
222
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
Fig.3. Measured load-CMOD
curve for the tested X-80
pipeline girth weld using
clamped SE(T) specimens with
a/W = 0.4 and three-point SE(B)
specimen with a/W = 0.25.
Mathias et al. [22] provide the tensile properties for the
tested pipeline girth weld and base material which include:
ПѓWMys = 715MPa
ПѓWMuts = 750 MPa
ПѓBMys = 609 MPa
ПѓBMuts = 679 MPa.
Here, σys and σuts represent the material’s yield stress
and tensile strength, and WM and BM denote the
weld metal and the base plate.
Fig.4. J-resistance curves for the
clamped SE(T) specimens with
a/W = 0.4 and H/W = 10.
Specimen geometries
Mathias [23] conducted unloading compliance tests at
room temperature on weld-centreline-notched SE(T)
specimens with fixed-grip loading to measure tearing
resistance curves in terms of J – Δa data. The clamped
SE(T) specimens have a fixed overall geometry and crack
length to width ratio defined by a/W = 0.4, H/W
= 10 with thickness B = 14.8 mm, width W = 14.8
mm (W = B) and clamp distance H = 148 mm (refer
to Fig.2a). Here, a is the crack depth and W is the
223
is
t
rib
ut
io
n
3rd Quarter, 2013
mode on the J – Δa data. We first draw attention to
the load-carrying capacity for the bend and tension
configurations. Figure 3 shows a typical load-CMOD
curve measured from tests of the SE(T) specimen with
H/W = 10 and a/W = 4, and the SE(B) specimen
with S/W = 4 and a/W = 0.25. The strong effect
of loading mode (tension vs bending) associated with
specimen geometry is evident in this plot. At similar
levels of CMOD, the applied load for the SE(T)
specimen increases approximately by a factor of four
compared to the load response for the SE(B) specimen.
co
py
-n
ot
f
specimen width which is slightly smaller than the pipe
wall thickness, tw. The UC tests at room temperature
were also conducted on weld-centreline-notched SE(B)
specimens with a/W = 0.25, specimen thickness B =
14.8 mm, width W = 14.8 mm (B = W) and span S = 4W
(refer to Fig.2b). Conducted as part of a collaborative
research programme at the University of SГЈo Paulo on
structural-integrity assessment of marine steel-catenary
risers (SCRs), testing of these specimens focused on the
development of accurate procedures to evaluate crackgrowth resistance data for pipeline girth welds. Mathias
et al. [22] provided more details on the materials used
and fracture testing of the X-80 pipeline girth weld.
or
d
Fig.5. J-resistance curves
for the three-point SE(B)
specimens with a/W = 0.25.
Sa
m
pl
e
All specimens, including the SE(T) configuration, were
pre-cracked in three-point bend conditions. After fatigue
pre-cracking, the specimens were side-grooved to a net
thickness of approx. 85% the overall thickness (7.5%
side-groove on each side) to promote uniform crack
growth along the crack front, and tested following the
general guidelines described in ASTM E1820 [6]. Records
of load vs CMOD were obtained for the specimens
using a clip gauge mounted on knife edges attached
to the specimen surface.
Crack-growth-resistance curves
Effect of specimen geometry on J-resistance curves
The framework for determining J-resistance curves
based on CMOD from conventional fracture specimens
described previously provides the basis for evaluating
the ductile-fracture response of the tested material and
assessing effects of specimen geometry and loading
Evaluation of the crack-growth-resistance curve follows
from determining J and О”a at each unloading point of
the measured load-displacement data based upon the
previous formulations for the О·-factors and compliance
functions. Figures 4 and 5 compare the measured
crack-growth-resistance curves for the SE(T) and SE(B)
specimens: it can be seen that the resistance curves for
the shallow-crack SE(B) specimen are comparable to the
J-R curves corresponding to the deeply-cracked SE(T)
specimen. Here, the average J-values at fixed amounts
of crack growth, О”a, for both crack configurations are
reasonably similar. Another significant features associated
with these plots include:
• The average slope of the J-R curve for the
SE(B) specimen (which is also related to
the material’s tearing modulus) is slightly
higher than the corresponding slope for the
SE(T) specimen.
• The estimated value of the J-integral at onset
of ductile tearing, JIc, is fairly independent of
specimen geometry and loading mode.
The Journal of Pipeline Engineering
rib
ut
io
n
224
Configuration
Measured post test
Compliance estimation
ot
f
Specimen
or
d
is
t
Fig.6. J-resistance curves for
the SE(T) specimen and SE(B)
specimen based upon О·-factors
for homogeneous materials and
overmatched welds.
Deviation
(%)
a0 (mm)
af (mm)
О”a (mm)
af (mm)
О”a (mm)
8.79
3.13
8.77
3.11
0.78
8.66
2.55
8.56
2.45
3.87
6.29
9.32
3.03
9.20
2.92
3.75
6.70
10.59
3.89
10.59
3.89
0.03
SE(T) H/W=10
5.66
SET2 H10
SE(T) H/W=10
6.11
SET3 H10
SE(T) H/W=10
SET4 H10
SE(T) H/W=10
SEB1
SE(B) a/W=0.25
4.38
6.28
1.90
5.89
1.39
26.84
SE(B) a/W=0.25
4.99
7.00
2.01
6.38
1.48
26.37
SE(B) a/W=0.25
4.48
6.65
2.17
6.12
1.78
17.97
SEB4
SE(B) a/W=0.25
3.75
6.50
2.75
5.84
1.94
29.45
SEB5
SE(B) a/W=0.25
3.93
6.25
2.32
5.30
1.65
28.88
py
m
pl
e
SEB3
co
SEB2
-n
SET1 H10
Sa
Table 1. Predicted and measured crack extension for tested fracture specimens.
Unfortunately, the measured resistance curves are perhaps
somewhat more scattered than we would expect for
these specimens, particularly for the bend configuration.
While we did not thoroughly investigate such behaviour,
the crack-front measurements conducted by Mathias
[23] revealed a somewhat highly uneven crack advance,
thus providing some explanation for the scatter of the
measured resistance curves. However, it is evident that
the J-resistance data for the SE(B) configuration compare
relatively well with the SE(T) specimen results.
Effect of weld strength overmatch on J-resistance
curves
The effect of weld-strength mismatch on the fractureresistance response characterized by the J - О”a data is
examined here for the tested SE(T) and SE(B) specimens
with weld-centreline notch. The primary interest is to
assess the potential deviation that arises from evaluating
the J-resistance curves using О· equations developed for
homogeneous materials.
225
Figure 6 compares the J-resistance curves for the
shallow-crack SE(B) specimen and deep-crack SE(T)
specimen based on О·-factors for homogeneous materials
and overmatched welds, as represented by open
and solid symbols. The О·-values for the overmatch
condition are determined from using the estimation
Eqns 14 and 15 provided previously, with My =
1.18 (refer to Eqn 16). To facilitate comparison,
only the lowest and highest resistance curves for
these crack configurations are included in the plot.
It can be seen that the fracture-resistance curves
derived from О·-factors for overmatched welds
are practically indistinguishable from the curves
evaluated with О·-factors for homogeneous materials.
Here, the use of О·-factors for homogeneous
materials (i.e. not taking into account the degree of
weld strength overmatch) leads to slightly
non-conservative (higher) estimates of the resistance
curve (we should emphasize that the larger the levels
of weld-strength mismatch, the larger the degree of
non-conservativeness).
or
d
is
t
rib
ut
(a)
Crack-length estimates
(b)
py
-n
ot
f
After testing, all specimens were subjected to heattinting treatment (300В°C for 30 min), and then air
cooled before being broken apart. Following standard
methods based on the nine-point average technique,
such as the procedure given by ASTM E1820 [6], the
initial and final crack length measured after the test
by means of an optical method were compared with
crack length estimates derived from the UC method.
Table 1 provides the predicted and measured crack
extension for all tested fracture specimens, where the
deviation is defined as:
io
n
3rd Quarter, 2013
co
Δameasured – Δapredicted/ Δameasured
(c)
e
The significant features that emerge from these results
include:
Sa
m
pl
• predictions of crack extension based on the UC
procedure for the SE(T) specimens are in close
agreement with experimental measurements with
a level of accuracy of approx 5%;
• crack-extension predictions for the shallow-crack
SE(B) configuration derived from the UC
procedure are not in good agreement with
the measured amount of ductile tearing; here,
the UC method underestimates the nine-point
average crack extension by approx. 25-30%, which
produces apparent higher J-resistance curves.
The crack-growth behaviour for the shallow-crack SE(B)
configuration can be explained in terms of the uneven
crack advance and a rather irregular crack-front profile
observed in these specimens. Figure 7 shows typical
crack surfaces for the SE(B) and SE(T) configurations:
(d)
Fig.7.Typical fracture surfaces of tested crack configurations: (a,
b) three-point SE(B) specimen with a/W = 0.25; (c, d) clamped
SE(T) specimen with a/W = 0.4.
226
The Journal of Pipeline Engineering
co
py
-n
This study described an experimental investigation of
the ductile-tearing properties for a girth weld made
of an API 5L X-80 pipeline steel and experimentally
measured crack growth resistance curves (J – Δa curves).
Testing of the pipeline girth welds used side-grooved,
clamped SE(T) specimens and shallow-crack-bend SE(B)
specimens with a weld-centreline notch to determine
the crack-growth-resistance curves based upon the
unloading compliance (UC) method using a single
specimen technique. The work described here supports
the following conclusions:
m
pl
e
• Shallow-crack SE(B) specimens (a/W = 0.25)
provide crack-growth-resistance curves which are
comparable to the J-resistance curves for deepcrack SE(T) specimens (a/W = 0.4). Despite
the relatively larger scatter of the J – Δa
data, the fracture resistance for the shallowcrack SE(B) configuration at a fixed amount
of crack growth, О”a, is relatively similar to
the corresponding fracture resistance for the
SE(T) specimen.
• Levels of weld-strength overmatch within the
range of 10-20% (approx.) overmatch do not
significantly affect J-resistance curves derived
from using О·-values applicable to homogeneous
materials. While the fracture resistance curves
based on О·-values for homogeneous materials
are slightly higher than those based on О·-factors
for overmatched weldments, differences are
nevertheless small and within acceptable limits.
Sa
io
n
rib
ut
ot
f
Concluding remarks
While the analyses described here clearly provide
support for using shallow-crack-bend configurations as
an alternative fracture specimen to measure crack-growth
properties for pipeline girth welds and similar structural
components, they are also suggestive of the need for more
experimental studies to validate the UC-based procedure
for estimating J-resistance curves of SE(B) configurations.
In particular, more-accurate techniques for crack-length
estimations in small-size-bend specimens appear central
to develop a robust and efficient J-resistance evaluation
procedure. Additional work is in progress along these lines
of investigation.
is
t
The UC procedure described previously to estimate the
current crack length involves the assumption of a straight
crack front. Consequently, the compliance equations
described by Eqns 12 and 13 should be viewed as
idealized solutions providing estimates for the average
crack extension. Moreover, the inaccurate estimate of
crack extension resulting from these analyses is also
suggestive of a strong effect of the bend-loading mode
on crack-length predictions. Indeed, previous studies
[24, 25] have already indicated that use of the UC
method with three-point-bend specimens underestimates
crack extension when compared with optically measured
values of crack length; this effect appears to be more
pronounced for SE(B) specimen with a shallow crack.
• Crack-extension predictions based on the
UC procedure agree well with experimental
measurements for the SE(T) specimens. In
contrast, the unloading-compliance method
underestimates the nine-point average crack
extension for the shallow-crack SE(B) specimen
by 25-30% (approx.). This rather strong
under-prediction of crack extension for this
crack configuration produces apparent higher
J-resistance curves and, at the same time,
underlies some limitations of current UC
estimation equations to predict crack length
in small-sized-bend specimens.
or
d
it can be seen that the bend specimens exhibited a
highly non-uniform fatigue pre-crack compared to the
SE(T) specimens. Such a feature could be caused by
unexpected misalignment between the specimen and
the rollers, thereby affecting the crack-tip stresses and
strains driving the ductile-fracture process. However,
crack tunnelling is less pronounced for the SE(B)
configuration than for the SE(T) specimen after some
amount of ductile tearing.
Acknowledgments
This investigation is primarily supported by Fundação de
Amparo Г Pesquisa do Estado de SГЈo Paulo (FAPESP)
through Grant 2009/54229-3 and by AgГЄncia Nacional
de PetrГіleo, GГЎs Natural e BiocombustГ­veis (ANP).
The work of CR is also supported by the Brazilian
Council for Scientific and Technological Development
(CNPq) through Grants 304132∕2009-8 and 476581∕20095. The authors acknowledge Tenaris-Confab Brasil and
Lincoln Electric Brasil for providing support for the
experiments described in this work.
References
1. J.W.Hutchinson, 1983. Fundamentals of the
phenomenological theory of nonlinear fracture
mechanics.’ J. Applied Mechanics, 50, pp 1042-1051.
2. U.Zerbst, R.A.Ainsworth, and K.-H.Schwalbe,
2000. Basic principles of analytical flaw assessment
methods. Int. J. Pressure Vessels and Piping, 77, pp
855-867.
3. T.L.Anderson, 2005. Fracture mechanics: fundaments
and applications. 3rd Edn, CRC Press, New York.
4. British Standard Institution, 2005. Guide to
methods for assessing the acceptability of flaws
in metallic structures, BS7910.
5. American Petroleum Institute, 2007. API Standard
579-1/ASME FFS-1, Fitness-for-service, 2nd Edn.
6. ASTM, 2011. ASTM E1820: Standard test method
for measurement of fracture toughness
3rd Quarter, 2013
227
or
d
is
t
rib
ut
io
n
17.G.H.B.Donato and C.Ruggieri, 2006. Estimation
procedure for J and CTOD fracture parameters using
three point bend specimens. Proc. International
Pipeline Conference (IPC 2006), American Society
of Mechanical Engineers, Calgary, Canada.
18.C.Ruggieri, 2012. Further results in J and CTOD
estimation procedures for SE(T) fracture specimens
- Part I: homogeneous materials. Eng Fract. Mech.,
79, pp 245-265.
19. X.-K.Zhu, B.N.Leis, and J.A.Joyce, 2008. Experimental
estimation of J-R curves from load-CMOD record
for SE(B) Specimens. J. ASTM International, 5, JAI
101532.
G.H.B.Donato, R.Magnabosco, and C.Ruggieri,
20.
2009. Effects of weld strength mismatch on J and
CTOD estimation procedure for SE(B) specimens.
Int. J. Fracture, 159, pp 1-20.
21.M.Paredes and C.Ruggieri, 2012. Further results
in J and CTOD estimation procedures for SE(T)
fracture specimens - Part II: center cracked welds.
Eng Fract. Mech., 89, pp 24-39.
22. L.L.S.Mathias, D.F.B.Sarzosa, and C.Ruggieri. Effects
of specimen geometry and loading mode on crack
growth resistance curves of a high-strength pipeline
girth weld. Int. J. Pressure Vessels and Piping (submitted
for publication).
23.L.L.S.Mathias, 2013. Experimental evaluation of J-R
curves in X80 pipeline girth welds using SE(T)
and SE(B) fracture specimens. Master’s Thesis,
Polytechnic School, University of SГЈo Paulo (in
Portuguese).
24.P.A.J.M.Steekamp, 1988. J-R curve testing of threepoint bend specimen by the unloading compliance
method. Fracture Mechanics: 18th Symposium,
ASTM STP 945, American Society for Testing and
Materials, Philadelphia, pp 583-610.
25.J.Dzugan, 2003. Crack length calculation by the
unloading compliance technique for Charpy size
specimens. Wissenschaftlich-Technische Berichte
FZR-385, Forschungszentrum Rossendorf (FZR).
Sa
m
pl
e
co
py
-n
ot
f
7. G.H.B.Donato and C.Ruggieri, 2008. Constraint
effects and crack driving forces in surface cracked
pipes subjected to reeling. Proc. ASME International
Conference on Offshore Mechanics and Arctic
Engineering (OMAE 2008), American Society of
Mechanical Engineers, Lisbon, Portugal.
8. B.Nyhus, M.Polanco, and O.Orjasæter, 2003. SENT
specimens as an alternative to SENB specimens for
fracture mechanics testing of pipelines. Proc. ASME
International Conference on Offshore Mechanics
and Arctic Engineering (OMAE 2003), American
Society of Mechanical Engineers, Cancun, Mexico.
9. S.Cravero and C.Ruggieri, 2005. Correlation of
fracture behavior in high pressure pipelines with
axial flaws using constraint designed test specimens
- Part I: Plane-strain analyses. Eng Fract. Mech., 72,
pp 1344-1360.
10.
L.A.L.Silva, S.Cravero, and C.Ruggieri, 2006.
Correlation of fracture behavior in high pressure
pipelines with axial flaws using constraint designed
test specimens - Part II: 3-D effects on constraint.
Idem, 73, pp 2123-2138.
11.D.Y.Park, W.R.Tyson, J.A.Gianetto, G.Shen, and
R.S.Eagleso, 2010. Evaluation of fracture toughness
of X100 pipe steel using SE(B) and clamped SE(T)
single specimens. Proc. International Pipeline
Conference, Calgary, Canada.
12.
International Organization for Standardization,
2010. Metallic materials в€’ method of test for the
determination of quasistatic fracture toughness of
welds, ISO 15653.
13.J.D.G.Sumpter and C.E.Turner, 1976. Method for
laboratory determination of Jc. Cracks and Fracture,
ASTM STP601, American Society for Testing and
Materials, pp 3-18.
14. S.Cravero and C.Ruggieri, 2007. Estimation procedure
of J-resistance curves for SE(T) fracture specimens
using unloading compliance. Eng Fract. Mech., 74,
pp 2735-2757.
15.Ibid., 2007. Further developments in J evaluation
procedure for growing cracks based on LLD and
CMOD data. Int J. Fracture, 148, pp 387-400.
16.X.-K.Zhu and J.A.Joyce, 2012. Review of fracture
toughness (G, K, J, CTOD, CTOA) testing and
standardization. Eng Fract. Mech., 85, pp 1-46.
io
n
rib
ut
is
t
or
d
Houston, February 10–13, 2014
-n
ot
f
.
.
r
e
e
f
n
n
c
o
e
C
Exhibition
s
e
s
r
u
Co
its 26 year, the PPIM Conference is recogniz
ed as
Now entering
tional forum for sharing and learning ab
a
n
r
e
t
n
i
t
s
o
m
e
r
out best
the fo
aintenance and condition-monitoring te
m
e
m
i
t
e
if
l
n
i
s
e
chnology
practic
for natural gas, crude oil and product pipelines.
www.clarion.org
Sa
m
pl
e
co
py
th
The international gathering of the global pigging industry!
Organized by
3rd Quarter, 2013
229
is
t
Tubular Goods Research Institute, CNPC, Xi’an, China
rib
ut
by Dr He Li*, Qiang Chi, Jiming Zhang,Yang Li, Lingkang Ji,
and Chunyong Huo
io
n
Fracture-toughness evaluations
by different test methods for the
Chinese Second West-East gas
transmission X-80 pipeline steels
F
-n
ot
f
or
d
RACTURE RESISTANCE DETERMINATION was one of the most important aspects during the
research and development of the Chinese 2nd West-East X-80 pipeline steels. More than 30 kinds
of longitudinal submerged-arc-welded (LSAW) X-80 pipes were tested using the 2-mm striker Charpy
V-notch (CVN) test, the 8-mm striker CVN test, and the instrumented drop-weight tear test (DWTT).
It was found that the threshold energy was about 200 J for the 2-mm striker and 8-mm striker CVN
tests. Below this threshold, the difference between the 2-mm striker CVN energy and the 8-mm striker
CVN energy was small. Above the threshold, however, the difference between the 2-mm striker CVN
energy and the 8-mm striker CVN energy increases as the CVN energy increases. It was also found that
there is a linear relation between the DWTT energy and the 8-mm striker CVN energy when the latter
is lower than 400 J/cm2. Otherwise, the DWTT energy trends to reach a plateau.
I
e
co
py
N ORDER TO MEET the sharply increased gas
consumption in China, high-grade X-80 linepipes
have been used for the second West-East gas pipeline
(WEGP). During the research and development of
Chinese X-80 linepipe steels for the WEGP, fractureresistance determination was been taken as one of
the most important issues.
Sa
m
pl
World-wide, Charpy V-notch (CVN) impact tests are
undertaken as the standard method to determine the
fracture resistance, and ASTM E23 [1] and ISO 148
[2] have been issued by American Society for Testing
and Materials (ASTM) and International Organization
for Standardization (ISO), respectively. In China, the
national standard GB/T 229 [3] has been widely
used with procedures and specifications similar to
ISO 148. Unfortunately, there are two different specimen
striker/hammer designs: ASTM E23 adopts the Charpy
striker with a tup radius of 8 mm, while ISO 148 and
*Corresponding author’s contact details:
tel: +86 29 8872 6155
email: [email protected]
GB/T 229 adopt the Charpy striker with a tup radius
of 2 mm. Consequently, the equivalence of the test
results is questionable. Some previous studies showed
that there is an increasing CVN energy difference
between the 2-mm striker and the 8-mm striker after
a threshold energy. The threshold energy appears to
be material-dependent and is related to the fracture
characteristics of the material [4, 5].
On the other hand, the CVN specimens were
considered by some researchers to be too small to
develop significant propagation and therefore might
be inappropriate to quantify fracture resistance for
today’s linepipes [6]. This leads to the adoption of the
drop-weight tear test (DWTT) energy as an alternative
fracture resistance [7]. However, the choice of DWTT
or CVN energy for fracture control of pipelines is
still controversial.
In the present study, a series of tests were performed
for the 2nd WEGP X-80 pipeline steels. The energy
difference determined by the 2-mm and 8-mm striker
CVN tests, as well as the energy difference determined
by the 8-mm striker CVN test and instrumented
DWTT, are discussed and clarified.
230
The Journal of Pipeline Engineering
Nbпј‹V
Ceq
+ Ti
C
Si
Mn
P
S
Mo
Cu
Ti
Nb
Requirements of 2nd
WEGP, maximum
0.09
0.42
1.85
0.022
0.005
0.35
0.3
0.025
0.11
–
–
0.23
Experimental
materials,
average
0.05
0.24
1.76
0.012
0.001
0.28
0.23
0.01
0.073
0.106
0.46
0.19
Pcm
io
n
Element
Table 1.The chemical composition of 2nd WEGP X-80 pipeline steels (wt %).
ASTM E23-06
rib
ut
GB/T 229-2007 ISO148
Allowable
deviation (mm)
Nominal
dimension (mm)
Allowable
deviation (mm)
Length (L)
55
В±0.6
55
+0, -2.5
Height (H)
10
В±0.075
10
В±0.075
Width (W)
10
В±0.11
Angle of notch (o)
45
В±2
Height below notch
8
В±0.075
45
В±1
В±0.075
–
–
В±0.025
0.25
В±0.025
В±0.42
–
–
В±2В°
ot
f
90В°
В±2В°
90В°
В±2В°
90В°
±10’
Distance of plane of symmetry of
notch from ends of test piece
27.5
Angle between plane of symmetry of
notch and longitudinal axis of test
piece
90В°
-n
0.25
or
d
10
Radius of curvature at base of notch
Angle between adjacent longitudinal
faces of test piece
is
t
Nominal
dimension (mm)
pl
e
co
py
Table 2. Dimensional requirements of Charpy V-notch specimens according to GB/T 229-2007 and ASTM E23-06.
(b)
m
(a)
Sa
Fig.1. Photos of (a) CVN specimen and (b) pressed-notch DWTT specimen.
Materials and experimental
procedures
The 2nd WEGP longitudinal submerged-arc welded
(LSAW) X-80 pipes with an outside diameter of 1219
mm (48 in) and wall thickness of 18.4 mm were used
as the experimental materials. The average chemical
composition of the linepipe material is shown in
Table 1. Initially, the materials were tested by a 750-J
CVN machine using 2-mm and 8-mm impact strikers
according to GB/T 299 and ASTM E23, respectively.
Among these Charpy impact tests, three sets were carried
out at temperatures of 20oC, 0oC, -10oC, -20oC, -40oC,
and -60 oC, while other tests were carried out at room
temperature (approx. 20oC).Secondly, the materials were
tested by a 50,000-J instrumented DWTT machine at
room temperature (approx. 20oC). The load-displacement
curves were recorded during the tests. The CVN
specimens had a size of 55 mm x 10 mm x 10 mm
with a 2-mm machined notch, as shown in Fig.1a.
3rd Quarter, 2013
231
CVN energy (J)
Specimen 1
Temperature (ЛљC)
Specimen 2
Specimen 3
8-mm striker
2mm striker
8mm striker
2mm striker
8mm striker
20
305
400
263
389
302
431
0
300
420
270
396
295
-10
295
385
250
349
288
-20
290
350
208
278
-40
295
356
207
189
-60
285
330
98
128
io
n
2-mm striker
396
rib
ut
415
340
165
217
171
158
is
t
291
or
d
Table 3. Comparison between CVN energy determined by the 2-mm and 8-mm strikers for specimens #1, #2, and #3.
ot
f
The DWTT specimens had a size of 305 mm x 76.2
mm x 18.4 mm with a 5-mm pressed notch, as shown
in Fig.1b. The DWTT specimens can be seen to be
much larger than the CVN specimens.
py
-n
Table 2 shows the dimension requirements by GB/T
229-2007 (or ISO148) and ASTM E23-06, and Fig.2
shows the schematic of the 2-mm and 8-mm strikers.
As shown in Table 2, the Charpy test standards GB/T
229-2007 (or ISO148) and ASTM E23 have similar
size requirements; the most significant difference is the
pendulum striker radius, as shown in Fig.2.
co
The relationship between the 2-mm
and 8-mm striker CVN energy
Sa
m
pl
e
Table 3 lists the CVN energy tested by the 2-mm and
8-mm strikers at six different temperatures for specimens
#1, #2, and #3. Hereafter, KV2 is used to indicate the
CVN energy tested by the 2-mm striker, while KV8 is
used to indicate the CVN energy tested by the 8-mm
striker. Figure 3 shows that for specimen #1, the KV2
values at different temperatures are almost equal (approx.
290 J), while the KV8 (approx. 400 J) values at higher
temperatures (0oC and 20oC) are obviously higher than
those (approx. 340 J) at lower temperatures (-60oC,
-40oC, and -2oC). And, KV8 is higher than KV2 at all
test temperatures from -60oC to +20oC. For specimens
#2 and #3, the trend of CVN change with temperature
is similar. At lower temperatures (-60oC and -40oC),
the KV2 and KV8 are almost equal, while at higher
temperatures (-10oC, 0oC, and 20oC), KV8 is much higher
than KV2. It is noted that the KV8 and KV2 values
have the similar ductile-brittle transition behaviour, and
both KV8 and KV2 reach the upper shelf energy at
temperatures above -10oC.
(a)
(b)
Fig.2. Schematic of Charpy impact strikers: (a) 2-mm striker
used in ISO148; and (b) 8-mm striker used in ASTM E23.
The Journal of Pipeline Engineering
rib
ut
io
n
232
py
-n
ot
f
or
d
is
t
(a)
Sa
m
pl
e
co
(b)
(c)
Fig.3. Average CVN energy versus
temperature for (a) specimen #1,
(b) specimen #2, and (c) specimen
#3, tested by the 2-mm and 8-mm
strikers.
233
rib
ut
io
n
3rd Quarter, 2013
co
py
-n
ot
f
or
d
is
t
Fig.4. CVN energy determined by
the 8-mm striker vs CVN energy
determined by the 2-mm striker.
pl
e
Fig.5. DWTT total energy versus
CVN energy (KV8).
Sa
m
Figure 4 shows the relationship between KV2 and KV8
for the 30 specimens tested at room temperature in the
present study. A general equivalence between the KV2
and KV8 is observed for the CVN energy up to about
200 J, after which KV8 is greater than KV2 with the
difference increasing with increasing energy. The present
results are consistent with Naniwa’s observations [8] on
carbon and low-alloy steels. The energy of 200 J as
the threshold was also reported in Ref.9. Nevertheless,
other thresholds such as 60 J [10] and 100 J [11] have
been reported for different steels. So, it is reasonable to
assume that the threshold energy is material-dependent.
Li’s study [12] showed that the KV8 is almost the same
as the KV2 if the specimens were broken into two parts;
otherwise, the KV8 is significantly larger than KV2 if
the specimens were unbroken. Similar observations have
been found in the present study: (a) when both KV2
and KV8 were below 200 J, all of the specimens could
be completely broken; (b) when both KV2 and KV8
were above 200 J, the specimen occasionally could not
be completely broken; and (c) when the KV2 is above
300 J, the specimen frequently could not be completely
broken. It is believed that the energy fraction of
specimen bending and plastic deformation will increase
with the total energy increasing, and consequently the
large deformation causes the specimen not to be broken.
Meanwhile, the 8-mm striker will cause a much higher
bending and plastic energy dissipation due to the larger
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
234
edge radius. So, it is not surprising that the KV8
has a higher value and the difference between KV8
and KV2 increases with increasing energy/temperature.
Once the KV8 energy is greater than 400 J/cm2, the
CVN specimens cannot frequently be completely broken
in the test, and Fig.7 shows the comparison between
the unbroken specimens and the fully broken specimens.
In contrast, all of the DWTT specimens could be
fully broken. So, when the KV8 energy is greater than
400 J/cm2, the major energy dissipation in the CVN
specimens might not be the crack-propagation energy,
but the crack-initiation energy, plastic-deformation energy,
and friction dissipation between the specimen and the
anvils. It is believed that the real crack-propagation
resistance may not increase with increasing KV8. On
the other hand, since the fracture-propagation energy
is still the major part of the DWTT total energy,
the plateau in the DWTT total energy occurs at very
high energies for the Chinese X-80 steels. It is noted
that this observation is subject to further verification.
ot
f
Relationship between the 8-mm
striker CVN energy and the DWTT
energy
or
d
Fig.6. DWTT propagation energy
versus CVN energy (KV8).
Sa
m
pl
e
co
py
-n
25 specimens were tested by both the 8-mm striker
CVN test and the instrumented DWTT with the
standard pressed notch. As shown in Fig.5, the
DWTT total energy has a good linear relation with
the KV8 energy for the KV8 below 400 J/cm2. The
linear relationship can be expressed as DWTTtotal = 2.3
CVN + 63.04 (J/cm2), which is almost the same as
Wilkowski’s result of DWTTtotal = 2.53 CVN + 63.04
(J/cm2) [7]. In Wilkowski’s study, the maximum value
of KV8 is below 320 J/cm2, and thus only a linear
relation between the DWTT total energy and the KV8
energy was obtained. In contrast, the present study
has a maximum KV8 of 571 J/cm2, and so the initial
linear relationship, an overall non-linear relationship,
and a final plateau when the KV8 energy is above
400 J/cm2, are observed.
Furthermore, the DWTT-propagation energy is calculated
as the area under the load-displacement curve by
integration between the maximum load and a load
of 0.3 times the maximum load. As shown in Fig.6,
a good linear relation is again observed and can be
expressed as DWTTpropagation = 0.9 CVN + 30 for KV8
below 400 J/cm2. The slope in Fig.6 is lower than
that in Fig.5; it seems that the propagation energy
fraction in the DWTT total energy decreases with KV8
increasing, and thus the DWTT deformation energy
increases with the increase of the total Charpy energy.
Conclusion
More than 30 kinds of X80 LSAW pipe steels made
for the Chinese 2nd WEGP were experimentally
investigated by different test methods using the 2-mm
striker CVN test, the 8-mm striker CVN test, and the
instrumented DWTT. From the tests and discussions
presented in this work, the following conclusions
are obtained:
• The differences between KV2 and KV8 can
be ignored for Charpy impact energy below
200 J. However, when the impact energy is
greater than 200 J, the differences between
KV2 and KV8 increase with increasing energy,
and this is usually accompanied by the increase
of unbroken specimen occurrence.
• Both KV2 and KV8 have the similar change
trends with the change of test temperature.
235
is
t
rib
ut
io
n
3rd Quarter, 2013
co
py
-n
ot
f
or
d
(a)
Fig.7. Photos of fracture appearance
for (a) unbroken CVN specimens, and
(b) fully broken CVN specimens.
m
pl
e
(b)
Sa
• The DWTT total energy has a linear relationship
with the KV8 for the KV8 energy values
less below 400 J/cm2, and is described by
DWTTtotal = 2.3 CVN + 63.04 (J/cm2). However,
for KV8 energies above 400 J/cm2, the DWTT
total energy trends to reach a plateau.
• The DWTT propagation energy has a linear
relationship with the KV8 for the KV8 energy
values below 400 J/cm2 and is described
by DWTTpropagation = 0.9 CVN + 30 (J/cm2).
For KV8 energies above 400 J/cm2, the
DWTT propagation energy also trends to reach
a plateau.
Acknowledgments
This project was supported by Natural Science Basic
Research Plan in Shaanxi Province of China (Program
No. 2011JQ6017) and by China National Petroleum
Cooperation (CNPC).
References
1. ASTM E23-06. Standard test methods for notched
bar impact testing of metallic materials.
2. ISO 148-1:2006. Metallic materials – Charpy
pendulum impact test method.
236
The Journal of Pipeline Engineering
rib
ut
io
n
8. T.Naniwa, M.Shibaike, et al., 1990. Effects of the
striking edge radius on the Charpy impact test.
ASTM STP 1072, pp 67-80.
9. C.N.McCowan, J.Pauwels, G.Revise, and H.Nakano,
2000. International comparison of impact verification
programs. ASTM STP 1380, pp 73-89.
10.O.L.Towers, 1993. Effects of striker geometry on
Charpy results. Metal Construction, 15, 11, pp 682686.
11.T.A.Siewert and D.P.Vigliotti, 1995. The effect of
Charpy V-notch striker radius on the absorbed
energy. ASTM STP 1248, pp 140-152.
12.H.P.Li, X.Zhou, and W.C.Xu, 2011. Correlation
between Charpy absorbed energy using 2mm and
8mm striker [J]. J.ASTM International, 8, 9, 1-4.
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
3. GB/T 229-2007. Metallic materials – Charpy
pendulum impact test method.
4. R.K.Nansta and M.A.Sokolov, 1995. Charpy impact
test results on five materials and NIST verification
specimens using instrumented 2-mm and 8-mm
strikers. ASTM STP 1248, pp 111-139.
5. M.Tanaka, Y.Ohno, H.Horigome, et al., 19995.
Effects of the striking edge radius and asymmetrical
strikes on Charpy impact test results. Idem, pp
153-167.
6. G.M.Wilkowski, W.A.Maxey, and R.J.Eiber, 1978.
What does the Charpy test really tell us? American
Society for Metals, pp 201-225.
7. G.M.Wilkowski, D.L.Rudland, H.Xu, and
N.Sanderson, 2006. Effect of grade on ductile
fracture arrest criteria for gas pipelines. Proc.
International Pipeline Conference, Calgary, Paper
No: IPC2006-10350.
3rd Quarter, 2013
237
CTOD and pipelines: the past,
present, and future
by Dr Philippa Moore* and Dr Henryk Pisarski
io
n
TWI Ltd, Cambridge, UK
C
-n
ot
f
or
d
is
t
rib
ut
RACK-TIP-OPENING displacement (or CTOD) has been the most widely used fracture-toughness
parameter within the oil and gas industry for nearly 50 years. Originally developed from research at
TWI in the UK during the 1960s, CTOD was an ideal parameter for characterizing the fracture toughness
of medium-strength carbon-manganese steels used in pressure vessels, offshore platforms, and pipelines
where the application of linear-elastic fracture mechanics was insufficient to account for their ductility.
Once fracture-toughness testing (CTOD testing) became standardized within BS 7448, ASTM E1290, ISO
12135, and ISO 15653, the CTOD concept enjoyed an established international reputation.The development
of standardized fitness-for-service assessment procedures, initially through the use of the CTOD design
curve, and then to use of the failure-analysis diagram approach described in BS 7910, also allowed CTOD
to be used directly to determine tolerable flaw sizes to assess the structural integrity of welds. In more
recent times, single-edge-notched tension specimen (SENT) testing has been enthusiastically adopted by the
pipeline industry in place of the traditional single-edge-notched bend (SENB) specimen used for standard
CTOD tests. However, currently there is no national standard describing SENT testing, although this is being
developed. SENT testing is particularly advantageous when pipeline girth welds are subjected to plastic
straining, and a number of assessment procedures based on CTOD have been and are being developed to
define strain capacity and flaw-acceptance criteria.
C
Sa
m
pl
e
co
py
RACK-TIP-OPENING DISPLACEMENT (or CTOD)
has been the most widely used fracture toughness
parameter within the oil and gas industry for nearly
50 years. Originally developed from research at TWI
in the UK during the 1960s, CTOD was an ideal
parameter for characterizing the fracture toughness of
medium-strength carbon-manganese steels used in the
manufacture of pressure vessels where the application
of linear-elastic fracture mechanics was insufficient to
account for their ductility. The development of North
Sea oil and gas from the 1970s onwards hastened the
application of CTOD testing and analysis concepts for
application to the construction of steel-jacket production
platforms, and pipelines. The fracture-toughness testing
of single-edge-notched bend specimens (or the �CTOD
test’ as it is sometimes called) is the standard method
to measure it. However, as further progress is made in
the development of fracture mechanics, both testing and
assessment, the CTOD concept must change and adapt
to keep up. To illustrate the story of CTOD over the
years we have highlighted some of the key people at
TWI who have been involved in this work; however it
*Corresponding author’s contact details:
tel: +44 1223 899000
email: [email protected]
must be recognized that the research and development
of this field has been the collaboration of a much larger
group of engineers from many institutions worldwide.
The origins of the CTOD concept
Fracture mechanics as an engineering discipline was
conceived just after the Second World War as a result
of the Liberty Ship fractures. Of 2700 ships fabricated
using the new welding technology during the war, around
400 had fractures, 90 were considered serious, and about
ten ships fractured completely in half [1]. This failure
rate had driven the US Naval Research Labs to research
the effect of cracks in steels, and by the 1950s it had
developed the linear-elastic fracture mechanics’ (LEFM)
description of cracks in brittle materials in work led by
Dr George L Irwin [2]. However, the stress-based LEFM
did not sufficiently describe the behaviour of more ductile
materials, such as medium-strength structural steels.
The UK had chosen to begin its own investigations
into brittle-fracture issues after the War, driven by the
UK Admiralty Advisory Committee on Structural Steels,
who held conferences at the University of Cambridge in
1945 and 1959 attended by many of those who would
become eminent in the fields of structural engineering,
metallurgy, and fracture mechanics, including George
238
The Journal of Pipeline Engineering
io
n
In the case of the CTOD test, the specimen size is
usually representative of the full material thickness.
The CTOD test piece originally had a saw-cut notch
but later used fatigue pre-cracking to produce a sharp
notch [5]. The crack mouth is instrumented with a
clip gauge to measure the crack-mouth opening, and
then loaded under quasi-static three-point bending to
enable a load versus crack-mouth opening displacement
trace to be plotted (Figs 2 and 3).
Fig.1. Alan Wells at TWI, UK, in the 1990s.
Sa
m
pl
e
co
py
-n
ot
f
Irwin and Dr Alan Wells from the British Welding
Research Association [3]. Alan Wells (Fig.1) had taken a
sabbatical at the US Naval Research Lab in 1954 and
worked with George Irwin in that time. After his return
to the UK and the British Welding Research Association
(BWRA), which later became TWI, Alan Wells proposed
an alternative model of fracture to LEFM in 1961 [4].
Wells developed the crack-opening displacement (COD),
later the crack-tip-opening displacement (CTOD), model of
fracture mechanics from an observation of the movement
of the crack faces apart during plastic deformation of
notched test pieces. He showed that fracture would take
place at a critical value of COD, and for calculations
below general yield, this was proportional to the square
of the critical stress intensity factor divided by the yield
strength. Furthermore, he showed that the critical value
of COD determined in bend specimens and wide-plate
specimens (representing structural components) of the
same thickness were equivalent. Thus he was able to
demonstrate transferability of fracture toughness determined
from test specimens to other structural geometries. This
was to have far-reaching implications on the development
of fitness-for-service concepts for welded structures for the
avoidance of fracture. As a result of this, the CTOD
parameter was used extensively in the UK for elastic-plastic
fracture mechanics (EPFM) analysis of welded structures
from the 1960s especially once the development of North
Sea oil reserves in the 1970s was driving much of the
fracture research at that time [1].
or
d
is
t
rib
ut
As confidence grew in the ability of the small-scale CTOD
test to predict the fracture conditions of a crack in a
full-scale structure, the test method became standardized.
BSI published a Draft for Development (DD19) on applied
force for CTOD testing in 1972. This became a standard
in 1979 as BS 5762 [6], which described �Methods for
crack opening displacement (COD) testing’. This was then
superseded by BS 7448 Part 1 in 1991 [7] as a �Method
for determination of KIC, critical CTOD and critical J
values of metallic materials’. Part 2 of BS 7448 provided
an equivalent method for welds in metallic materials
when it was published in 1997 and was largely based on
TWI’s experience in testing welds. This was developed
further by an ISO committee, with TWI representation
by Henryk Pisarski (Fig.4), and in 2010 BS 7448
Part 2 was superseded by BS EN ISO 15653 [8],
although Part 1 still remains current for CTOD testing
of parent metals.
Standard methods for CTOD testing
In essence, the fracture-toughness test specimen comprises
a rectangular bar of material that is notched into the
appropriate region (with respect to a welded joint).
Once fracture-toughness testing became standardized
within BS 7448 [7], ASTM E1290 [9], E1820 [10], ISO
12135 [11], and ISO 15653 [8], the CTOD concept
enjoyed an established international reputation. CTOD
had been established as the fracture-toughness parameter
for the oil and gas industry so thoroughly, that often
the phrase �CTOD test’ has been used interchangeably
with the more precise �fracture-mechanics’ test’ by
those in that industry.
Definition of CTOD
In the days of development of the CTOD testing
standard BS 5762 within TWI, both Mike Dawes
(Fig.5) and Alan Wells put forward formulae to
determine CTOD from the test result, based respectively
on either the load and crack-mouth opening, or
the crack-mouth opening alone. The Dawes
approach [12] which combined separate elastic and
plastic components of the crack-tip-opening
displacement was that which was ultimately adopted
by the British Standard, and in the early editions of
ASTM E1290.
The equation to determine CTOD from bend specimens
in the current fracture-toughness-testing standards
ISO 12135 and BS 7448 Part 1 comprises an elastic
component and a plastic component which are added
together. The elastic part is based on the applied force
239
rib
ut
io
n
3rd Quarter, 2013
co
py
Fig.3. Example of a load vs crack-mouth
opening displacement (CMOD) trace
measured during a fracture-mechanics’
test (�CTOD test’) of a single-edge-notched
bend specimen.
-n
ot
f
or
d
is
t
Fig.2. Fracture-mechanics’ test (�CTOD test’)
of a single-edge-notched bend specimen
instrumented with a double-clip gauge.
Sa
m
pl
e
(F) and a function of initial crack length to specimen
width ratio (a0/W), as well as the specimen dimensions,
while the plastic component is determined using the
plastic component of the clip-gauge displacement (Vp)
and the height of the clip gauge above the crack mouth
(z), in addition to specimen dimensions. The current
CTOD equation in ISO 12135 and BS 7448 Part 1
for bend specimens is given in the following formula:
2
пЈ®пЈ« S пЈ¶
F
пЈ« a пЈ¶ пЈ№ пЈ® (1 в€’ОЅ ) пЈ№
пЈє
CTOD or Оґ = пЈЇпЈ¬ пЈ·
Г— f пЈ¬ 0 пЈ·пЈє пЈЇ
0.5
пЈ­ W пЈё пЈєпЈ» пЈЇпЈ° 2Пѓ YS E пЈєпЈ»
пЈЇпЈ°пЈ­ W пЈё ( BBNW )
0.4 (W в€’ a0 )V p
+
0.4W + 0.6a0 + z
2
(1)
where S is the loading span of the specimen, W is
the specimen width, B is the thickness, BN is the net
section thickness (accounting for side grooving), ОЅ is
Poisson’s ratio, and E is Young’s modulus.
The CTOD can also be related to the stress-intensity factor,
KI, using the following formula, given in BS 7910 [13]:
CTOD or Оґ =
K I2
(2)
mПѓ YS E '
where E’ is the Young’s modulus under plain strain
conditions, equal to E/(1-ОЅ2), and m is a geometric
and material factor which is usually between 1 and 2.
In the new editions of BS 7910 which is due to be
published in 2013, a more-precise value of m is given
which is a function of the ratio of material yield to
tensile strength for deeply notched specimens.
The crack-tip-opening displacement can be conceptually
understood as the amount that a crack tip needs to
be opened up (or the distance the crack faces need
to be moved apart) before unstable propagation of the
crack occurs (Fig.6a). However, several alternatives have
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
240
Fig.5. Mike Dawes at TWI in the 1980s.
been put forward as to how to exactly define CTOD,
illustrated in Fig.6. For materials with lower ductility,
the original definition (Fig.6a) is fine, but for moreductile materials, unstable fracture may not occur and
the CTOD is determined from the point of maximum
load in the test: the CTOD can be considered the
opening displacement of the deformed crack at the
tip position of the original crack (Fig.6b). A third
definition is that most often used when performing
numerical models of cracks, which defines the CTOD
as the displacement at the points where a 90В° angle
at the crack tip intersects with the crack sides (Fig.6c).
to be used to more accurately determine tolerable flaw
sizes to assess the structural integrity of welds. Published
standard fitness-for-service (FFS) assessment procedures
also cemented the power of fracture-mechanics’ testing.
Successful experience using the CTOD concept to
determine tolerable flaw sizes over almost a decade led to
the development of an FFS Published Document (PD 6493)
by the British Standards Institution. Initially published as
PD 6493:1980 [17] �Guidance on methods for assessing
the acceptability of flaws in fusion welded structures’,
the procedure was revised in 1991 and subsequently
became a standard, BS 7910 [13], in 1999.
-n
ot
f
or
d
Fig.4. Henryk Pisarski at TWI in 2013.
py
Fitness-for-service assessment
Sa
m
pl
e
co
One of the earliest codified applications of the CTOD
concept was to provide alternative flaw-acceptance criteria
to those in Appendix A of API 1104 [15] in the 1970s.
The Welding Institute’s CTOD design curve was developed
by Mike Dawes (Fig.5) and colleagues, at a time when
very little guidance was available on the application
of elastic-plastic fracture-mechanics’ (EPFM) analyses
to common materials, particularly to welded structures
with high residual stresses and stress concentrations.
The CTOD design curve [16] was intended to provide
a logical, simple, and rapid means of determining the
allowable crack sizes in welded structures subjected to
normal design loads. The importance of CTOD was
that it could be used directly to calculate the maximum
tolerable flaw size for a given weld.
The CTOD design curve approach was intended to
be used as a first, coarse, filter in fitness-for-purpose
assessments. The subsequent development of standardized
fitness-for-service assessment procedures, initially through
the use of the CTOD design curve and then to the use
of the failure-analysis diagram approach, allowed CTOD
FFS assessment allowed larger flaws to be shown
as tolerable, for example in offshore platforms and
pipelines, compared to the small flaw sizes permitted
by applying �workmanship’ flaw limitations imposed by
welding standards. At TWI, John Harrison (Fig.7) had
performed numerous FFS assessments using both the
CTOD design curve and PD6493, and demonstrated the
effectiveness of these methods to industry, particularly
for oil and gas [18]. He became heavily involved in the
standardization of PD 6493 into BS 7910. These FFS
methods were used for a number of offshore installations
in the 1970s and 1980s where it had been necessary
to demonstrate avoidance of brittle fracture in as-welded
joints in thicker section (40-120 mm). Similar to pressurevessel practice, highly stressed welds for operation at
sub-zero temperatures would normally be subjected to
post-weld heat treatment (PWHT) when section thicknesses
exceeded 40 mm, and avoiding PWHT needed this kind
of rigorous justification. These issues were addressed in
greater detail in the UK Department of Energy’s Guidance
Notes on the design, construction, and certification of
offshore structures, which was first published in 1990
[19]. This document had a large influence on the design
of steel jackets operating in the North Sea. The sections
241
(a)
(b)
(c)
io
n
3rd Quarter, 2013
rib
ut
Fig.6. Some definitions of CTOD [14]: (a) early idealization; (b) CTOD at original crack-tip position; (c) CTOD at positions
subtending 90В° at crack tip.
ot
f
or
d
is
t
dealing with toughness requirements for steels, avoidance
of brittle fracture, and post-weld heat treatment were
largely based on work conducted by John Harrison and
Henryk Pisarski [20], and utilized concepts described
in PD 6493. Where problems had arisen with flaws
having been detected and repair strategies needing to be
decided, an FFS assessment could justify whether repair
without subsequent PWHT was acceptable or not. With
the standard methods, performing an FFS assessment
became a regular part of the preparation for any new
pipeline installation in order to set the fabrication flawacceptance criteria.
SENT testing
co
py
-n
The desire to extend the application of pipelines to
ever-more-challenging loading conditions during both
during installation and operation has been driving the
development of fitness-for-service methods to reduce
their over-conservatism while remaining confident that
the structure will be safe. Pipelines intended for deeper
water, higher pressure, or installation methods or upset
conditions (such as ground movement) involving strains
beyond yield, impose greater challenges to integrity.
Sa
m
pl
e
The development of fracture-toughness testing had
traditionally been through the use of the deeplynotched bend (SENB) tests, which are intended to
impose a high degree of crack-tip constraint, and hence
provide a lower-bound estimate of fracture toughness.
Recognition that flaws in pipeline girth welds are
subjected to lower crack-tip constraint has led to the
development of the single-edge-notch tension (SENT) test
which has lower constraint than the SENB specimen.
Collaboration between Henryk Pisarski (Fig.6) at TWI
in the UK and SINTEF and DNV in Norway, along
with a group of industry partners, produced guidance
for fracture control for pipeline-installation methods
introducing cyclic plastic strain, which became DNV’s
Recommended Practice DNV-RP-F108 in 2006 [21]. The
method intended for pipe installation by methods such
as reeling used SENT specimens to measure fracture
toughness from a notched specimen whose constraint
more closely matched that of a flaw in a girth weld.
Fig.7. John Harrison at TWI in 2007.
The method used multiple-specimens to produce a J-R
curve which was then used in an assessment procedure
based on BS7910 to generate flaw-acceptance criteria.
The higher value of fracture toughness that can be
obtained from a SENT specimen compared to a SENB
specimen has led to a rapid growth of interest in using
these specimens for other fracture-mechanics’ assessment
and testing. SENT testing has been enthusiastically
adopted by the offshore pipeline industry (for example,
in DNV-OS-F101) and gradually been accepted by the
pipeline industry in general. However, the initial testing
procedures for SENT specimens gave the fracture toughness
exclusively in terms of J-R curves rather than CTOD.
Although developed for pipeline installation, there is
growing evidence that the biaxial-loading experience by
pipelines during operation may also exhibit similar ductiletearing resistance as the R-curve measured from SENT
testing [22], making SENT specimens appropriate for
analysis of pipeline operation as well. In its Appendix
A, the DNV-OS-F101 standard for submarine pipeline
The Journal of Pipeline Engineering
rib
ut
io
n
242
or
d
CTODmat ≤ 0.1 mm; X = 1.8
0.1 mm < CTODmat < 0.4 mm; X = 1.9 - CTODmat
CTODmat ≥ 0.4 mm; X = 1.5
CTODmat =
-n
CTOD from SENT tests
to be iteratively improved using the following clauses to
determine the parameter X.
ot
f
systems [23] describes a fracture-mechanics’ method to
determine the acceptability of flaws. The 2012 edition
requires the fracture-toughness testing to be performed
on SENT specimens, and expresses the fracture-toughness
requirements in terms of J. It allows the fracture toughness
to be expressed as CTOD only if the procedure for
calculating CTOD is demonstrated to be conservative.
is
t
Fig.8. Comparison of methods to
determine CTOD from J relative
to double-clip measurements of
CTOD for SENT specimens.
m
pl
e
co
py
Despite its growing popularity, there is currently no
standardized procedure for carrying out SENT tests, or
for determining CTOD in a SENT specimen. This gap
in CTOD knowledge led to a study by TWI to validate
methods for determining CTOD in the SENT specimen
[24]. The work compared direct measurements of CTOD
made by silicone-crack infiltration to a finite-element
model of the SENT specimen to predict CTOD, and
the double-clip gauge method to determine CTOD using
the equivalent triangles rule. These methods were also
compared to CTOD from J equations in recent literature
to improve upon the over-conservative equation given in
the 2010 edition of DNV-OS-F101 [25].
Sa
Using silicone-rubber crack infiltration allowed direct
measurement of CTOD to be made from replicas of
the SENT specimen notch, although the method is not
practical for routine testing. FEA models also give a
reliable way to determine CTOD, but require too much
analytical processing to be practical for determining
CTOD for every fracture-mechanics’ test. These methods
were used to compare the effectiveness of other simpler
methods for calculating CTOD.
DNV-OS-F101:2012 [23] now also gives a revised method
to calculate CTODmat from Jmat (Equn 3), which uses the
material yield strength (YS) and an estimate of CTOD
J mat
X .Пѓ YS (1 в€’ОЅ 2 )
(3)
Both the FEA model and the crack-infiltration methods
agreed fairly well with the double-clip method, giving
confidence that the double-clip method can give reliable
values of CTOD for SENT specimens with a/W ratios
of between 0.3 and 0.5.
When the equations for calculating CTOD from J for
SENT specimens were compared against the double-clip
(Fig.8), the one given by Shen and Tyson [26] offers
the best alternative method to calculate CTOD from
J compared to the over-conservative approach given in
DNV OS F101 from 2010. The Moreira and Donato
method [27] may show more benefit when applied
to weld specimens, not just parent metals. When the
newer method from DNV-OS-F101 from 2012 is also
included, the improvement in accuracy of the new
formula can be seen, but seems to be still consistently
slightly over-conservative.
The intention of this comparison was to provide
confidence in the value of CTOD that is determined
when using SENT-test specimens, so that the validation
of the SENT-specimen approach keeps pace with the
pipeline industry’s need to continue to define fracture
toughness in terms of CTOD, while using the most
modern test methods.
3rd Quarter, 2013
243
io
n
rib
ut
ot
f
Nevertheless, the CTOD fracture-toughness parameter is
being withdrawn from the fitness-for-service assessment
standard BS7910 in the 2013 edition, which will directly
use only K to determine the fracture ratio. However, the
intuitive understanding of CTOD as fracture toughness
by a wide range of industries will mean that it will
continue to be used to describe fracture toughness for
some time to come. Indeed, there is current discussion
on whether J or CTOD are the appropriate fracturecharacterizing parameters when considering assessments
in the plastic regime. For example, the strain-based
assessment procedure used by ExxonMobil uses CTOD
[26]. Pipeline-assessment research and development
continues towards strain-based methods and the CTOD
parameter is well suited for this application.
is
t
Further research into SENT testing with the intention of
developing a complete and thorough testing standard is
being carried out at TWI as part of a Group Sponsored
Project, while research on SENT testing is also active in
the US, Brazil, and Canada. The authors of this paper
are involved in the British Standards’ committee which
will eventually publish the SENT testing standard as BS
8571 in the near future.
5. P.Houldcroft, 1996. Fifty years of service to
industry – a brief and occasionally lighthearted
history of BWRA and The Welding Institute. TWI.
6. BS 5762:1979. Methods for crack opening displacement
(COD) testing. British Standards Institution.
7. BS 7448-1:1991. Fracture mechanics toughness tests:
Part 1: Method for determination of Kic, critical
CTOD and critical J values of metallic materials.
Idem.
8. BS EN ISO 15653:2010. Metallic materials - method
of test for the determination of quasistatic fracture
toughness of welds. Idem.
9. ASTM, 2008. ASTM E1290-08e1: Standard test
method for crack-tip opening displacement (CTOD)
fracture toughness measurement. (Withdrawn 2013),
American Society for Testing and Materials.
10.Ibid., 2011. ASTM E1820-11e1: Standard test method
for measurement of fracture toughness. Idem.
11.ISO 12135:2002, 2008. Metallic materials - unified
method of test for the determination of quasistatic
fracture toughness. International Standards
Organization.
12.M.G.Dawes, 1979. Elastic-plastic fracture toughness
based on COD and J-contour integral concepts. In:
Elastic-plastic fracture. ASTM STP 668, American
Society for Testing and Materials, pp 307-333.
13.BS 7910:2005. Guide to methods for assessing the
acceptability of flaws in metallic structures. British
Standards Institution.
14.J.D.Harrison, 1980. The “state of the art” in crack
tip opening displacement (CTOD) testing and analysis.
TWI Members Report 108/1980, April.
15.API 1104:1973. Standard for welding pipelines and
related facilities. 13th Edn (superseded), American
Petroleum Institute.
16.F.M.Burdekin and M.G.Dawes, 1971. Practical use of
linear elastic and yielding fracture mechanics with
particular reference to pressure vessels. In: Proc.
Institution of Mechanical Engineers Conference
on Practical Application of Fracture Mechanics to
Pressure Vessel Technology, London, 3-5 April, pp
28-37.
17.PD 6493:1980. Guidance on some methods for
the derivation of acceptance levels for defects in
fusion welded joints (superseded). British Standards
Institution.
J.D.Harrison, M.G.Dawes, G.L.Archer, and
18.
M.S.Kamath, 1979. The COD approach and its
application to welded structures. In: Elastic-plastic
fracture, ASTM STP 668, American Society for
Testing and Materials, pp 606-631.
UK Department of Energy, 1984 and 1990.
19.
Standard: Offshore installations: guidance on design,
construction and certification. 3rd and 4th Edns,
HMSO, London.
20.J.D.Harrison and H.G.Pisarski, 1986. Background to
new guidance on structural steel and steel construction
standards in offshore structures. HMSO, London.
or
d
The future
py
-n
TWI is proud of its history and involvement in the
story of CTOD, and of individual colleagues whose work
in those early years allowed CTOD to become such a
versatile fracture parameter, well established within the
oil and gas industry and beyond. The present is a time
of further changes in both fracture-mechanics’ testing
and assessment and standardization, but offers a bright
future for further research and development in this field,
not just at TWI, but worldwide.
co
Acknowledgements
pl
e
The authors are grateful to Dr Mike Dawes for providing
a photograph and some biography text for this paper,
despite his retirement.
m
References
Sa
1. T.L.Anderson, 1995. Fracture mechanics –
fundamentals and applications. 2nd Edn, CRC Press.
2. G.Irwin, 1957. Analysis of stresses and strains near
the end of a crack traversing a plate. J. Applied
Mechanics, 24, pp 361-364.
3. J.Knott, 1997. From CODs to CODES (the realisation
of fracture mechanics in the UK): Fracture research
in retrospect – an anniversary volume in honour of
R.George Irwin’s 90th Birthday, Ed. H.P.Rossmanith,
A.A.Balkema, Netherlands, ISBN 9054106794.
4. A.A.Wells, 1961. Unstable crack propagation in metals:
cleavage and fast fracture. Proc. Crack Propagation
Symposium, Cranfield, UK, 2, p 210.
244
The Journal of Pipeline Engineering
or
d
is
t
rib
ut
io
n
Dr Henryk Pisarski is a TWI Technology Fellow and
works in the Structural Integrity Technology Group
of TWI. He has been with TWI since 1973. His
interests include the application of fracture-mechanics’
based assessment and testing methods to assure the
integrity of welded structures with respect to fracture
avoidance and to demonstrate fitness-for-service. He
is also involved in the development of strain-based
procedures for the assessment of pipeline girth welds.
He has managed projects applying fracture-mechanics’
testing and assessment methods to a wide range of
engineering structures including ships, offshore structures,
subsea components, pipelines, and pressure vessels. In
addition, he has contributed to standards’ bodies on
fracture-toughness testing and flaw assessment (ISO
15653, BS 7448, and BS 7910). He has published a
number of papers on these subjects and also carried out
expert-witness work. Currently, he is the UK delegate to
IIW Commission X (Structural performance of welded
joints – fracture avoidance) and is contributing to the
revision of BS 7910 (flaw assessment).
Dr Mike G Dawes was involved in the development
of the CTOD approach to fracture avoidance at TWI
from 1968 until he retired in 2000. By that time his
CTOD design curve and fracture-toughness test method
relationships were used in all the corresponding BSI,
ASTM, and ISO standards. Much of his work, including
his PhD thesis, was concerned with applications to
welded-metal structures. This included development of
the local compression treatment, which for the first
time enabled acceptable shapes of fatigue pre-cracks
to be obtained in fracture-toughness-test specimens
extracted from as-welded joints. In 1993, in recognition
of his work on international standards, he received the
ASTM’s highest award, the Award of Merit, and the
title of Honorary Fellow.
-n
ot
f
21.DNV-RP-F108:2006. Recommended Practice F108:
Fracture control for pipeline installation methods
introducing cyclic plastic strain. January.
22. K.A.MacDonald, 2011. Fracture and fatigue of welded
joints and structures. Chapter 2 Constraint fracture
mechanics: test methods. By K.A.MacDonald, E.Ostby,
and B.Nyhus, Woodhead Publishing Ltd.
23.
DNV–OS-F101:2012. Offshore Standard F101:
Submarine pipeline systems. August.
24.P.Moore and H.Pisarski, 201. Validation of methods
to determine CTOD from SENT specimens. In:
Proc. ISOPE-2012, The 22nd International Offshore
(Ocean) and Polar Engineering Conference, Rhodes,
Greece, 17-22 June.
25.DNV-OS-F101:2010. Offshore Standard: Submarine
pipeline systems (superseded).
26.G.Shen and W.R.Tyson, 2009. Evaluation of CTOD
from J-integral for SE(T) specimens. In: Proc. Pipeline
Technology Conference, Ostend, 12-14 October.
27. F.Moreira and G.Donato, 2010. Estimation procedures
for J and CTOD fracture parameters experimental
evaluation using homogeneous and mismatched
clamped SE(T) specimens. In: Proc.ASME 2010
Pressure Vessels and Piping Conference. PVP2010,
Bellevue, Washington, USA, 18-22 July.
28.H.Tang, M.Macia, K.Minnaar, P.Gioielli, S.Kibey,
and D.Fairchild, 2010. Development of the SENT
test for strain-based design of welded pipelines. In:
Proc. 8th International Pipeline Conference, Calgary,
Canada, 27 Sept.-1 Oct.
Appendix – Biographies of TWI
engineers cited
Sa
m
pl
e
co
py
Dr Alan Wells died at the end of 2005. By the time
he formally retired from TWI in the summer of 1988,
Alan Wells had notched up 25 years at TWI’s site
in Great Abington, the last 11 as Director General.
His relationship with TWI, in the guise of the British
Welding Research Association, dated back to 1950, a
few months after completing his PhD in soil mechanics.
As a Fellow of the Royal Society, Dean of Faculty for
four of his years at Queen’s University in Belfast, a
recipient of the Order of the British Empire, and one
of the very few non-Americans to have contributed
to the US Navy’s brittle-fracture work, he believed his
career peaked relatively early. In 1954 he was able to
accept an invitation to work with Dr George Irwin at
the Naval Research Laboratory in Washington, DC.
That placement was key in his role in developing the
CTOD concept, but in addition Alan Wells also worked
to establish the Wells wide-plate test. Wells later spent
13 years as the Head of Civil Engineering at Queen’s
University Belfast before returning to TWI as Director
General. Upon retirement he was involved in several
major failure investigations including a fracture under
hydrostatic test of one of the Sizewell power station
boilers and the Westgate bridge collapse in Australia.
Dr John Harrison joined TWI in 1962, when he
worked on fatigue of welded structures, concentrating
in particular on the significance of weld defects, the
topic upon which he obtained his PhD from Cambridge
University. He then became responsible for TWI’s
fracture research, eventually being promoted Head of
Engineering Research, covering design engineering,
fatigue, fracture, and non-destructive testing. In 1985,
the Engineering and Materials Group was formed,
comprising the Design Engineering, Fatigue, Fracture,
Non-Destructive Testing, and Materials Departments,
and John was appointed Group Manager of a total of
90 staff. John Harrison has published over 50 papers
on fatigue and fracture and has been extensively
involved in national and international committees. In
1988 he resigned Chairmanship of Commission XIII
of the International Institute of Welding which deals
with the fatigue behaviour of welded structures and
components, a position which he held for 15 years.
He retired from TWI in 1998.
3rd Quarter, 2013
245
Use of curved-wide-plate (CWP)
data for the prediction of
girth-weld integrity
io
n
by Dr Rudi M Denys*, Dr Stijn HertelГ©, and Dr Antoon A Lefevre
rib
ut
Laboratorium Soete, Universiteit Gent (UGent), Gent, Belgium
A
E
ECA produces conservative estimates provided the input
data and the variability of these data are known. The
conservatism increases for defects in overmatched welds
[21,24,34,36].
ot
f
VEN WHEN THE most stringent welding procedures
are used, the occurrence of girth-weld defects is
unavoidable during the construction of a pipeline. The
acceptance or rejection of a detected defect can be
assessed using workmanship or fitness-for-service based
acceptance criteria [1-20]. However, the curved-wide-plate
(CWP) test is an alternative tool to determine the effect
of a defect on weld performance [21-36].
or
d
is
t
MULTITUDE OF interdependent material and geometric factors determine the response of a girth weld
containing a defect under installation or service loads. A carefully designed curved-wide-plate (CWP)
test enables a direct assessment of these factors. Consequently, the CWP test provides, as a predictor of
the failure conditions, a tempting tool to assess girth-weld integrity, establish material requirements, and
validate numerical models or fracture-mechanics’-based defect assessments. However, multi-disciplinary
skills are required to explore this potential. This paper outlines the CWP testing requirements and the
material data required to obtain representative information.The evolution, and the current and future roles,
of CWP testing are also discussed.
-n
The CWP test, developed in 1979 by the Soete Laboratory
at the UGent, Belgium, incorporates all factors affecting
weld integrity and allows for the determination of the
critical tensile load and strain capacity at fracture or
at load instability. Accordingly, the CWP test provides
a convenient tool to quantify the safety margin or the
ratio between the predicted and the critical defect size
involved in an ECA.
pl
e
co
py
For conventional stress-based designs (axial remote strain
< 0.5%)1, acceptance of a girth-weld defect is usually
assessed against conservative weld-quality or workmanship
(WMS) based go / no-go acceptance criteria. These
criteria have no scientific basis and do not account for
pipe grade, weld-strength mismatch, and wall-thickness
effects. However, the application of these criteria prevents
slipshod welding practices.
Sa
m
If the defect is outside the WMS limits, contemporary
welding standards allow that the rejected defect may be
acceptable when assessed on the basis of a performancebased or fitness-for-service acceptance criterion. A fitnessfor-service assessment, or an engineering-critical assessment
(ECA), uses fracture-mechanics’-based analytical equations
and / or a plastic collapse criterion to determine the
allowable dimensions of a planar defect. By design, an
The 0.5% strain limit coincides with the strain at which the value of the
specified minimum yield strength (SMYS) is defined.
1
*Author’s contact details:
email: [email protected]
After a brief discussion on the conservatism of ECA
methods, the next sections are focused on the historical
and future role of CWP testing in material selection
and defect acceptance. Further, a brief review of the
key factors to be considered in CWP testing is given.
In this context, many of useful papers dealing with the
details of the subject matter have been published. For
space reasons, however, only a selected number of these
papers is cited.
Conservatism of ECA methods
The quest for perfecting the accuracy of the predicted
allowable defect limits has led to the development
of a multiplicity of ECA methods. In particular, the
pipeline industry can make use of both generic [9-12]
and pipeline-specific ECA methods [13-20] which give
different allowable defect sizes as the provisions, the
assumptions, and the treatment of the input data differ
246
The Journal of Pipeline Engineering
rib
ut
io
n
a. the predicted dimensions when the test is intended
to determine the safety margin;
b. the worst-case dimensions that could be
encountered or escape detection; or
c. for strain-based designs, the test performance
criterion.
is
t
For stress-based designs, it is recommended to select defect
dimensions ensuring the onset of pipe yielding. Note also
that the effects of weld reinforcement, weld-joint design
(shape of the weld bevel), and weld misalignment on
test performance are directly accounted for in the test.
The test is conducted at the minimum design temperature
until failure occurs in the notched section or by necking
in the weakest pipe. When the toughness of the notch
region shows transitional behaviour at the minimum
design temperature, a lower test temperature is used to
ensure lower-bound results. Note that specimen cooling
needs due attention [48-49].
ot
f
If traditional ECA methodologies are used as a
design tool, the user has to accept that the inherent
conservatism of an ECA may produce over-conservative
requirements which may lead to difficulties in weldingprocedure qualification testing. In addition, it may
also be a challenge to obtain and ensure the required
toughness for higher-grade pipe in production welds if
the assessment is uniquely focused on the three-factor
relationship between toughness, applied stress, and
defect size. This problem can be solved if the many
more interdependent factors affecting the complex
relationship between crack-driving force and fracture
resistance or fracture toughness are accounted for in
the assessment. For example, the crack-driving force
can be reduced by using overmatched weld metals
[47-48]. In this context, it should also be noted that
the latest ECA updates are mainly based on numerical
simulations involving material models which do not
necessarily represent the stress-strain characteristics of
contemporary high-strength pipe materials. Therefore,
the veracity of the assumptions for modern high-strength
pipeline materials also requires more experimental
validation.
Depending on the properties of the weld region, the
notch is placed at the weld-metal centreline or along
the fusion line (HAZ) in either the weld root or cap.
Multiple notched specimens can also be tested. The
selection of the notch dimensions can be based on:
or
d
[37-46]. Consequently, there exists no generally accepted
ECA. This situation confuses potential users, and in
particular the regulatory bodies.
co
CWP test
py
-n
As no single ECA provides information on the safety
margin, ECA defect assessments do not necessarily
produce the most economic solution. To overcome
this limitation, the CWP test, as a predictor of the
failure conditions, is considered to be fit for setting
allowable defects for a pre-determined level of safety.
The CWP test can also be considered as a necessary
step to rationalize existing ECA methods.
Sa
m
pl
e
The CWP test involves an axially tensile-loaded curved
(un-flattened) girth-welded pipe segment containing
either a real defect or an artificial circumference
surface-breaking crack-like notch1 at mid-length along
the weld. As discussed in a later section, the test
incorporates all influential factors intervening in the
fracture/failure process. The overall dimensions of the
test specimen are typically 1200 mm (axial length) by
420 mm (circumferential direction). Comparison of CWP
and full-scale pipe bend-test data has demonstrated that
testing of an axially tensile-loaded pipe segment with
an arc length of 300 mm (or about 10% of the pipe
circumference in the gauge section) is a conservative
means to determine the failure characteristics of
defective girth welds in a large-diameter ( > 36-in)
pipeline [23,27].
In CWP testing, the term �notch’ represents a real defect, crack, flaw, or
any other discontinuity.
2
To obtain structurally representative results, the CWP
specimen must also be carefully instrumented. The analysis
of the test results is a further challenge, since the effect
of the multiple interdependent factors on the test results
have to be accounted for [51-61]. Hence, users not well
versed in the factors affecting girth-weld performance might
unintentionally either produce structurally irrelevant CWP
test data or not fully explore the potential of a CWP test.
CWP testing
Since its inception in 1979, UGent uses the CWP
test to study girth-weld integrity for both stress-based
and strain-based (axial remote strain > 0.5%) designs.
At present, more than 1100 CWP tests on old-vintage
low-strength and modern high-strength pipe grades up
to API 5L X-120 have been conducted [32].
Historical developments in CWP testing
The results of the early-1980s CWP tests demonstrated that:
a. low fracture toughness (CTOD < 0.05 mm)
does not automatically disqualify girth welds
overmatched in strength;
b. the first-generation, CTOD-design-curve based,
ECA defect-acceptance methods over-estimate
the effect of toughness on defect tolerance; and
c. toughness, applied stress, and defect size are
not the only factors determining weld integrity
[2-7, 21,24].
3rd Quarter, 2013
247
CWP tests on matched welds also illustrated that the
highly constrained standard three-point bend CTOD
toughness test underestimates the fracture resistance of
tensile loaded girth welds [24]. In 1992, CWP tests
on girth welds of moderate toughness confirmed that
the workmanship limits for stress-based designs are
very safe because CWP failure occurred after pipe
yielding [22].
Since 2001, CWP instrumentation and material-testing
requirements have gradually been fine-tuned to cover
these issues [52-61].
Current use of CWP testing for stress-based designs
Nowadays, the CWP test is applied to:
a. determine the safety margin implied by the
commonly used ECA defect assessments for
stress-based designs;
b. study factors affecting girth-weld integrity
that cannot simply be modelled (such as
tearing behaviour, weld-strength mismatch, and
heterogeneity;
c. determine the axial strain capacity of a girth weld
with postulated defect dimensions [21-36]; and
d. optimize material requirements for both stressand strain-based designs.
io
n
In the mid 1990s, using the results of 186 CWP tests,
UGent developed, on behalf of EPRG:
co
py
is
t
-n
ot
f
The analysis of 93 multiple-notched CWP specimens
in 2005 led to the development of girth-weld-specific
defect-interaction criteria which are now incorporated in
the recently revised 2013 ERPG guidelines [29]. These
guidelines also include the defect length-acceptance limits
for 4- and 5-mm deep defects which can be applied
for pipe grades up to X-80 [14]. It is of interest to
note that the application of the EPRG defect-acceptance
criteria is less complicated than the ECA methods
using CTOD or J as fracture-resistance parameter [9-12,
15-20]. In 2011, 480 CWP test results were used to
develop, through a lower-bound curve-fitting, the UGent
defect-acceptance model for strain-based designs [62-63].
Furthermore, UGent also performs low-temperature
(-50/-60В°C) CWP tests to quantify the defect sizing
capabilities of AUT inspection [28,31]. In contrast to
the classical �salami’ technique, the low-temperature
CWP test provides a direct visual access to the whole
defect and prevents the actual defect size from being
incorrectly estimated. In addition, the low-temperature
CWP test allows the assessment of the severity of
natural defects at lower-shelf toughness conditions
[28,31,36].
or
d
CWP tests on 150-mm wide specimens extracted from
thin-walled (5-mm) small-diameter pipe were used to
adapt the ECA-based defect-acceptance procedures for
small-diameter pipelines in the Australian pipeline
code [26].
rib
ut
a. the world-wide accepted EPRG 30/40-J Charpy-V
girth-weld toughness requirement where the axial
(tensile and bending) strain is less than or equal
to 0.5%; and
b. the EPRG-Tier 2 length limit for 3-mm deep
defects in pipe grades up to API 5L X-70 [13-14].
m
pl
e
The results of the EPRG studies, the introduction of
high-strength pipeline steels, and the increasing demand
to study girth-weld integrity for strain-based designs,
paved the way for the world-wide use of the CWP
test [25, 30-36]. These research efforts disclosed that:
Sa
a. the level of weld-metal strength mismatch has a
strong effect on the crack driving force;
b. the pipes at either side of a girth weld have
different tensile properties;
c. strain-hardening capacity as measured by the
Y/T ratio has a strong effect on ductile tearing
behaviour and weld performance; and
d. the correct interpretation of CWP test
performance of specimens extracted from
production welds or high-strength pipes requires
significantly more pipe material and weld metal
tensile testing than specified in existing material
testing standards [49].
In contrast with an ECA assessment, wide-plate-test
results allow for the determination of the critical
conditions and the selection of the desired safety
factor. Therefore, CWP testing is increasingly applied
to formulate tailor-made weld-defect-acceptance criteria
when production girth welds fail to meet the specified
toughness or strength mismatch requirements, and an
ECA, using the material properties measured during
weld qualification or production testing, produces unduly
restrictive defect limits. In addition, the decision for
conducting CWP testing during construction is also
driven by either the costs involved by production
delays or the costs associated with the re-qualification
of the materials until satisfactory results are obtained.
CWP testing and strain-based design
Numerical studies and pressurized (bi-axially loaded)
full-scale tension (FST) tests have shown that internal
pressure causes a higher axial stress for an equal axial
strain [64-70]. This effect elevates the crack-driving force
and reduces the strain capacity in the post-yield loading
range. This means that the FST is the most suitable
test to assess defect acceptance for strain-based designs.
However, FST testing is expensive and, moreover, the
material properties controlling the crack-driving force in
the FST test have to be derived from a dummy weld.
The Journal of Pipeline Engineering
Fig.1.Typical sampling
plan (not to scale).
or
d
The UGent guidelines provide information on the
specimen design, instrumentation, testing requirements,
and the material properties that need to be measured
for the interpretation of the CWP failure characteristics.
The guidelines illustrate that the CWP test cannot be
standardized in the same way as, for example, a CTOD
test. Unlike the CTOD test, many more factors affect
CWP test performance, among which are:
ot
f
Even if the dummy weld is made with pipe pup pieces
sampled from the same (parent) pipe and the same
welding procedure, the inherent weld-to-weld variability
makes it difficult to obtain the required information
with sufficient precision. The problem is that the
actual material properties determine test performance
in the post-yield loading range [52,67,70].
is
t
rib
ut
io
n
248
Sa
m
pl
e
co
py
-n
Since, leaving aside the difference in crack-driving force,
the CWP and FST test performance are controlled
by the same factors, it is recommended to use the
(uni-axially loaded) CWP test for strain-based designs.
However, this option requires the internal pressure
effect to be accounted for in the translation of the test
results. For matching welds, the measured strains have
to be reduced by a factor of two [62]; the correction
decreases for overmatched welds [63]. Besides the
practical and economic advantages, CWP testing allows
the pipe material and weld-metal properties affecting
CWP strain capacity to be obtained from test specimens
taken out adjacent to the CWP specimen. Because of
the better accessibility of the weld-root region, it is
also easier to place the flaw/notch tip in the target
weld metal (WM) and HAZ microstructure. Considering
the variability of the material properties around the
circumference, and since several CWP specimens can
be removed from a single weld, it is also possible to
conduct a sensitivity analysis.
Guidance for CWP testing
Using UGent’s vast experience in CWP testing for
both stress-based and strain-based designs, the authors
developed the UGent guidelines for CWP testing in
2009 [49]. Currently, DOT-PHMSA and PRCI are also
preparing a CWP testing procedure, although to date
there is no generally accepted standard for CWP testing.
• notch location and dimensions (length and
height);
• ratio of notch depth to wall thickness;
• geometric aspects such a pipe-wall thickness
variations, high-low weld misalignment, weldbevel geometry, and weld reinforcement;
• toughness properties of the notched region;
• post-yield stress-strain responses (strain-hardening
characteristics, Y/T ratio, etc.) of the materials
in the notched region and the (remote) pipe
sections;
• uniform strain (uEL) capacity of the pipe metal;
• ductile-tearing resistance of the material in the
notched region;
• level of yield, flow, and tensile strength mismatch.
In the following, the UGent CWP testing guidelines and
the recent modifications that require due considerations
are briefly discussed.
Material characterization and specimen sampling
Literature on CWP testing reveals that different CWP
test specimen dimensions are used [58], and the
differences in instrumentation complicate the comparison
of published CWP results. Added to this, the material
properties controlling CWP performance are either
249
is
t
rib
ut
io
n
3rd Quarter, 2013
Fig.2. CWP dimensions and instrumentation layout for deformation measurements.
or
d
specimen. The minimum length of the prismatic test
section is 3W, with the girth weld at mid-length. This
length avoids the yielding bands emanating from the
notch tips and the specimens ends affecting the strain
measurements in the pipe body.
ot
f
not reported or poorly documented. The issue is
that the tensile and toughness properties vary in the
circumferential direction [6, 10-11]. This explains, for
example, why the interpretation of CWP results based
on material properties inferred at a distance away from
the CWP specimen may lead to misleading conclusions.
Consequently, the pipe-material and weld-metal properties
in the immediate vicinity of the CWP specimen need
to be known in the evaluation of CWP test results.
That is, material characterization in terms of tensile,
hardness, and toughness cannot be derived from the
testing of randomly sampled specimens. This problem
can be simply solved when the small-scale material
characterization testing samples are taken out adjacent
to the CWP specimen (Fig.1).
Medium CWP specimens
py
-n
As discussed, W is normally equal to 300 mm, and
this arc length ensures that the CWP test conservatively
models the failure behaviour of girth welds in a largediameter pipe under bending loads. For some users, the
load capacity of the test machine is a limiting factor
for the testing of 300-mm wide specimens. When a
medium, narrow, sub-sized, or �mini’ CWP specimen
is tested, the test provides a distorted picture of the
strain capacity and tearing behaviour occurring in wider
specimens. Decreasing specimen width causes an increase
of the net-to-gross section stress ratio with the result
that the critical condition in a narrow specimen occurs
at a lower level of applied load, applied strain, and
CMOD, while the onset of ductile tearing occurs at a
higher level of applied (gross) stress. However, medium
CWP specimens provide structurally relevant information
for small-diameter pipe when the arc length (width) of
the prismatic section meets the 10%-circumference rule.
m
pl
e
co
Pipes of different tensile properties are welded in
a random order in the pipeline string, and so it is
incorrect to assume in a CWP test assessment that the
pipes on either side of the girth weld have the same
or very comparable tensile properties. As shown in
Fig.1, the pipe-material tensile properties at either side
of the girth weld must also be determined.
Sa
CWP test specimen
For large-diameter transmission pipelines, the following
practical guidelines can be used to generate structurally
representative CWP test results.
Specimen dimensions
The nominal dimensions of the �standard UGent’
dog-bone-shaped CWP specimen are shown in Fig.2.
The minimum CWP dimensions are 1.4W (loading
pull-tabs) by 4W (overall specimen length), where W
is the width (arc length) of the prismatic part of the
Notch location
The notch is oriented parallel to the welding direction.
In the region of the notch, the weld reinforcement is
ground flush to facilitate the accurate placement of the
notch. For HAZ testing, the notch can be placed in
the HAZ of either the strongest or weakest pipe. As
discussed in a later section, hardness testing can serve
to assist notch-tip placement. Similarly, when significant
wall-thickness variations exist, the HAZ of the thinnest
The Journal of Pipeline Engineering
tests showed that for notch-area ratios smaller than 7%
failure occurs by plastic collapse. For such toughness
levels, crack tip blunting causes a fatigue crack to act in
a similar manner to a machined notch. Consequently,
when the CVN toughness of the notch region meets
this threshold, a machined notch is used, made with
a thin (0.15/0.20-mm wide) jewellers’ cutting wheel or
by electro-erosion.
ot
f
or thickest pipe can be tested. When the HAZ of the
strongest (and/or thickest) pipe is tested, the effect of
the minimum level of weld-strength overmatch on strain
capacity is evaluated. One can also test the HAZ/weldmetal interface of the weakest (thinnest) pipe, as the
highest deformations will be concentrated in this area.
Fig.3. Digital-image
correlation picture
of longitudinal strain
distribution in a
field weld.
or
d
is
t
rib
ut
io
n
250
-n
Notch dimensions
Sa
m
pl
e
co
py
The selection of the notch dimensions (height and
length) are based on either the allowable defect size
as predicted by a fracture-mechanics’-based assessment,
the worst-case weld anomaly that might occur, or the
maximum flaw dimensions that could escape detection
by the inspection method used. When the latter two
options are used, the non-destructive inspection tolerance
is added to the selected notch dimensions. Further,
it is of interest to mention that a remote strain of
at least 0.5% will be obtained for notch area ratios
smaller than 7% of the load-bearing cross-sectional
area provided the threshold notch toughness of 30 J
(min.) / 40 J (average) is met [13,51-52]. The 7% limit
assumes matched or overmatched welds and applies for
pipe metal Y/T ratios not exceeding 0.90. For Y/T
ratios greater than 0.90, the notch-area ratio must be
reduced, while for overmatched welds, larger notch-area
ratios can be tested. Moderate to ample cap/root weld
reinforcement in thin-wall pipe has a similar effect as
weld-strength overmatch, and thus equally allows for a
relaxation of the 7% limit.
Notch-tip acuity
The early CWP tests were carried out with a pre-fatigued
sharpened notch. Provided that the Charpy toughness
(CVN) of the notch area exceeds the 30/40-J threshold,
Deformation measurements
To capture the effects of the strength differences of
the neighbouring pipes, and the possible wall-thickness
differences (which can vary up to 1.5mm in the pipeline),
the overall remote axial strain (gauge length: l0) as well
as the strains of the pipe sections at either side of the
girth weld (gauge lengths lA and lB, Fig.2) are measured.
For room-temperature CWP tests, a more comprehensive
picture of the complex strain distribution occurring
in a CWP specimen / field weld can be obtained
by means of digital-image correlation (DIC), Fig.3
[60], which illustrates that a relatively small difference
in yield strength (22 MPa in the case shown) can
cause very large differences in remote strains; the
weaker pipe strains three times more than the
stronger pipe.
The crack-mouth-opening displacement (CMOD) is
monitored by a clip-on measuring device which follows
the relative displacement of two steel pins straddling the
notch at mid-length. However, CMOD is not directly
comparable with CTOD, as measured in a CTOD bend
test. For an experienced eye, a CMOD-elongation/strain
plot provides a simple means to explain the effects on
strain capacity of toughness, weld-strength mismatch,
251
or
d
Curve A (failure mode: net section collapse) – undermatched girth weld
is
t
Fig.4. Dependence of the CMOD on remote strain and weld-strength mismatch:
rib
ut
io
n
3rd Quarter, 2013
Curves B through F (failure after remote yielding)
M = level yield of strength mismatch
overmatched welds: MF > ME > MD > MC > MB
ot
f
Curve F: CWP specimen failed by necking in the pipe.
Note:The pipe material in girth weld E exhibited discontinuous yielding (LГјders plateau).
Using established relationships between hardness and
tensile strength, hardness-test results allow a quick
screening to determine the possible strength differences
between adjacent pipe metals, to identify the weakest
pipe, and to obtain an estimate of the level of tensile
strength mismatch, and thus to verify whether the girth
weld is under- or overmatched.
py
-n
and other factors. For example, Fig.4 shows the CMOD
remote-strain responses of CWP specimens containing an
identical notch where the girth welds were made with
consumables of different strengths.
These records provide experimental evidence that:
Sa
m
pl
e
co
a. the relationship between toughness in terms of
CMOD and CWP strain capacity strongly depends
on the level of strength mismatch;
b. strength mismatch has a clear effect on strain
capacity;
c. reliance on toughness alone as the quantifying
parameter in a defect assessment can disqualify
acceptable welds; and
d. toughness requirements can be related to weldstrength mismatch (overmatched welds require less
toughness than matched welds). In this respect,
note that the beneficial strength mismatch effects
on strain capacity gradually break down with
increasing defect size.
Macrosectioning and hardness testing
Transverse-weld cross section(s) are extracted to illustrate
the configuration and sequence of the weld runs within
the girth weld, to facilitate notch-tip placement in the
CWP specimen and to perform hardness tests.
Hardness testing is performed according to current
welding-qualification standards. Using a 5-kg pyramiddiamond indenter (HV5), the indentations are made
along traverses both at the cap and root side, 2 mm
below the pipe surface in the unaffected pipe material,
the weld metal, and the HAZ at either side of the
girth weld. Alternatively, a hardness map provides a far
better picture of the hardness variation across the girth
weld, Fig.5: a typical hardness map involves between
700 and 1200 HV5 measurements.
The girth-weld hardness maps shown in Fig.5 provide
information with respect to the strength variations in
the pipe materials, the heat-affected zones, and the weld
metal. For example, both parts of the figure illustrate that
the pipes have different tensile properties, and that the
weld metal is matched (left) and undermatched (right)
in tensile strength. Also, Fig.5b shows that, because
of the large hardness differences in the through-wall
direction, the root area in the SMAW weld is either
The Journal of Pipeline Engineering
ot
f
or
d
is
t
rib
ut
io
n
252
-n
Fig.5. Macrosections and corresponding hardness maps of a GMAW and SMAW girth weld.
py
matched (right) or undermatched in tensile strength (left),
whereas the weld cap is overmatched. Furthermore, for
the GMAW weld, the HAZ regions are overmatched
to both the pipe materials and the girth weld (Fig.5a).
co
Material tensile properties
Sa
m
pl
e
Unless the CVN toughness of the weld region is lower
than the 30/40-J threshold, the CWP specimen usually
fails by plastic collapse in either the notch section or by
necking in the soft pipe. In this case, test performance is
controlled by the axial tensile properties of the girth weld
and the adjacent pipe pup pieces. The quantification of
these failure modes requires the stress-strain response and
the strain-hardening rate or Y/T ratio of the materials. The
information is also essential to facilitate the quantification
of the level of yield, flow, and tensile-strength mismatch
[47-48]. The percentage elongation at the uniform strain
of the pipe materials, uEL, is another essential factor
because it relates to the achievable strain capacity.
Pipe-metal tensile tests
Full-thickness longitudinal specimens, extracted from both
pup pieces in the axial direction, are tested. Round-bar
(RB) specimens may not be used because the material
is taken from the mid-wall thickness location so that
the �stronger’ inner and outer pipe surfaces are not
tested. The use of full-thickness specimens can avoid an
overestimated level of strength mismatch. In addition,
since pipes in the coating-aged condition can exhibit
higher yield strength and Y/T ratio values and lower
uniform elongations than bare – uncoated – pipes, the
test coupons should have undergone a thermal cycle as
used in the plant and/or field coating [69].
All-weld-metal tensile tests
The tensile properties of the girth weld are determined
by either all-weld-metal RB or rectangular specimens.
It must also be ensured that the parallel length of the
specimen consists entirely of weld metal corresponding to
a particular extraction position. Hardness measurements
can be used to assist the placement of RB specimens
in the through-thickness direction.
Practical considerations
The variation of the tensile properties of the pipe in
the axial and circumferential directions is often neglected
in the assessment of CWP test results. The associated
difference in post-yield stress-strain response is another
253
rib
ut
io
n
3rd Quarter, 2013
Fig.6.The effect of sampling position on weld yield strength and Y/T ratio of an undermatched weld (specimens were taken
from the 9 and 10 o’clock positions).
or
d
is
t
threshold by a factor of two, it is considered that the
effect of sampling location on toughness does not need
further consideration.
When the CWP test is aimed at the validation of
numerical or analytical defect assessments, complementary
fracture-mechanics’ tests are performed to determine the
fracture toughness in terms of K, J, and/or CTOD.
When ductile tearing intervenes in the failure process,
other fracture-mechanics’ parameters such as the R-curve
can be used to characterize material toughness. For girthweld defects, the SENT test is a better tool than the
SENB test to estimate the onset of stable tearing and
tearing resistance under tensile load. The single-specimen
R-curve testing requires measurements of crack growth
by means of techniques such as unloading compliance
or (direct-current) potential drop. Similar crack-growth
measurements can be performed on CWP tests [71-72].
co
Toughness testing
py
-n
ot
f
significant factor requiring consideration [48,55,57]. The
scatter of the all-weld (AW) metal tensile properties is an
even more complicated issue. Aside from the variations
around the circumference, the microstructural variations
within the weld deposit or through-thickness direction
make it more difficult to determine representative AW
tensile properties [48]. The nature of this variation also
depends on the welding process. For SMAW welds, the
tensile properties of the weld root are lower than those
of the fill passes. The weld passes in the root region
of narrow-gap GMAW welds are stronger than the
fill and cap passes. Consequently, specimen geometry,
sampling location, and specimen dimensions affect the
measured AW tensile properties [36,48]. Therefore, a
post-test verification of the microstructure(s) sampled by
AW specimen is recommended when significant strength
variations are measured, Fig.6.
Sa
m
pl
e
Toughness testing can be limited to Charpy V-notch
(CVN) toughness testing because correlations between
CWP performance and CVN impact energy have shown
that the CVN test is suitable for predicting the expected
CWP failure behaviour [51-52]. As discussed, failure by
plastic is ensured if the toughness of the weld region
exceeds the 30/40-J threshold. However, SENT tests are
recommended to assess the tearing resistance of high
Y/T ratio welds.
In the material-screening phase, CVN toughness testing
is concentrated on the girth-weld centreline and the
fusion line of the low-strength pipe, as the highest plastic
strains in CWP testing occur in this area. Unless the
weld cap requires a specific assessment, it is a standard
practice to focus testing on the weld root.
The CVN impact test is conducted at the minimum
design temperature using full-size standard (10 x 10 mm2)
test pieces. If the CVN properties exceed the 30/40-J
CWP test performance criteria
Upon completion of the test, the overall and remote
strains at specimen failure are either compared to a predetermined performance criterion or used as input in
a fracture-mechanics’-based assessment (ECA validation).
After the test, the notched region and the fracture surfaces
of the broken specimens are subjected to a fractographic
examination. The broken as well as unbroken HAZ
specimens are then sectioned for post-test metallography
to identify the microstructure sampled by the notch tip.
Strain criterion
Unless the CWP data are used to validate fracturemechanics’ methods, it is safe for stress-based designs
to require that the remote longitudinal strain meets
or exceeds the 0.5% level [13,14]. In this context,
the 0.5% criterion has been applied to establish the
Charpy requirements of pressure-vessel steels [73]. The
rationale for adopting the 0.5% performance criterion
The Journal of Pipeline Engineering
rib
ut
io
n
254
-n
Post-test fractography and metallography
or
d
circumferential surface-breaking defect(s) for both stressbased and strain-based designs. Since the CWP test
captures the influential factors affecting weld performance,
the test data play a useful role in identifying the
limitations inherent in the commonly used ECAbased defect assessments and associated material-test
requirements. At this time, CWP testing of a carefully
designed and instrumented specimen is often used as
a reliable tool for developing tailor-made weld-defectacceptance criteria, which account for the specified
girth-weld performance requirements and the available
material properties. However, for strain-based designs,
the effect of internal pressure on the crack-driving force
must be accounted for by applying a correction factor.
ot
f
is both to provide an adequate margin of safety for
conventional stress-based pipeline designs, and also to
exclude toughness-dependent fracture. As discussed, for
strain-based designs, the CWP test does not capture the
internal pressure effects on the crack-driving force. For
matched welds, this deficiency can be accounted for by
reducing the measured strains by a factor of two; this
factor can be reduced for overmatched welds [49-50].
is
t
Fig.7. Cross sections of valid and invalid notch locations in CWP specimens which failed by pipe necking. Note that significant
blunting occurred prior to failure.
pl
e
co
py
For broken specimens, the initial notch geometry,
the extent of blunting and ductile tearing, the crack
dimensions at failure, and the fracture faces are evaluated
under a stereomicroscope. In addition, macrographic and
micrographic examinations of HAZ-notched specimens
are performed to verify whether the notch tip effectively
intercepted the target microstructure (validity check), and
to evaluate the fracture path in the through-thickness
direction relative to the fusion boundary. The photographs
in Fig.7 illustrate the locations of the notch tips in two
unbroken CWP specimens.
Sa
m
UGent’s experience has shown that the CWP test
results of CGHAZ / fusion-line notched specimens are
�metallurgically valid’ when the distance between the
target microstructure (the fusion line) and the original
(machined or pre-fatigued) notch tip is smaller than
0.50mm (Fig.7a). When this requirement is not achieved
(Fig.7b), engineering judgment is needed to quantify the
practical significance of the CWP test result.
Concluding remarks
This paper gives a brief overview of the developments
and potential of CWP testing to assess the integrity
of girth welds containing either a single or a multiple
The correct interpretation of CWP test results requires
that the material properties controlling test performance
are known. In particular, the material characterization
in the vicinity of the CWP specimen is critical to
the assessment of CWP test results. The issue is that
the quantification of the pipe material and weld-metal
properties requires significantly more testing than specified
in existing material-testing standards. Concluding matters,
the CWP test is not a routine test. Its execution is
complex and the analysis and understanding of the
test results requires a multi-disciplinary expertise and
a wide range of skills because multiple interdependent
factors influence the CWP test performance.
References
1. R.Denys, 1999. ECA-based girth weld defect acceptance
levels. Pipeline & Gas Journal, Dec., 226, 9, p 60, Part I.
2. E.L.von Rosenberg, 1980. Alternative girth weld
defect assessment criteria for pipelines. Proc. Conf.
on Pipeline and Energy Plant and Piping, pp 319328. Permagon Press.
3rd Quarter, 2013
255
or
d
is
t
rib
ut
io
n
23.P.Hopkins, G.Demofonti, G.Knauf, and R.Denys,
1991. An experimental appraisal of the significance
of defects in pipeline girth welds. Proc. EPRG/
NG18 8th Biennal JTM on Line Pipe Research,
Paris, May 14-17, paper 20.
24.R.Denys, 1993. Fracture behavior of large diameter
pipeline girth welds: effect of weld metal yield
strength – Volume II. AGA Catalog No. 202-922.
F.Minami, M.Ohata, M.Toyoda, T.Tanaka,
25.
K.Arimochi, A.G.Glover, and T.H.North, 1995.
The effect of weld metal yield strength on the
fracture behaviour of girth welds in grade 550
pipe. Proc. Pipeline Technology Conf., pp441-461,
Ostend, Belgium.
26.G.F.Bowie, F.J.Barbaro, C.J.Moss, L.Fletcher, and
R.Denys, 1997. Defect acceptance criteria in thin
walled pipes girth welds. Proc. 2nd Symposium
on Mis-Matching of Interfaces and Welds, pp573586, GKSS Geesthacht.
27.
W.Schipaanboord, R.Denys, A.Lefevre, and
P.Roovers, 1999. Validation tests to sSupport the
EPRG – Tier 2 guidelines. Proc. 12th Biannal
JTM on Pipe Line Research, Groningen, 17-21
May.
28.R.Denys, A.Lefevre, C.De Jaeger, and S.Claessens,
2000. Failure characteristics and defect tolerance
levels of girth welds in large diameter X65/
X70 steel pipelines: experimental verification
through wide plate testing and comparison with
ECA prediction models. Proc. 3rd International
Pipeline Technology Conference, Elsevier Science,
Amsterdam, I, pp151-167, Brugge, Belgium.
29.R.Denys, W.De Waele, and A.Lefevre, 2005. Input
to code revision for interaction of girth weld
defects under plastic collapse conditions. Contract
GRI 8766, January.
30.H.G.Pisarski, A.Muhammed, and A. Staccy, 2000.
Using wide plate test results to improve predictions
from probabilistic fracture mechanics. ECF 13,
San Sebastian, Spain.
31.R.M.Andrews and L.L.Morgan, 2003. Integration
of automated ultrasonic testing and engineering
critical assessment for pipeline girth weld defect
acceptance. 14th JTM on Line Pipe Research,
Berlin, paper 36.
32. M.L.Macia, S.D.Papka, D.P.Fairchild, R.Denys, and
A.Lefevre, 2004. Verification of fracture toughness
and defect tolerance of X120 girth welds. Proc.
4th International Pipeline Technology Conference,
I, Ostend, Belgium, 9-13 May.
33.M.V.Hukle, A.M.Horn, D.S.Hoyt, and J.B.LeBleu,
Jr., 2005. Girth weld qualification for high strain
pipeline applications. ASME, Int. Conf. on OMAE,
Paper No 2005-67573, pp1-6, Halkidiki, Greece.
R.Denys, S.HertelГ©, and M.Verstraete, 2010.
34.
Longitudinal strain capacity of GMAW welded
high niobium (HTP) grade X80 steel pipes. Proc.
of HSLP, Xi’an, China.
Sa
m
pl
e
co
py
-n
ot
f
3. API, 1983. API-1104 Standard for field welding
pipe lines and related facilities – Appendix A.
4. BSI, 1984. BS-4515 Specifications for field welding
of carbon steel pipelines – Appendix H. British
Standards Institution, London.
5. CSA, 1986. CSA-Z184 Appendix K, 1986, Standard
for welding pipelines and related facilities – Appendix
K. Canadian Standards Institution, Mississauga.
6. P.D.Hilton and R.A.Mayville, 1985. An evaluation
of girth weld defect acceptance criteria. Arthur
Little Inc., Cambridge, MA.
7. BSI, 1980 and 1991. BSI-PD 6493, 1980 and 1991,
Guidance on some methods for the derivation of
acceptance levels for defects in fusion welded joints.
British Standards Institution, London.
8. British Energy, 2005. R6-Rev 4 Assessment of the
integrity of structures containing defects.
9. BSI, 2005. BS 7910:2005 Guide on methods for
assessing the susceptibility of flaws in structures,
incorporating amendment No.1. British Standards
Institution, London.
10.
API, 2000. API-579 Fitness-for-service: API
recommended practice.
11.M.Kodak, S.Webster , J.J.Janosch, R.A.Ainsworth,
and R.Koers, 2008. FITNET fitness-for-service (FFS)
procedure (Vol.1). GKSS Research Centre, Geesthacht.
12.M.Kodak, L.Hadley, S.Szavai, Y.Tkach, and N.Taylor,
2008. Ibidem, Annex (Vol.2).
13.G.Knauf and P.Hopkins, 1996. EPRG guidelines on
the assessment of defects in transmission pipeline
girth welds. 3R International, 35, 10-11, pp 620-624.
14.R.M.Denys, R.M.Andrews, M.Zarea, and G.Knauf,
201. EPRG Tier 2 guidelines for the assessment of
defects in transmission pipeline girth welds. Proc.
IPC, Calgary, Canada.
15. Standards Australia, 2002. AS 2885.2-2002 Pipelines –
gas and fluid petroleum, Part 2: welding. Homebush,
New South Wales.
16.EN 12732, 2002. Gas supply systems – welding steel
pipework – functional requirements. London.
17.BSI, 2004. BS 4515 Specification for welding of
steel pipelines on land and offshore. Carbon and
carbon manganese steel pipelines. British Standards
Institution, London.
18. API, 2005. API-1104 Welding of pipelines and related
facilities. 20th Edn, Washington, USA, November.
19.Det Norske Veritas, 2007. DNV-OS-F101 Submarine
pipeline systems, offshore.
20.CSA, 2007. Z662 Oil and gas pipeline systems.
Mississauga, Ontario.
21.B.Lian, R.Denys, and L.Van de Walle, 1986. An
experimental assessment on the effect of weld metal
yield strength overmatching in pipeline girth welds.
Proc. 3rd Conf. on Welding and Performance of
Pipelines, TWI, London, November.
22.R.Denys, 1992. Fracture behaviour of girth welds
containing natural defects and comparison with existing
workmanship standards. AGA No. 202-009, pp102, 9 Feb.
256
The Journal of Pipeline Engineering
or
d
is
t
rib
ut
io
n
49.R.Denys and A.Lefevre, 2009. Gent guidelines for
curved wide plate testing. Proc. 5th International
Pipeline Technology Conference, Ostend, Belgium,
12-14 October.
50.Y.-Y.Wang, M.Liu, Y.Chen, and D.Horsley, 2006.
Effects of geometry, temperature, and test procedure on
reported failure strains from simulated wide plate tests.
Proc. IPC, paper IPC2006-10497, Calgary, Canada.
51.R.Denys, 1995. Toughness requirements for pipeline
integrity. Proc. 13th OMAE Conference, III,
Copenhagen.
R.Denys and A.Lefevre, 2006. Material test
52.
requirements for strain-based pipeline design. Proc.
Int. Symposium on Microalloyed Steels for the Oil
and Gas Industry, Araxa, MG, Brazil, TMS (The
Minerals, Metals & Materials Society) Publication,
Warrendale, USA, ISBN 978-0-87339-656-1, pp513-533.
53.
D.P.Fairchild, W.Cheng, S.J.Ford, K.Minnaar,
N.E.Biery, A.Kumar, and N.E.Nissley, 2008. Recent
advances in curved wide plate testing and implications
for strain-based design. Int. J. of Offshore and Polar
Engineering, 18, 3, pp161-170, September.
54.S.HertelГ©, W.De Waele, R.Denys, and M.Verstraete,
2010. Investigation of pipe strain measurements in
a curved wide plate specimen. Proc. IPC, paper
IPC2010-31292, Calgary, Canada.
55.R.Denys, S.HertelГ©, and M.Verstraete, 2010. Strain
capacity of weak and strong girth welds in axially
loaded pipelines. Proc. IPTC 2010, International
Pipeline Conference, Beijng, China.
T.Weeks, D.McColskey, M.Richards, Y.-Y.Wang,
56.
H.Zhou, W.Tyson., and M.Quitana, 2011. Weld
design, testing, and assessment procedures for highstrength pipelines curved wide plate tests. US DOT
Contract No. DTPH56-07-T-000005, final report to
US DOT and PRCI, September.
57.S.HertelГ©, W.De Waele, R.Denys, and M.Verstraete,
2011. Sensitivity of plastic response of defective
pipeline girth welds to the stress-strain behaviour of
base and weld metal. Proc. 30th OMAE Conference,
paper 2011-49239, Rotterdam, June.
58. S.HertelГ©, M.Verstraete,K.Van Minnebruggen, R.Denys,
and W.De Waele, 2012. Curved wide plate testing
with advanced instrumentation and interpretation.
Proc. IPC, paper IPC2012-90591, Calgary, Canada.
59. S.HertelГ©, W.De Waele, R.Denysm and M. Verstraete,
2012. Full-range stress-strain behaviour of contemporary
pipeline steels: Part I: model description. Int. J. Press
Vessels and. Pipe, 92, 34-40.
60.S.HertelГ©, W.De Waele, R.Denys, and M.Verstraete,
2012. Investigation of strain measurements in a
(curved) wide plate specimen using digital image
correlation and finite element analysis. J. Strain Anal.
Eng. Des., 47, 5, 276-88.
61. S.Hertelé, N.P.O’Dowd, R.Denys, K.Van Minnebruggen,
and W.De Waele, 2013. Effects of pipe steel
heterogeneity on the tensile strain capacity of a flawed
pipeline girth weld. Submitted to Eng. Fract. Mech.
Sa
m
pl
e
co
py
-n
ot
f
35. Y.-Y.Wang, M.Liu, X.Long, M.Stephens, R.Petersen,
and R.Gordon, 2011. Validation and documentation
of tensile strain limit design models for pipelines.
US DOT Contract No. DTPH56-06-T000014, 2
August.
36.T.Swankie, 2012. Validation of assessment methods
for production scale girth welding of high strength
steel pipelines with multiple pipe sources. US DoT,
DTPH56-07-T-000006 WP #275, March.
37.G.M.Wilkowski and R.J.Eiber, 1978. Review of
fracture mechanics approaches to defining critical
defect size girth weld discontinuities. Weld. Res. Bull.,
329, July.
38.A.H.Roodbergen and R.Denys, 1987. Limitations
of fitness-for-purpose assessment of pipeline girth
welds. Int. Conf. on Pipeline Technology, AIM,
Rome.
39.J.R.Gordon, 1993. A review of fracture assessment
procedures and their applicability to welded
structures: fracture control in welded structures: codes
and standards. Special Issue: Canadian Metallurgical
Quarterly, 32, 3, July-September, pp195-204.
40.P.M.Scott, 1993. Interpretative study of published
and recent research on the applicability and
limitations of current fracture prediction models
for girth welds. Ibidem, pp223-238.
41.R.Denys, A.Lefevre, J.T.Martin, and A.G.Glover,
1995. Yield strength mis-match effects on fitness
for purpose assessments of girth welds. PRC/EPRG
10th Biennial Joint Technical Meeting on Line Pipe
Research, AGA, Paper 3, pp1-15, Cambridge, UK.
P.M.Scott, T.L.Anderson, D.A.Osage, and
42.
G.M.Wilkowski, 1998. Review of existing fitnessfor-service criteria for crack-like flaws. WRC Bulletin,
430, 155p April.
Y.-Y.Wang, M.Liu, and D.Rudland, 2007. A
43.
comprehensive update in the evaluation of pipeline
weld defects. US DOT Agreement # DTRS5603-T-0008, PRCI Contact # PR-276-04503.
44.A.Cosham, 2008. ECAs: are they fit-for-purpose?
OPT 2008, Amsterdam, Netherlands, February.
45.G.Wilkowsky, D.Rudland, D.Shim, and F.W.Burst,
2008. An advanced integration of multi-scale
mechanics and welding process simulation in weld
integrity assessment. Final Technical Report DOE
DE-FC36-04Go14040, June.
46.R.Denys and W.De Waele, 2009. Comparison of
API 1104 - Appendix A Option 1, and EPRG-Tier
2 defect acceptance limits. Proc. Pipeline Technology
Conference, Ostend, Belgium.
47.R.Denys, W.De Waele, and A.Lefevre, 2004. Effect
of pipe and weld metal post-yield characteristics on
plastic straining capacity of axially loaded pipelines.
IPC, paper IPC04-0768, Calgary, Canada.
48.R.Denys, 2008. Weld metal strength mismatch: past,
present and future. Proc. Int. Symp. to Celebrate
Prof. Masao Toyoda’s Retirement from Osaka
University, Osaka, Japan pp115-148.
3rd Quarter, 2013
257
or
d
is
t
rib
ut
io
n
68.X.Li, L.Ji, W.Zhao, and H.L.Li , 2007. Key issues
should be considered for application of strain-based
designed pipeline in China. Proc. ISOPE2007,
Lisbon, Portugal.
69. L.Fletcher, 2003. Strain limits for hydrostatic testing:
Revision to AS 2885.1. Issue Paper 1.12, Revision:
B, 20 June.
70.Y.-Y.Wang, M.Liu, J.Gianetto, and W.Tyson, 2010.
Considerations of linepipe and girth weld tensile
properties for strain-based design of pipelines. Proc.
IPC, Calgary, Canada.
71.W.Cheng, P.C.Gioielli, H.Tang, K.Minnaar, and
M.L.Macia, 2009. Test methods for characterization
of strain capacity – comparison of R curves from
SENT/CWP/FS tests. Proc. Pipeline Technology
Conference, Ostend, Belgium.
72. H.Tang, K.Minnaar, S.Kibey, M.L.Macia, P.C.Gioielli,
and D.P.Fairchild, 2010. Development of SENT
test procedure for strain-based design of welded
pipelines. Proc.IPC, Calgary, Canada.
73. R.Denys, 1990. Wide plate testing of weldments, Parts I,
II, and III. Fatigue and Fracture Testing of Weldments,
ASTM STP 1058, pp157-228, ASTM, Philadelpia.
Sa
m
pl
e
co
py
-n
ot
f
62.R.Denys, S.HertelГ©, M.Verstraete, and W.De Waele,
2011. Strain capacity prediction for strain-based
pipeline designs. Proc. Workshop Welding of High
Strength Pipeline Steels, Araxa, Brazil.
63.M.Verstraete, W.De Waele, R.Denys, and S.HertelГ©,
2012. Pressure correction factor for strain capacity
predictions based on curved wide plate testing.
Proc. IPC, paper IPC2012-90592, Calgary, Canada.
64.J.R.Gordon, N.Zettlemoyer, and W.C.Mohr, 2007.
Crack driving force in pipelines subjected to large
strain and biaxial stress conditions. Proc. IPC, 4,
pp3129-3140, Lisbon, Portugal.
65.K.Minnaar, P.C.Gioielli, M.L.Macia, F.Bardi, and
W.C.Kan, 2007. Predictive FEA modelling of
pressurized full-scale tests. ISOPE, Lisbon, Portugal.
66.S.Kibey, S.P.Lele, H.Tang, M.L.Macia, D.P.Fairchild,
W.Cheng, W.C.Kan, M.F.Cook, R.Noecker, and
B.Newbury, 2011. Full scale test protocol for
measurement of tensile strain capacity of welded
pipelines. Proc. ISOPE, Maui, Hawaii, USA.
67.D.B.Lillig, D.S.Hoyt, M.W.Hukle, J.Dwyer, and
A.M.Horn, 2006. Materials and welding engineering
for ExxonMobil high strain pipelines. Proc. 16th
Int’l Offshore and Polar Eng. Conf., pp8-17, San
Francisco, USA.
rib
ut
io
n
DON’T MISS
AN ISSUE!
Sa
m
pl
e
co
py
-n
ot
f
or
d
is
t
SUBSCRIBE TODAY
Purchase your Pipelines International subscription online at
pipelinesinternational.com/subscribe
3rd Quarter, 2013
259
is
t
1 Edison Welding Institute, Columbus, OH, USA
2 B N Leis Consultant, Inc, Worthington, OH, USA
rib
ut
by Dr Xian-Kui Zhu*1 and Dr Brian N Leis2
io
n
Ductile-fracture arrest methods
for gas-transmission pipelines
using Charpy impact energy or
DWTT energy
T
or
d
HE STANDARDIZED CHARPYV-notched (CVN) impact energy has been used by the pipeline industry
since the 1960s to characterize fracture toughness of pipeline steels, and is central to the fracturecontrol technology developed for gas transmission pipelines. The drop-weight tear test (DWTT) has been
standardized to assess fracture mode in such applications, with DWTT energy suggested as a means to
quantify toughness, although not standardized for such a use.
py
-n
ot
f
Battelle developed its two-curve model (BTCM) in the early 1970s to determine the required toughness
to arrest ductile fracture in gas transmission pipelines in terms of CVN impact energy.The BTCM has been
found viable for pipeline grades up to X-65, but issues have emerged in applications to higher grades. Thus
different correction methods were proposed over the years to improve the BTCM predictions.This paper
reviews the use of CVN and DWTT energy in conjunction with the BTCM to predict arrest toughness to
control running fractures in gas transmission pipelines, and evaluates correction methods adopted to extend
its use to X-80 and above. The correction methods include the Leis correction factor, the CSM factor, the
Wilkowski DWTT method, and others. These methods are evaluated through analysis and comparison of
predictions with full-scale experimental data. Suggestions to further improve the BTCM also are discussed.
D
Sa
m
pl
e
co
UCTILE FRACTURE PROPAGATION control is
a major concern for the safe design and operation
of modern gas transmission pipelines at high internal
pressure. Technology to ensure such control is critical
for the structural integrity and safety of the gas pipeline
because the possibility of a running fracture opens-up the
catastrophic long-running failure of a gas pipeline which
poses a societal threat involving the public and property,
as well as to the environment local to the pipeline.
Fracture resistance to a running fracture is an important
material property, with the minimum resistance to
affect arrest defining the arrest toughness. The absorbed
energy obtained in a Charpy V-notched (CVN) impact
test has been used to characterize the ductile fracture
resistance since the 1960s in the pipeline industry,
*Corresponding author’s contact details:
tel: +1 614 688 5135
email: [email protected]
while a drop-weight tear test (DWTT) has been used
to assess the fracture mode, with DWTT energy also
suggested to quantify toughness. Both CVN and DWTT
have played a key role in that context as part of the
technology developed for fracture-propagation control
(FPC) for gas transmission pipelines [1].
Based on the fracture resistance quantified by CVN
energy, Battelle developed its two-curve model (denoted
here as BTCM) in the mid-1970s to predict arrest
toughness for ductile pipeline steels. Maxey [2] detailed
the technology that underlies the BTCM and its
use for dynamic fracture control. The fracture and
the gas-decompression behaviours in a gas pipelines
were uncoupled, being described by two independent
curves. The gas-decompression curve was quantified
by the code GASDECOM [3] developed at Battelle,
which was valid for a wide range of gas compositions;
the fracture-resistance curve was described by a semiempirical fracture model in terms of CVN energy. It
has been found that the CVN-energy-based BTCM is
The Journal of Pipeline Engineering
rib
ut
io
n
260
co
py
-n
ot
f
Much has been done since the 1980s to improve the
predictability of BTCM in applications at the highertoughness levels commonly available for modern highstrength steels. Such work includes corrections and
correlations: viz. the Leis correction [4], CSM factor
[5], Wilkowski pressed-notch DWTT method [6], and
Japanese pre-cracked DWTT method [7]. In a soon-tobe-released update of the PRCI report on Fracture control
technology for transmission pipelines, Leis and Eiber [8]
outline the history, technology, methods, and progress
for ductile FPC for gas-transmission pipelines. Further
details of these topics can be found in that report.
design against this threat. In the 1960s, as the need for
such technology became clear, the potential of fracturemechanics’ technology was evident, but the capabilities
were rudimentary as compared to the complexity posed
by ductile running fractures. More critically, the CVN
test had been adopted by the pipeline industry to
quantify fracture resistance: accordingly, the technology
developed to quantify arrest toughness was formulated
in reference to CVN impact energy, as follows.
or
d
accurate only for pipeline grades up to X-65 with low
to intermediate fracture toughness, with issues developing
for higher toughness levels, which commonly develop
for higher-strength pipeline grades.
is
t
Fig.1. Fracture and gas-decompression
velocity curves.
Battelle two-curve model
Battelle’s two-curve model (BTCM) was presented by Maxey
[2] in 1974. This model considers the gas decompression
(fracture-driving curve) and the dynamic crack-propagation
resistance (fracture-resistance curve) as uncoupled processes,
with both related to decompressed pressure local to the
crack tip. The gas-decompression curve is determined
by the code GASDECOM developed at Battelle that is
viable for a wide range of gas compositions [3]. The
fracture-propagation curve (or fracture-resistance curve)
is determined by the following semi-empirical equation:
Sa
m
pl
e
This paper reviews the use of CVN and DWTT
energy in conjunction with the BTCM to predict
arrest toughness to control a running fracture in gastransmission pipelines, and evaluates correction methods
adopted to extend its use to X-80 and above. The
1/ 6
objective is to assess the utility of the methods available
Пѓf пЈ« P пЈ¶
=
V
C
в€’
1
(1)
пЈ¬
пЈ·
f
to assess arrest toughness for modern pipelines made
R пЈ­ Pa пЈё
of high-toughness and high-strength steels. Corrections
and correlations are evaluated through analysis and
comparison of predictions with full-scale experimental where:
data. Suggestions to further improve the BTCM also
are discussed.
Vf is the fracture propagation velocity in m/s
(or ft/s);
CVN impact energy and ductile
C is a backfill parameter with a constant
fracture arrest methods
value of 2.75 (or 0.648) and 2.34 (or 0.47),
respectively for no backfill in air conditions
Because high internal pressure can cause rapid axial
and soil backfill;
ductile fracture propagation along a gas-transmission
Пѓf = Пѓy + 69 MPa (or 10 ksi) is a flow stress
pipeline, technology has evolved as noted above to
in MPa (ksi);
3rd Quarter, 2013
261
et al. [10] curve-fit a simple formula to the results of
parametric BTCM analyses permitting direct calculation
of the arrest toughness for such applications. Based on
the results developed via the BTCM for a wide range
of mechanical properties and CVN values, the arrest
fracture toughness quantified relative to a 2/3-size CVN
specimen was obtained as a function of the initial hoop
stress and pipe geometry as:
io
n
R = CV/Ac is fracture resistance with CV the
full-size Charpy (CVN) impact energy at the
upper shelf in Joule (or ft-lb) and Ac the
ligament cross-section area of Charpy specimen
with a value of 80 mm2 (or 0.124 in2);
P is the instant decompressed pressure near
the crack tip in MPa (or psi); and
Pa=2tПѓa/D is the arrest pressure at the crack
tip in MPa (or psi).
The arrest hoop stress is determined by:
CV ( 2 /=
7.2 Г— 10в€’3 Г— Пѓ h2 ( Rt )
3)
пЈ®
пЈ№
пЈ® 2Пѓ f пЈ№
ПЂ RE
Пѓa = пЈЇ
]пЈє(2)
пЈє arccos пЈЇexp[в€’
2
24Пѓ f Dt / 2 пЈєпЈ»
пЈЇпЈ°
пЈ° 3.33ПЂ пЈ»
where CVN energy is in ft-lb, the hoop stress Пѓh is in
ksi, and the pipe radius R and wall thickness t are in
inches. Because of its simplicity, this simplified form of
the BTCM has been adopted in a variety of codes and
standards for gas-transmission pipeline design [for example,
see Ref. 8]. Concurrently many simplified equations (SEs)
have emerged in a form similar to Equn 3 through
work done in the North America, Europe, and Japan.
While their predictive quality varies, these SEs provide
a simple basis for estimating arrest toughness. Nine such
equations can be found in Ref. 8.
(US units) (3)
is
t
or
d
Since the mid-1970s, pipeline-steelmaking technology has
been improved remarkably and thus fracture toughness
of pipeline steels has been significantly increased. It
was found that the BTCM and all simplified models
predicted non-conservative arrest toughness in comparison
to measured Charpy energy for high-strength pipeline
steels with CVN energies larger than 70 ft-lb (or 95 J)
[4]. Therefore, a variety of corrections, correlations, and
modified methods have been proposed to improve the
BTCM predictions, as reviewed next.
ot
f
where E is the elastic modulus in GPa (or ksi), D is
the pipe diameter in mm (or inches), and t is the pipe
wall thickness in mm (or inches). Equn 2 was originally
developed in Ref. 9 for fracture-initiation control for
pipelines by use of the rudimentary fracture-mechanics’
method. In this equation, a linear relationship between
the CVN energy and the fracture toughness GC was
assumed. The linear relationship was obtained by
calibration of full-scale hydrostatic burst data for pipeline
steels dominated by grades ≤ X65.
rib
ut
1/ 3
Sa
m
pl
e
co
py
-n
When the fracture curve determined by Equn 2 and the
gas-decompression curve determined by GASDECOM are
tangent, as shown in Fig.1, the minimum fracture-arrest
toughness in terms of CVN energy is determined for
the pipeline under consideration. If the CVN toughness
increases, the fracture curve moves up to be above the
gas curve, and a running fracture will arrest because
the fracture velocity is slower than the gas-decompression
velocity at all pressure levels. On the other hand, if
the CVN toughness decreases, the fracture curve moves
down, and the running fracture will continue to propagate
because the fracture velocity is faster than the gasdecompression velocity. Note that the gas-decompression
code GASDECOM is quite general in its applicability
because it embeds the BWRS equations of state and
accounts for both single-phase and two-phase decompression
behaviours of gases. In contrast, Equns 1 and 2 used for
determining the fracture-resistance curve were calibrated
with experimental data available in the late 1960s to the
early 1970s, such that lower-toughness linepipe steels up
to grade X-65 were involved. Therefore, the BTCM may
be inaccurate when used beyond its calibration database.
Simplified equation to represent the BTCM
Use of the BTCM to predict the arrest toughness
depends on its software. While DYNFRAC developed
at Battelle [3] is available today, its initial use required
iterative computer-based analysis. To facilitate such analysis
when dealing with single-phase decompression, Maxey
Leis correction
It has been found the proportion of propagation to
initiation energy dissipated in the CVN specimen
varies greatly with increasing toughness for ductile
pipeline steels [for example, Ref.4]. Experimental data
showed this proportion was fairly constant for the
lower-toughness steels, with the propagation component
tending to zero as the toughness increases toward
250 ft-lb (340 J). From these observations, and
based on the energy-dissipation principle, Leis [4]
developed a correction to the BTCM for the
Alliance pipeline project. In this correction, the arrest
toughness in terms of CVN energy is found to be the
same as that determined by the BTCM if the measured
CVN energy is less than 95 J (or 70 ft-lb); otherwise,
a non-linear correction between actual arrest toughness
and the BTCM-predicted arrest toughness is needed.
The Leis correction can be expressed mathematically as:
( CV )arrest = ( CV )BTCM
for (CV) < 95 J (4a)
The Journal of Pipeline Engineering
rib
ut
io
n
262
or
d
The factor was determined as the
to affect a correct prediction via
the CSM factor is expressed as a
between the arrest toughness
predicted toughness:
multiplier required
the BTCM. Thus,
linear relationship
and the BTCM
ot
f
See also Equn 4b (below) where (CV)arrest is the CVN
full-size equivalent (FSE) energy in Joules required for
arrest, and (CV)BTCM is the CVN FSE arrest energy
in Joules, calculated using Equn 3. Because the ratio
of (CV)arrest/(CV)BTCM determines a correction factor to
the BTCM prediction, the Leis correction was often
referred to as the Leis correction factor.
is
t
Fig.2. Comparison of arrest Charpy (Cv)
toughness predictions by four improved
methods.
py
-n
Because the characteristics of the flow and fracture
responses of the steels involved in developing Equn
4 had been limited to grades X-70 and below, this
equation was limited in its use to X-70 and below
[8]. Care must be taken if this correction is used for
higher grades of X-80 and above.
co
Modified Leis correction
Sa
m
pl
e
Based on recent experimental burst data for X-70
and X-80 pipeline steels, Eiber [11, 12] showed that
the Leis correction is accurate for X-70 steels, but
was less so for some X-80 steels. If the coefficient of
0.002 in Equn 4b was empirically set at 0.003, Eiber
determined that the following modified equation better
predicted the arrest toughness for a set of full-scale
experimental data covering a range of X-80 pipeline
steels, shown in Equn 5 below.
(CV)arrest = k (CV)BTCM(6)
Such multiplicative factors were determined first for
X-80 steels, and then again as X-100 and X-120 pipes
were subjected to test. The constant factor k was found
to be 1.43 for X-80 and 1.7 or higher for X-100, as
given by Demofonti [13]. In contrast to the non-linear
form of the Leis correction factor, the CSM factor
is linear and specific to the grade and full-scale tests
that underlie its determination.
Statistical factor
As the CSM factor depends on full-scale burst tests
and is grade-specific, a more-general factor was sought
via a statistical analysis. Wolodko and Stephens [14]
at C-FER obtained a statistical correction for testing
involving single-phase decompression in grades from
X-70 to X-100 in the form:
CSM factor
= (1.5 + 0.29nsd )( CV ) BTCM (7)
( CV )arrest
CSM [13] proposed a simple factor determined by
comparing the actual arrest toughness based on results
of full-scale burst data to the corresponding BTCM
prediction of the minimum CVN arrest energy (CV)BTCM.
where nsd is the multiplier on the standard
deviation of the model error that can be selected to
achieve the desired probability of non-arrest of a
( CV )arrest = ( CV )BTCM + 0.002 ( CV )BTCM в€’ 21.18
2.04
for (CV) ≥ 95 J ( CV )arrest = ( CV )BTCM + 0.003 ( CV )BTCM − 21.18 for (CV) ≥ 95 J
2.04
(4b)
(5)
263
rib
ut
io
n
3rd Quarter, 2013
1/ 6
Vf = C
Пѓf пЈ« P пЈ¶
пЈ¬ в€’ 1пЈ·
K R пЈ­ Pa пЈё
(8)
ot
f
Figure 2 compares the arrest-toughness predictions by
the four correction methods, i.e. the Leis correction
in Equn 4, the modified Leis correction in Equn 5,
the CSM factor in Equn 6, and the statistical factor
in Equn 7. It is assumed here that X-80 pipeline
steels are considered in this figure, and so the factor
k = 1.43 in Equn 6, and the factor in Equn 7 is
taken as 1.935. For X-80 steels, the arrest toughness
generally ranges from 130 to 270 J, and the CSM
simple factor gives a reasonable prediction, as shown in
Ref. 13. On this basis, Fig.2 shows that the modified
Leis correction in Equn 5 predicts a good result of
arrest toughness that is close to the CSM factor result
from Equn 6. The original Leis correction in Equn
4 slightly underestimates – and the statistical factor
in Equn 7 overestimates – the arrest toughness for
X-80 gas pipeline steels. As such, the statistical-factor
equation not recommended be used, at least for X-80
pipeline steels.
Based on their test data, these authors modified
Equn 1 of the fracture curve as:
or
d
running fracture. For example, when nsd = 1.0, 1.5, and
2.0, corresponding to the probability of non-arrest of
16%, 6.7%, and 2.3%, the factor between the required
arrest toughness and the BTCM prediction used in
Equn 7 will be 1.79, 1.935, and 2.08, respectively.
is
t
Fig.3. Effect of speed-dependent
fracture toughness on fracture-speed
curves (from [17]).
m
pl
e
co
py
-n
where:
Sa
Backfill correction
In the fracture curve of the BTCM (Equn 1), the
effect of backfill on the fracture velocity is lumped
into a �backfill coefficient’ that is empirically
based, and does not distinguish between different
soil types or strengths. In Equn 1, a constant
power-law exponent of 1/6 is fixed for all kinds of
backfill. In order to improve this, and to consider
the backfill effect due to different backfill depths and
different soil types, Rudland and Wilkowski [15, 16]
conducted a series of burst tests for gas pipelines.
K = 0.275 Hactual / Hnominal + 0.725;
Hactual is the actual backfill depth used in the
burst test of gas linepipes; and
Hnominal = 30 in, as was employed in the early
gas burst tests conducted by Battelle in the
1970s to calibrate the backfill coefficient C.
This modification in Equn 8 appears reasonable, but
its practical use is limited because the burst tests in
calibration of Equn 8 were conducted only for pipes
with toughness CVN < 100 J.
Speed-dependent toughness method
In the original BTCM, the fracture toughness CVN
was assumed as a constant material resistance. Since
experiments showed that fracture toughness or resistance
is dependent on the fracture speed, Duan and Zhou
[17, 18] at TransCanada modified the fracture resistance
as a speed-dependent value:
R = R0V fв€’ a(9)
where:
;
Rref is a reference fracture resistance at the
reference speed Vref; and
О± is a fracture-speed-dependent index.
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
264
steels. Later, the DWTT was also evaluated as the basis
for measuring the ductile-fracture resistance. The larger
DWTT specimen with its full-size wall thickness has
been considered superior to the smaller CVN specimen
in quantifying fracture resistance for ductile pipeline
steels with high toughness and large plastic deformation,
particularly for modern high-strength pipeline steels.
Correlations between the DWTT and CVN energies
were developed as the basis for adapting the BTCM
for use with DWTT energy, as outlined next.
ot
f
Figure 3 show the effect of speed-dependent fracture
toughness on fracture-speed curves for an X-80 pipeline
steel, where О± = 0 represents the case of constant
fracture toughness as used in the BTCM. When О±
= 0.2, the predictions are in good agreement with
full-scale burst tests of X-80 pipes.
or
d
Fig.4. Correlation between Charpy and
DWTT specific energies (E/A) (from
[20], (1ft-lb/in2 = 0.0021J/mm2).
-n
Reformulated BTCM
Sa
m
pl
e
co
py
As the BTCM remains the only viable tool to quantify
arrest toughness, consideration has been given to its
reformulation to extent its use for tough higher-grade
linepipe steels. Very recently, in a work funded by the
China National Petroleum Cooperation [19], the present
authors adopted modern elastic-plastic fracture mechanics
and in part reformulated the original BTCM. This
reformulated BTCM better predicted arrest toughness
as compared to the modified Leis correction for high
pipeline grade X-80, and also was effective for X-100
when compared to full-scale burst-test data. A simplified
equation also was developed, whose predictions are
comparable to those by the Reformulated BTCM.
However, as this work 1s proprietary, the present
paper simply notes it, in passing.
Research exploring DWTT-based
ductile-fracture-arrest technology
Early DWTT-energy methods
A drop-weight tear test (DWTT) specimen was developed
at Battelle, which was the first candidate considered to
replace the Charpy impact specimen to quantify the
ductile-to-brittle transition temperature and determine
the fracture mode (brittle versus ductile) for pipeline
Early DWTT method at Battelle
Wilkowski et al. [20, 21] at Battelle developed a linear
correlation between the standard pressed-notch (PN)
DWTT energy density and the CVN energy density
for conventionally rolled steels and quenched and
tempered steels in the late 1970s:
пЈ«EпЈ¶
пЈ«EпЈ¶
= 3 пЈ¬ пЈ· + 300
пЈ¬ пЈ·
A
пЈ­ пЈё DWTT
пЈ­ A пЈёCVN
(ft-lb/in2)
(10)
where E is the total fracture energy in ft-lb, A is
fracture area of the specimen ligament in in2, and
E/A denotes the energy density (or the energy per
unit area) in ft-lb/in2.
Figure 4 shows the linear relationship between the
CVN and DWTT energy densities that is good for
these selected steels. When the minimum CVN energy
for a fracture arrest is predicted using the BTCM, the
minimum DWTT energy for the fracture arrest can be
determined from Equn 10.
265
is
t
rib
ut
io
n
3rd Quarter, 2013
Early DWTT method at British Gas
slope of the linear-correlation function continues to
decreases from 2.94 for grade X-60 to 1.91 for grade
X-100. Therefore, a general correlation between CVN
and DWTT is a non-linear function for high-toughness
pipeline steels. This observation is consistent with
that by Leis [23].
m
pl
e
co
py
-n
ot
f
Fearnehough et al. [22] at British Gas were among
the early investigators who developed test methods
and showed that the propagation energy in highertoughness pipeline steels was not linearly related to
the Charpy energy. Their work, which preceded that
of Wilkowski, involved a series of DWTT specimens
that were pre-cracked to different crack lengths – from
short to deep – under quasi-static loading, and then
impacted under drop-weight dynamic loading. This
test was called an �interrupted DWTT’ practice. The
energy density (i.e. E/A) was determined for this series
of interrupted DWTT specimens with differing crack
lengths, and compared with the Charpy energy density
for the same steel, with the result as shown in Fig.5.
This figure indicates that the two energy densities are
linearly related for the lower-toughness steels (i.e. data
in groups A and B). However, for the higher-toughness
steels (with CVN energy of 70 J or more) the E/ADWTT
energy becomes strongly nonlinear with E/ACVN (i.e.
data in group C).
or
d
Fig.5.Variation of DWTT (E/A)
and Charpy (E/A) as developed by
Fearnehough (from [22]).
Sa
Leis [23] discussed the linear correlation between DWTT
and CVN energies in Equn 10 for pipeline grades
up to X-70, and found that most burst-test data for
Alliance pipeline steel X-70 did not follow the linear
trend, but behaved in a non-linear relation similar to
that in Fig.5. This raises questions about generality
of any linear correlation. For modern high-toughness
pipeline steels with CVN energy larger than about
100 J, experiments showed that the relation between
DWTT and CVN energies deviates from the linearity.
Recently, Wilkowski et al. [6] showed that the pipeline
grade has significant effect on their correlation. The
Brittle-notch DWTT specimens
In the 1970s, investigators believed that the non-linear
relationship might be caused by the large initiation
energy obtained by the standard PN DWTT specimens
for tougher steels. Accordingly, Battelle modified
the standard PN DWTT specimen in an attempt to
reduce the initiation energy from the total DWTT
energy – so that the propagation energy is dominant –
through use of a brittle-notch (BN) DWTT specimen.
Figure 6 shows the results obtained by Wilkowski et
al. [20, 24] in 1977 using the BN DWTT specimens
for pipeline grades up to X-70. Based on these test
data, Wilkowski et al. proposed a curve-fitted nonlinear function between the BN and PN DWTT
energy densities:
0.385
пЈ«EпЈ¶
пЈ«EпЈ¶
= 175 пЈ¬ пЈ·
в€’ 1500 (11)
пЈ¬ пЈ·
пЈ­ A пЈё BN в€’ DWTT
пЈ­ A пЈё PN в€’ DWTT
where the energy densities and the constant are in
ft-lb/in2.
An alternative to the brittle-notch DWTT specimen was
a termed the static-precracked (SPC) DWTT specimen,
which is the interrupted test of Fearnehough, but uses
The Journal of Pipeline Engineering
is
t
rib
ut
io
n
266
пЈ№
175 пЈ® пЈ« E пЈ¶
пЈ«EпЈ¶
=
пЈ¬ пЈ·
пЈЇ1.3 пЈ¬ пЈ·
пЈє
3 пЈ° пЈ­ A пЈё DWTT пЈ»
пЈ­ A пЈёCVN (W 2000 )
Wilkowski DWTT methods
pl
e
co
py
-n
In order to reflect the non-linear relationship between
the DWTT and CVN energies, Wilkowski et al. [6]
proposed two new non-linear correlations. They
assumed that the standard PN DWTT specimen used
for fitting Equn 10 and the BN DWTT specimen
were equivalent (because both specimens have less
initiation energy in the total absorbed energy). In this
case, they replaced the PN DWTT energy in Equn 10
with the BN DWTT energy in Equn 11 and obtained
the following so-called Wilkowski 1977 prediction of CVN
energy from the standard DWTT tests:
0.385
в€’ 600.0(12)
Sa
m
пЈ№
175 пЈ®пЈ« E пЈ¶
пЈ«EпЈ¶
=
пЈ¬ пЈ·
пЈЇпЈ¬ пЈ·
пЈє
3 пЈ°пЈ­ A пЈё DWTT пЈ»
пЈ­ A пЈёCVN (W 1977 )
this statistical factor with Equn 12 gives the Wilkowski
2000 prediction of the CVN energy from the standard
DWTT tests:
ot
f
a fixed notch length. These DWTT specimens have
crack-like notches. It has been shown that the SPC
DWTT specimen leads to results similar to the BN
DWTT [6].
or
d
Fig.6.Variation of brittle-notch and pressed-notch DWTT
energies for pipeline grades up to X-70 (from [24]).
where the specific energy and the constant are in
ft-lb/in2, (E/A)DWTT is the total PN DWTT energy density,
and (E/A)CVN(W 1977) is the total CVN energy density.
When compared with the full-scale burst-test
results for pipeline steels of X-52, X-60, X-65, and
X-70, Wilkowski et al. [25] found in 2000 that
the estimation from Equn 12 resulted in an
overestimation of arrest DWTT energy in comparison
to the full-scale DWTT data. A statistical factor of the
overestimation was determined as 1.291. Combining
0.385
в€’ 600.0 (13)
This equation, as well as Equn 12, can be used to
estimate the required arrest toughness in terms of
DWTT energy when the minimum CVN energy is
predicted by the BTCM.
From Equns 12 or 13, Wilkowski et al. [6] also obtained
the arrest CVN energy by use of the BTCM-predicted
CVN toughness, although the authors did not describe
the procedure they used for such a prediction. As a
result, this predicted CVN result may rely on the
quality of the correlation of CVN energy with DWTT
energy, and embed the related uncertainty, which later
discussion in regard to Fig.10 indicates can be very large.
Kawaguchi DWTT method
Kawaguchi et al. [26] considered the Wilkowski DWTT
equations for use with X-80 linepipe steels. They found
that Equn 11 did not match their test data for X-80
because this curve-fit equation was based on test data
only up to X-70 (see Fig.6); they therefore proposed
the following relationship to correlate the SPC DWTT
and the PN DWTT energy densities:
0.9563
пЈ«EпЈ¶
пЈ«EпЈ¶
= 0.9431пЈ¬ пЈ·
пЈ¬ пЈ·
пЈ­ A пЈё SPC в€’ DWTT
пЈ­ A пЈё PN в€’ DWTT (ft-lb/in2) (14)
267
is
t
rib
ut
io
n
3rd Quarter, 2013
Wilkowski CVN correction
ot
f
Following Wilkowski’s assumption and concepts that
generated the non-linear correlation as Equn 12, using
Equns 10 and 14 Kawaguchi obtained the following
correlation between the Charpy and DWTT energy
densities for X-80 steels:
or
d
Fig.7. Comparison of three
correlations between DWTT
and Charpy specific energies.
-n
0.9563
2
в€’ 100 (ft-lb/in ) (15)
py
пЈ®пЈ« E пЈ¶
пЈ№
пЈ«EпЈ¶
= 0.3144 пЈЇпЈ¬ пЈ·
пЈ¬ пЈ·
пЈє
A
A
пЈ­ пЈёCVN
пЈ°пЈ­ пЈё DWTT пЈ»
The literature also includes a so-called Wilkowski
CVN correction equation. Motivated by the Leis
correction Equn 4b, Papka et al. [27] at ExxonMobil
proposed an alternative correction in 2003 in reference
to experimental data and the DWTT correlation
obtained by Wilkowski et al. [20]. From Equns 10
and 12, and assuming (Cv)(W 1977) = (Cv)BTCM, after the
DWTT term was eliminated, the following
�correction’ for the BTCM-predicted CVN arrest
energy was obtained (see Equn 16 below) where the
CVN energy and constant are in ft-lb.
co
Note that both Equns 14 and 15 are nearly linear
because the value of the exponent is close to unity.
Sa
m
pl
e
Figure 7 compares the four correlations between DWTT
and Charpy energy densities determined by the Battelle
linear correlation in Equn 10, the Wilkowski 1977
correlation in Equn 12, the Wilkowski 2000 correlation
in Equn 13, and the Kawaguchi correlation in Equn 15.
It is seen from this figure that Equns 12 and 13 have
a non-linear dependence between the DWTT and CVN
energy densities that diverges from the experimental trend
observed in Fig.4. Equation 15 is nearly linear with another
slope value, and this opens questions concerning the
viability of the three correlations proposed by Wilkowski
[6] and that of Kawaguchi [26], with further discussion
of this aspect following later.
( CV )arrest = 0.04133 пЈ°пЈ®0.138 ( CV )BTCM + 10.29пЈ№пЈ»
( CV )arrest = 0.056 пЈ®пЈ°0.1018 ( CV )BTCM + 10.29пЈ№пЈ»
2.597
2.597
Wolodko and Stephens [14] at C-FER in 2006 converted
this Wilkowski CVN �correction’ in Equn 16 from
Imperial units to SI units as shown in Equn 17
below, where the CVN energy and the constant are
in Joules. Both Equns 16 and 17 have been presented
for use in determining the arrest toughness when
the BTCM-predicted CVN toughness is used
applications involving high-strength pipeline steels.
Eiber [11, 12] considered Equn 17 in the evaluation
of arrest CVN toughness for X-70 and X-80 pipeline
steels, and showed its overestimation of arrest toughness
for such steels.
в€’ 12.4
(16)
в€’ 16.8 (17)
The Journal of Pipeline Engineering
rib
ut
io
n
268
where:
Vc is the fracture velocity in m/s;
Пѓf = (Пѓy + Пѓuts)/2 is the flow stress;
R = Dp/Ap is the material resistance in J/mm2;
Dp is the estimated total energy of PC DWTT
specimen in J;
Ap is the fracture area of PC DWTT specimen
in mm2;
P is the decompressed pressure at the crack
tip in MPa;
Pa is the arrest pressure in MPa;
D is the pipe diameter in mm; and
t is the pipe thickness in mm.
co
py
-n
ot
f
Figure 8 compares the predictions of minimum CVN
arrest toughness obtained by the four CVN-correction
methods in applications to X-80 pipeline steel: i.e. the
Leis correction factor in Equns 4, the Modified Leis
correction in Equn 5, the CSM factor in Equn 6,
and the Wilkowski CVN �correction’ in Equn 17. It
is apparent that the CSM factor prediction is most
accurate, which follows because it was developed based
on these data. Figure 8 indicates that the Wilkowski
�correction’ in Equn 17 gives a conservative result
that is larger than that obtained by the Modified Leis
correction in Equn 5 or the CSM factor in Equn
6. Because the assumption of (Cv)(W 1977) = (Cv)BTCM is
questionable, the resulting correction in Equns 16 or
17 is equally questionable, as are its predictions.
or
d
is
t
Fig.8. Comparison of arrest
CVN toughness predictions
by four CVN-correction
methods
Japanese HLP model
Sa
m
pl
e
In the late 1970s, in parallel to the work done at
Battelle on use of the DWTT specimens, Japanese
researchers [7, 28] began a large research programme
referred to as �HLP’. This involved extensive experimental
and analytical work that sought to extend the fracture
model developed in the BTCM. This work relied on
fracture resistance expressed in terms of pre-cracked
(PC) DWTT energy. For the soil backfill conditions,
fracture velocity in the PC DWTT-based HLP model
was determined as:
Пѓf пЈ« P пЈ¶
V f = 0.670
пЈ¬ в€’ 1пЈ·
R пЈ­ Pa пЈё
tПѓ f
пЈ®
Pa = пЈЇ0.382
D
пЈ°
0.393
(18)
пЈ®
пЈ№
3.81Г—107 R пЈ№(19)
]пЈє
пЈє arccos пЈЇexp[в€’
Пѓ 2f Dt пЈєпЈ»
пЈЇпЈ°
пЈ»
The adaptation of DWTT energy and recalibration of the
constant and exponent in the fracture velocity equation
for direct use with DWTT energy is one of the features
of the HLP model. In order to estimate associated arrest
toughness in terms of CVN energy, the Japanese HLP
Committee developed a correlation between the CVN
energy and PC DWTT energy in the form:
D p (estimate) = 3.29t1.5CV 0.544 (20)
This correlation was developed based on test results from
a variety of linepipes in grades from X-60 to X-100,
and with wall thicknesses from 10 mm to 32 mm [29].
Because Equn 18 was calibrated in reference to the
full-scale burst-test data for X-70 pipeline steels, its
prediction of arrest toughness in terms of DWTT energy
is accurate for X-70 pipeline steels, but inaccurate for
X-80 pipeline steels [29]. Thus, the applications of the
Japanese HLP model to high-strength pipeline steels of
grade X-80 and above are not recommended.
269
rib
ut
io
n
3rd Quarter, 2013
Modified HLP model
or
d
is
t
Fig.9. Actual vs predicted
CVN energies by BTCM
for X-80 and X-100
high-strength pipeline
steels (taken from [13]).
Оі=
3
пЈ« t/D пЈ¶
3.22 + 0.20 пЈ¬
пЈ· (25)
пЈ­ t0 / D0 пЈё
ot
f
To improve the HLP model, Makino et al. [29, 30]
investigated the effect of pipe geometry on the prediction
of arrest toughness. It was found that the accuracy of
arrest toughness prediction by the HLP model depends
on pipe diameter to wall thickness ratio D/t, with the
modified the HLP model referred to as the Sumitomo
model. A new fracture-velocity curve equation was
proposed in a general form:
3.42
-n
In these three expressions, the reference pipe diameter
was set as D0 = 1219.2 mm (48 in) and the reference
wall thickness was set as t0 = 18.3 mm.
ОІ
co
пЈ®
пЈ№
4.57 Г—107 R пЈ№
]пЈє(22)
пЈє arccos пЈЇexp[в€’
Пѓ 2f Dt пЈєпЈ»
пЈЇпЈ°
пЈ»
e
tПѓ f
пЈ®
Pa = Оі пЈЇ0.382
D
пЈ°
py
Пѓf пЈ« P пЈ¶
=
Vf О±
пЈ¬ в€’ 1пЈ· (21)
R пЈ­ Pa пЈё
Sa
m
pl
where the three parameters О±, ОІ, and Оі were assumed
as a function of the pipe diameter D and the wall
thickness t. Note that the coefficient 4.57 in Equn 22
is different from 3.81 in Equn 19. Based on available
full-scale burst-test data for high-strength pipeline steels
up to X-100, the curve fitting determined these three
parameters as:
1/ 4
пЈ« Dt пЈ¶
О± = 0.670 пЈ¬
пЈ·
пЈ­ D0t0 пЈё (23)
пЈ« DпЈ¶
ОІ = 0.393 пЈ¬ пЈ·
пЈ­ D0 пЈё
5/ 2
пЈ«t пЈ¶
пЈ¬ пЈ·
пЈ­ t0 пЈё
в€’1/ 2
(24)
The results in Refs 29 and 30 showed that use of
the newly developed Equns 21 to 25 led to improved
predictions of arrest toughness for high-strength pipeline
grades X-100 and X-120 in comparison to the predictions
by the original HLP model. However, comparisons of
Equns 2, 19, and 22 show that both HLP and improved
HLP models (a) use the same BTCM fracture model for
determining arrest pressure, and (b) assume that the fracture
resistance R = Dp/Ap from PC DWTT energy is equal to
R = Cv/Ac in terms of CVN energy. This implies that the
PC DWTT energy density is equal to the CVN energy
density – which is contrary to the non-linear relationship
between these two energy parameters expressed in Equn
20. Thus, the DWTT-based HLP model and the improved
HLP model appear to be viable for the grades used in
their calibration, but are likely to suffer the same issues
that occur for the BTCM in applications to circumstances
beyond their calibration database.
Further discussion of arresttoughness prediction methods
CVN energy-based methods
Figure 9 shows the measured CVN energy versus the
BTCM-predicted CVN energy for high-strength pipeline
steel grades X-80 and X-100 presented by Demofonti et
The Journal of Pipeline Engineering
rib
ut
io
n
270
The backfill-correction method proposed by Rudland and
Wilkowski [15, 16] considered this aspect over a limited
range of backfill depths and was focused on pipeline
steels with CVN energy less than 100 J. If the backfill
height used in an actual burst test for high-strength
steels is equal to the original backfill height as used
in the BTCM, Equn 8 reduces to Equn 1. As such,
there is little improvement made in such applications,
which implies this correction has limited general utility.
-n
ot
f
al. [13], where experimental burst data were extracted
from the CSM database. As evident in this figure, the
BTCM underestimates the actual arrest-CVN toughness
and leads to non-conservative predictions. Similar results
were observed for the simplified BTCM model. Thus,
as discussed in regard to Equn 6, the arrest toughness
predicted by the BTCM must be multiplied by 1.43
for X-80, and 1.7 for X-100.
or
d
is
t
Fig.10. Correlation between
PN DWTT and CVN
energies (from [13]).
Combining the results in Figs 2 and 9 shows that:
pl
e
co
py
a. the modified Leis correction in Equn 5 leads
to viable predictions of arrest CVN toughness
in comparison to the CSM linear-factor method;
b. the original Leis correction factor in Equn 4
slightly underestimates the actual arrest CVN
energy (this is not surprising because Equn 4 was
calibrated for grades up to X-70); and
c. the statistical factor method overly estimates the
arrest CVN energy.
Sa
m
While not evident in this paper due to proprietary
restrictions, it was found that the reformulated BTCM
provides viable predictions for the available full-scale
database. It follows that the modified Leis correction
factor and – once published – the reformulated
BTCM, are the best methods to predict the arrest for
X-80 steels. However, these methods are not broadly
validated for predictions of the arrest toughness for
X-100 steels, and therefore more work is needed in
the context of such grades. This is particularly the case
in applications where the flow response of the steel
shows limited strain hardening, the extensive presence
of splits, and limited strain to failure – or has limited
through-thickness strength.
The speed-dependent toughness method proposed by
TransCanada considered the effect of fracture speed on
the fracture resistance in terms of CVN or DWTT energy,
which is consistent with the common understanding
that dynamic fracture toughness depends on the strain
rate. While such considerations offer the potential to
improve predictions based on the BTCM through use
of a speed-dependent resistance, it is not clear how to
quantify in general the reference resistance, the reference
speed, and the fracture speed and a constant index.
DWTT energy-based methods
Figure 10 shows broad experimental relations between
the standard PN DWTT energy density and the CVN
energy density for various high-strength pipeline grades.
This figure includes data for X-80 and X-100 high-strength
pipeline steels as developed by Demofonti et al. [13] (for
which the full-scale data were extracted from the CSM
database), along with the linear correlation of Equn 10.
It is apparent that the experimental trend in Fig.10
begins to deviate strongly from linear response at a
CVN energy density of approx. 120 J/cm2 (or 570 ftlb/in2), which corresponds to a FSE CVN energy of
3rd Quarter, 2013
271
io
n
rib
ut
References
1. X.-K.Zhu, 2013. Existing methods in ductile
fracture propagation control for high strength
gas transmission pipelines. Proc. ASME Pressure
Vessel and Pipeline Conference (PVP 2013), July,
Paris, France.
2. W.A.Maxey, 1974. Fracture initiation, propagation,
and arrest. Proc. 5th Symposium on Line Pipe
Research, November, Houston, USA.
3. R.J.Eiber, T.A.Bubenik, and W.A.Maxey, 1993.
Fracture control technology for natural gas pipelines.
NG-18 Report 208, Pipeline Research Council
International, Project PR-3-9113, Battelle.
4. B.N.Leis et al., 1998. Relationship between apparent
Charpy Vee-notch toughness and the corresponding
dynamic crack-propagation resistance. International
Pipeline Conference, Calgary Canada, pp. 723-732.
5. G.Mannucci, G.Demofonti, D.Harris, L.Barsanti,
and H.-G.Hillenbrand, 2001. Fracture properties of
API X100 gas pipeline steels. Proc. 13th Biennial
Joint Technical Meeting on Pipeline Research, 30
April-4 May, New Orleans, USA.
6. G.Wilkowski, D.L.Rudland, H.Xu, and N.Sanderson,
2006. Effect of grade on ductile fracture arrest
criteria for gas pipelines. Proc. International Pipeline
Conference, Canada. Paper IPC006-10350.
7. E.Sugie et al., 1982. A study of shear crack
propagation in gas-pressurized pipelines. J. Pressure
Vessel Technology, 104, p 338.
8. B.N.Leis and R.J.Eiber, 2013. Fracture control
technology for transmission pipelines. Battelle
Report on PRCI Projects PR-003-00108 and PR003-084506 (update of Reference 3).
py
-n
ot
f
Beyond the extent of linearity between the CVN and
DWTT energy densities, the trend of the experimental
results shown in Fig.9 becomes non-linear, and this
bent-over trend follows the same tendency to the
experimental trend presented for the interrupted (precracked) DWTT results shown in Fig.5. In contrast,
the three correlations between DWTT and CVN energy
densities proposed by Wilkowski and by Kawaguchi have
a non-linear dependence that deviated from linearity
in the opposite direction – that is it is bent-up. This
opens to question the utility of Equns 12 to 15: similar
uncertainty exists in regard to Equns 16 and 17.
It was noted that the modified Leis correction coupled
with the BTCM and a newly developed reformulated
BTCM provide viable methods to determine arrest
toughness, at least for high-strength pipeline steels (X80). However, given that such methods are not broadly
validated for predictions of the arrest toughness for
X-100 steels, more work is needed in the context of
such grades. This is particularly the case in applications
where the flow response of the steel shows limited
strain hardening, the extensive presence of splits, and
limited strain to failure – or has limited throughthickness strength.
is
t
In reference to Equn 10, the just-noted limit on
linearity at a CVN E/A of approx. 120 J/cm2 (or
570 ft-lb/in2) corresponds to a PN DWTT E/A of
420 J/cm2 (2000 ft-lb/in2). It is evident in Fig.6 that
the BN DWTT E/A response also begins to deviate
from linearity as a function of CVN E/A at roughly
this same level (i.e. 2000 ft-lb/in2). It follows that
the trends for the PN DWTT and the BN DWTT
do not differ significantly as a function of the CVN
E/A. The onset of this nonlinearity occurs at a FSE
CVN energy of approx. 95 J (or a CVN E/A of
approx. 120 J/cm2 (570 ft-lb/in2)), or a PN DWTT
energy density of 420 J/cm2 (approx. 2000 ft-lb/in2).
This implies that little to no benefit accrues to the
use of the BN (or a PC) DWTT in lieu of a PN
DWTT or a CVN specimen. In turn, this indicates
that the use of such a geometry/test practice will not
offset the inherent calibration issues with the BTCM
as the toughness increases.
The results showed that some of the existing non-linear
models for correlating DWTT and CVN energy densities
are open to question. For example, the trends for the
PN DWTT and the BN DWTT were shown do not
differ significantly as a function of the CVN E/A,
which opens to question the utility of the BN DWTT
practice, and correlations that embed it. In turn, it
was indicated that the use of such a geometry/test
practice will not offset the inherent calibration issues
problems with the BTCM as the toughness increases.
or
d
approx. 95 J (or 70 ft-lb). Leis and Eiber [8] present
an equally extensive dataset in the context of grade
X-65 and below, which also shows this same trend.
pl
e
co
In addition to the discussions and analyses above,
other relevant discussions can be found in a recent
review by Mannucci and Demofonti [31]. In particular,
these authors discussed the applications of the CVN
energy and the DWTT energy in the ductile-fracture
propagation control for X-80 gas transmission line pipes.
m
Conclusions
Sa
This paper discussed work regarding the use of CVN
and DWTT energy in applications to the arrest of
ductile-fracture propagation, and evaluated related
methods to quantify arrest toughness for hightoughness, higher-strength, gas-transmission pipelines.
The BTCM was reviewed along with corrections and
correlations to offset its shortcomings as toughness
increases. These included the Leis correction, the CSM
factor, and Wilkowski’s DWTT correlations, which
were evaluated through comparison with the full-scale
fracture-propagation test data for higher-grade pipeline
steels including X-80 and X-100.
272
The Journal of Pipeline Engineering
or
d
is
t
rib
ut
io
n
21. G.Wilkowski, 1979. Fracture propagation toughness
measurements. Proc. 6th Symposium on Line Pipe
Research, Paper K, Houston, USA.
22.G.D.Fearnehough, D.T.Dickson, and D.G.Jones,
1976. Dynamic toughness determination in ductile
materials. The Dynamic Fracture Toughness
Conference, London, UK.
23.B.N.Leis, 2002. Evolution of line-pipe steel and
its implications for transmission pipeline design.
Proc. International Pipeline Conference, Calgary,
Canada.
24.G.Wilkowski, W.A.Maxey, and R.J.Eiber, 1978.
Problems in using the Charpy, dynamic tear test
and drop weight tear test for high toughness
quenched and tempered steels. In: What does the
Charpy test really tell us? ASM.
25.G.Wilkowski, Y.Y.Wang, and D.L.Rudland, 2000.
Recent development on determining steady-state
dynamic ductile fracture toughness from impact
tests. Proc. 3rd International Pipeline Technical
Conference, Bruges, Belgium, 1, pp 359-386.
26.S.Kawaguchi et al., 2004. Application of X80 in
Japan: fracture control. Proc. 4th International
Conference on Pipeline Technology, Ostend,
Belgium.
27.S.D.Papka et al., 2003. Full-size testing and analysis
of X120 linepipe. Proc. 13th International Offshore
and Polar Engineering Conference, Hawaii, USA.
28.H.Makino et al., 2001. Prediction for crack
propagation and arrest of shear fracture in ultrahigh pressure natural gas pipelines. ISIJ Int., 41,
pp 381-388.
H.Makino, I.Takeuchi, and R.Higuchi, 2009.
29.
Fracture arrestability of high pressure gas
transmission pipelines by high strength pipelines.
Proc. Pipeline Technology Conference, Ostend,
Belgium.
30.R.Higuchi, H.Makino, and I.Takeuchi, 2009. New
concept and test method on running ductile fracture
arrest for high pressure gas pipelines. Proc. 24th
World Gas Conference, Buenos Aires, Argentina.
31.G.Mannucci and G.Demofonti, 2011. Control of
ductile fracture propagation in X80 gas linepipe.
Journal of Pipeline Engineering, 10, pp 133-145.
Sa
m
pl
e
co
py
-n
ot
f
9. W.A.Maxey, J.F.Kiefner, R.J.Eiber, and A.R.Duffy,
1972. Ductile fracture initiation propagation and
arrest in cylindrical vessels. Fracture toughness, ASTM
STP 514, Part II, pp 347-362.
10. W.A.Maxey, J.F.Kiefner, and R.J.Eiber, 1975. Ductile
fracture arrest in gas pipelines. Final Report for
PRC Project NG-18, Report no 100, December.
11.R.J.Eiber, 2008. Fracture propagation – 1: Fracturearrest prediction requires correction factors. Oil &
Gas Journal, 106, 39, 20 October.
12.Idem, 2008. Ibidem, 40, 27 October.
13.G.Demofonti, G.Mannucci, and P.Roovers, 2007.
Existing methods for the evaluation of material
fracture resistance for high grade steel pipelines. Proc.
PRCI-EPRG-APRA 16th Biennial Joint Technical
Meeting on Pipeline Research, March, Canberra,
Australia.
14.J.Wolodko and M.Stephens, 2006. Applicability of
existing models for predicting ductile fracture arrest
in high pressure pipelines. Proc. International Pipeline
Conference, Calgary, Canada. Paper IPC2006-10110.
15.D.L.Rudland and G.Wilkowski, 2007. Effects of
backfill soil properties and pipe grade on ductile
fracture arrest. Proc. PRCI-EPRG-APRA 16th Biennial
Joint Technical Meeting on Pipeline Research,
March, Canberra, Australia.
16.Idem, 2006. The effects of soil properties on the
fracture speeds of propagating axial crack in line
pipe steels. Proc. International Pipeline Conference,
Calgary, Canada. Paper IPC2006-10086.
17.D.M.Duan and J.Zhou, 2009. Speed dependent
fracture toughness and the effect on fast ductile
fracture propagation in gas pipelines. Proc. 12th
International Conference on Fracture, July, Ottawa,
Canada.
18.D.M.Duan, J.Zhou, D.J.Shim, and G.Wilkowski,
2010. Effect of fracture speed on ductile fracture
resistance – Part 2: Results and application. Proc.
8th International Pipeline Conference, Calgary,
Canada.
X.-K.Zhu and B.N.Leis, 2012. Fracture arrest
19.
toughness analysis for high-grade gas pipelines.
Battelle Final Report to Tubular Goods Research
Center of China National Petroleum Cooperation,
August.
20. G.Wilkowski, W.A.Maxey, and R.J.Eiber, 1977. Use
of a brittle notch DWTT specimen to predict
fracture characteristics of line pipe steels. ASME
1977 Energy Technology Conference, Paper 77Pet-21, Houston, USA, September
Are you up to speed?
Training courses – 2013
io
n
0 0 0 2 0 1 2
6th Pipeline Technology Conference (Ostend, Belgium)
20-23 October
2nd Best Practices in Pipeline Operations and Integrity Management Conference
(Bahrain)
11-12 November
DOT Pipeline Safety Regulations - Overview and Guidelines for Compliance
(Houston)
11-12 November
Pipeline Repair Methods, Hot Tapping, and In-Service Welding (Houston)
11-13 November
Pipeline Integrity Management (Houston)
11-15 November
Onshore Pipeline Engineering (Houston)
11-15 November
November Practical pigging training course (Houston)
11-15 November
Subsea Production Systems Engineering (Aberdeen)
11-15 November
Advanced Pipeline Risk Management (Houston)
11-15 November
Stress Corrosion Cracking in Pipelines (Houston)
18-20 November
ot
f
or
d
is
t
Subsea production systems (Houston)
6-9 October
-n
NOV 2013
rib
ut
TRAINING
OCT 2013
1-4 October
2013
18-22 November
Subsea Pipeline Engineering (Houston)
Defect Assessment Calculations Workshop (Houston)
Pigging & In-line Inspection (Calgary)
2-4 December
Defect Assessment in Pipelines (Calgary)
4-6 December
Pipeline Integrity Management (Calgary)
4-6 December
Defect Assessment Calculations Workshop (Calgary)
5-6 December
Geohazard Management for Pipeline Engineers (Calgary)
Sa
m
pl
e
DEC 2013
2-3 December
co
py
20-22 November
Defect Assessment in Pipelines (Houston)
Working with a faculty of 38 leading industry experts, Clarion and Tiratsoo Technical are privileged to provide some of the best
available industry based technical training courses for those working in the oil and gas pipeline industry, both onshore and offshore.
Complete syllabus and registration details for each course are available at:
www.clarion.org
e
pl
m
Sa
py
co
io
n
rib
ut
is
t
or
d
ot
f
-n

Mode I Critical Stress Intensity Factor of Beech Wood (Fagus

Emended diagnosis of Xeronycteris vieirai (Mammalia

application of the , parameters in fatigue crack growth modelling δ δ

TASSEL 3.0 / 4.0 Pipeline Command Line Interface: Guide to using

Sample copy not for distribution - Pipes Pipelines International

Pfizer Pipeline as of November 6, 2014

Pipeline Technology Journal 2/2014

CORK UNIVERSITY HOSPITAL PATHOLOGY USER HANDBOOK

You`re Invited to a Preview of “Against Hate”

Forward-Looking Statements

Ladies and Gentlemen, I am a small business owner in the Charlotte

Pelvic Acetabular Fracture Treatment Solutions - Stryker

PAS / GSK Pharmacist of the Year 2015 Nomination Form

Confession and Absolution

Light tackle tournament puts off turning 50 for another year