Nicholas Ohms Tearing resistance of high strength linepipe steels Academic year 2014-2015 Faculty of Engineering and Architecture Chairman: Prof. dr. ir. Patrick De Baets Department of Mechanical Construction and Production Master of Science in Electromechanical Engineering Master's dissertation submitted in order to obtain the academic degree of Counsellors: Diego Belato Rosado, Dr. Stijn Hertelé Supervisor: Prof. dr. ir. Wim De Waele
70
Embed
Tearing resistance of high strength linepipe steels Nicholas Ohmslib.ugent.be/fulltxt/RUG01/002/224/782/RUG01-002224782_2015_000… · Supervisor: Prof. dr. ir. Wim De Waele. The
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Nicholas Ohms
Tearing resistance of high strength linepipe steels
Academic year 2014-2015Faculty of Engineering and ArchitectureChairman: Prof. dr. ir. Patrick De BaetsDepartment of Mechanical Construction and Production
Master of Science in Electromechanical EngineeringMaster's dissertation submitted in order to obtain the academic degree of
Counsellors: Diego Belato Rosado, Dr. Stijn HerteléSupervisor: Prof. dr. ir. Wim De Waele
The author gives permission to make this master dissertation available for consultation and to
copy parts of this master dissertation for personal use.
In the case of any other use, the copyright terms have to be respected, in particular with regard
to the obligation to state expressly the source when quoting results from this master
dissertation.
Nicholas Ohms, May 2015
i
Acknowledgments First of all I would like to thank my supervisor prof. dr. ir. Wim De Waele for his counseling and
effort he put into this thesis. Especially the time he put into correcting the SCAD paper and his
critical and constructive feedback he gave during presentations and meetings.
A lot of gratitude goes to my counselor ir. Diego Belato Rosado. Without his time and effort, I
would not have been able to deliver this thesis. Even though he is a metallurgical engineer, he
put a lot of energy in studying mechanical aspect of engineering to be able to give excellent
counseling. His support during the tests not only made it possible to perform the tests within a
reasonable amount of time, it also gave welcome company during the long hours spend in the
lab.
I would also like to give thanks to the technical staff of Labo Soete. First of all be because they
flawlessly made the test specimens. Furthermore because they were always prepared to give
help or tips when using machines or equipment within the lab.
Special thanks go to Tony Lefevere for creating the notches with high precision, to Hans Van
Severen for mounting the correct clamps onto the test rigs, for perfectly breaking the specimens
in half for post mortem measurements and for always finding a solutions to whatever
mechanical problem that occurred, to Chris Bonne for his help with electrical related problems,
to Wouter Ost for providing the necessary materials and door key and to Georgette D’Hondt for
taking care of all administrative work.
I would also like to give many thanks to dr. ir. Stijn Hertelé. Without his vast amount of
knowledge and experience in the field of SENT testing, this work would not have been able. His
help with the DIC training is also invaluable. Also ir. Nahuel Micone derserves a word of thanks
for always being prepared to help whenever I had a question.
Thanks also go to my fellow master thesis students at Labo Soete for providing help and often
pleasant chats.
Many thanks go to my friends, who I will not name to avoid hurting someone by forgetting a
name. Their company always gave much needed relaxation.
Finally my utmost gratitude goes to the persons who are closest to me: my parents for giving me
the opportunity to study and for always supporting me no matter the outcome, my sister
Charlotte for her support and good company. And last but not least, my girlfriend Caressa for her
unconditional support during good and bad times and for always believing in me, which gave me
strength when I needed it.
Nicholas Ohms, May 2015
ii
Tearing resistance of high strength linepipe steels
Nicholas Ohms
Supervisor: Prof. dr. ir. Wim De Waele
Counsellors: Diego Belato Rosado, Dr. Stijn Hertelé
Master's dissertation submitted in order to obtain the academic degree of Master of Science in Electromechanical Engineering
Department of Mechanical Construction and Production Chairman: Prof. dr. ir. Patrick De Baets Faculty of Engineering and Architecture Academic year 2014-2015
Summary This thesis investigates the tearing resistance of high strength linepipe steels. This is done by
performing small scale tests on base material specimens. The used test method is the single
edge notch tension (SENT) method. A square bar specimen is obtained from a full pipe and has a
machined notch applied to it in order to simulate a defect. The specimen is then put under
tension to test its resistance to ductile tearing.
The measuring is done using the unloading compliance (UC) technique, which uses the material’s
compliance to estimate the crack growth. This method allows the construction of resistance
curves, which are a critical tool in the strain based design. The UC method has already been used
in previous studies at the UGent and therefore test procedures are readily available. SENT tests
will be performed on 4 different pipeline materials: two spiral formed pipes and two UOE pipes.
For both types of pipes, two different steel grades are examined. For each specimen, two
different notch configurations are tested to investigate the material’s heterogeneity effect on
the Tearing resistance. Digital image correlation (DIC) tests are also performed to evaluate the
strain patterns.
The method delivered accurate resistance curves for most materials. This was validated by
comparing the test results with physically measured values from the fracture surfaces of the
tested SENT specimens. The difference in resistance curves could either be credited to the
material’s tensile properties, the material’s heterogeneity or the microstructural composition of
These defects may grow under increasing load and can
cause the material to fail prematurely. It is economically
impossible to repair all these defects and even practically
impossible to even detect all of them. In order to assess the
severity of the defects, resistance curves (R-curves) need
to be studied. These curves are constructed by putting the
crack tip opening displacement (CTOD) in function of the
ductile crack extension Δa. Theses curves show how well a
material shows resistance to further tearing if an initial
defect is present. Based on these curves, decisions can be
made to what extent certain defects can be ignored,
possible saving millions of dollars. This thesis aims to
determine R-curves for different high strength materials
and based on the results try to determine the main
determining parameters for tearing resistance. The R-
curves are obtained using SENT tests.
II. MEASURING METHODS
A. Unloading compliance (UC)
The unloading compliance method aims to estimate the
ductile crack extension of the SENT specimen. This is
done by measuring the material’s compliance C, since an
increase in compliance indicates crack growth. During the
test, the specimen is partially unloaded and loaded again at
fixed CMOD intervals. From each of these unloading
cycles the compliance is obtained by measuring the slopes
of said unloading cycle (𝐶 =∆𝐶𝑀𝑂𝐷
∆𝐹). The crack depth can
then be calculated from these compliance values using
analytical formulas. The CMOD and CTOD values are
obtained by using the double clip gauge method.
B. Digital image correlation (DIC)
DIC allows the calculations of the specimen’s strains by
tracking the surface displacements by using cameras.
Specialized software is able to track deformations in a
speckle pattern which needs to be applied to the material’s
surface. The cameras are programmed to take pictures at
the top of a loading cycle and at the bottom of an
unloading cycle. The software is then able to compare
these obtained images to a reference image. From this the
strain fields can be calculated and visualized.
C. Post mortem fracture surface measurements
In order to evaluate the accuracy of the tests, physical
measurements need to be made. This is done by measuring
iv
the fracture surfaces of the tested specimens. The
specimens are broken in two to make microscopic pictures
of the fracture surfaces. Prior to breaking, the specimens
need to be heat tinted to create a color difference between
the ductile tear and the rest fracture after breaking. The
crack extension can be measured using the nine points
average method.
III. EVALUATION METHODS
A total of 24 SENT specimens are tested using the
unloading compliance method. There are four different
pipe materials. Two API-5L X70 steel grade pipes and two
API-5L X80 steel grade pipes. For each of these grades, a
spiral formed pipe and a UOE formed pipe are taken. For
each material two different notch locations are studied and
for each notch location three tests are performed, giving
the total of 24 tests. To the results of these tests, R-curves
were fitted according to the ASTM E1820 standard. The
obtained R-curves of the X80 spiral pipe were discarded
due to extreme high scatter in the obtained test values. The
other three materials gave satisfying results. Comparison
with the measured values gave an average accuracy of
0.28mm (9%).
IV. INFLUENCES ON THE R-CURVES
A. Tensile properties
The results showed a difference between the different
materials. The mean cause for these difference can be
found in the tensile properties of the different materials.
The results for the TT configuration are given in Figure 1.
Figure 1 Resistance curves for the TT configuration of the 7S, 7U
and 8U material.
The 7U material has the best tearing resistance even
though it is the weakest material stress wise. It does
however have the lowest yield-to-tensile ratio (0.77) and
the highest uniform elongation (10.2%), which are
favorable values for the tearing resistance. The 7S material
scored much worse due to discontinuous yielding
behavior. This material has a Lüders plateau in its stress-
strain curve which is detrimental for the tearing resistance.
The X80 spiral pipe, which gave non-consistent R-curves,
has a high yield-to-tensile ratio (0.91). This causes a lot of
local deformation around the crack tip, which distorts the
CTOD and compliance measurements, explaining the high
scatter.
B. Heterogeneity
Not only a difference between the different materials
was noticed, the different notch configurations of the same
material also gave different R-curves. Figure 2
demonstrates the difference between the two notch
configurations for steel 8U.
Figure 2. Resistance curves for the TT and ID notch
configuration for the 8U material.
This difference is caused by heterogeneous material
properties within the specimen. In order to analyze this
heterogeneity, hardness maps have been made of each
material. The hardness map of this material showed a
softer central region between harder outer layers, probably
caused by the rolling process of the steel plates from which
the pipe was made. The ID notch crack propagates into
this softer region, which has a lower tearing resistance.
The TT notch also traverses this region, but only partially.
A large part of the crack front also traverses the harder
zones, which yield an overall better tearing resistance. The
7S and 8S also showed a softer central region. The 7U
however did not have outspoken hardness regions.
V. DIC RESULTS
The DIC method was used to validate the CTOD
measurements obtained by the double clip gauge method.
This was done by measuring the δ5 definition of the CTOD
with the software. Both obtained values were in good
agreement with each other. Although the CTOD obtained
from the clip gauges was larger for higher CTOD values,
but this is an inherent trait of the measuring method. An
attempt to analyze the blunting using the DIC data did not
yield usable results, because the blunting effect is a very
small phenomenon located around the crack tip.
ACKNOWLEDGEMENTS
The author would like to acknowledge the support of
prof. dr. ir. Wim De Waele, ir. Diego Belato Rosado and
the technical staff of Labo Soete for their help with either
theoretical or practical work.
v
List of contents Acknowledgments .......................................................................................................................................... i
Summary ....................................................................................................................................................... ii
List of contents .............................................................................................................................................. v
List of symbols ............................................................................................................................................. vii
1.3.4 SENT test............................................................................................................................... 10
2 Test methods ....................................................................................................................................... 13
3 Test specimens .................................................................................................................................... 24
3.1 Pipe forming process ................................................................................................................... 24
3.4 Test rigs ........................................................................................................................................ 29
4 Evaluation of the test results .............................................................................................................. 29
List of figures ............................................................................................................................................... 58
List of tables ................................................................................................................................................ 60
vii
List of symbols a Crack size
a0 Initial crack depth
a0,9p Initial crack depth measured by nine points average method
a0,meas Total measured initial crack size
a0q Original crack size estimated from compliance
af Final crack length
af,9p Final crack length measured by nine points average method
API American Petroleum Institute
ASTM American Society for Testing and Materials
B Thickness of SENT specimen
Be Effective thickness of SENT specimen
BN Net thickness of SENT specimen
BSI British Standard Institution
C Compliance
C1, C2 Constants used in the calculation of a0q
Cc Compliance corrected for rotation and necking
CCT Center crack tension
CDF Crack driving force
CL Center line
CMOD Crack mouth opening displacement
CT Compact tension
CTOD Crack tip opening displacement
CTODini Crack tip opening displacement at crack initiation
dF Offset between remote and ligament force
DCPD Direct current potential drop
DENT Double edge notch tension
DIC Digital image correlation
em Uniform elongation
E Young’s modulus
E’ young’s modulus corrected by Poisson ratio
EPFM Elastic plastic fracture mechanics
F Force
Fmax Maximum force
Fr Rotation and necking correction factor
Fres Resulting reaction force in remaining ligament
h Stress triaxiality
h1 Height of lower clip gauge
h2 Height of upper clip gauge
H Daylight gripping length of SENT specimen
ID Inner diameter
J J-integral
L Total length of SENT specimen
LEFM Linear elastic fracture mechanics
m Parameter in function of the material and the configuration
viii
Mres Resulting bending moment
MAG Metal active gas welding
OD Outer diameter
OM Optical microscopy
P Load
PY Limit load
Pm Maximum load
ri Constant used for crack size estimation
Rt;0.5 Stress corresponding to a plastic strain of 0.5%
SAW Submerged arc welding
SENB Single edge notch bending
SENT Single edge notch tension
SMYS Specified minimum yield strength
STD Standard deviation
TT Through thickness
U Dimensionless parameter for crack size estimation
UC Unloading compliance
UGent Ghent University
UOE U bending, O forming and expansion
V1 Displacement measured by lower clip gauge
V2 Displacement measured by upper clip gauge
W Width of SENT specimen
W.T. Wall thickness
Y/T Yield-to-tensile ratio
αδ Curve fitting parameter for CTOD resistance curve
δ CTOD obtained from the double clip gauge method
δ5 CTOD measured using two point located 5 mm apart, cross the crack tip
δ90 CTOD evaluated using double clip gauge and 90° intercept definition
δS Scatter on tearing resistance curve
Δa Amount of ductile crack extension
Δa9p Ductile crack extension measured by nine points average method
Δab Blunting distance
Δab,9p Blunting distance measured by nine points average method
ΔaUC Ductile crack extension measured by unloading compliance
εmax Tensile strain capacity
ηδ Curve fitting parameter for CTOD resistance curve
ν Poisson’s coefficient
σ0 Yield strength (0.2% offset)
σe Equivalent Von Mises stress
σm Hydrostatic stress
σTS Ultimate tensile strength
σY Effective yield strength
1
1 Introduction
1.1 Situating the subject
Figure 1. World energy consumption by source in 2012 [1].
Although a lot of effort is put in the development of renewable and cleaner energy, the vast majority of
the world’s energy is still obtained from non-renewable sources such as fossil fuels. Figure 1 shows that
less than 10% of the world energy consumption comes from renewable sources, whereas 87% comes
from fossil fuels. The constant increase of the world population and the rise of new economies give a
growing demand for these resources. As current resource sites are becoming depleted, a constant
search for new sources is being conducted. The high demand and increasing scarcity also cause a price
increase for fossil fuels. The combination of these factors makes it economically justified to search for
these resources in remoter areas. The transportation of the resources from the origin to the refinery is
mainly done by pipelines. To access remote locations, these pipelines often have to traverse harsh
environments. These environments include discontinuous permafrost, offshore ice gouging, seismic
active areas, landslides and high temperature/pressure operations [2]. Most of these events are related
to displacements, which cause large plastic deformations.
Figure 2. Trans-Alaska pipe line traversing an arctic region [3].
1.2 Strain based design The traditional approach for designing pipelines is stress based. The hoop stress in the pipe, which is
caused by the internal pressure, is limited to a certain design stress. This design stress is taken at a
2
certain percentage of the minimum yield strength (typically taken at 72% or 80% [4]). The yield stress is
often defined as the stress that occurs at 0.5% total strain (Rt;0.5) [5]. If we take into account that the
yield-to-tensile ratio (Y/T) of pipelines is often around 0.8 to 0.9, then the design stress of the pipeline
rarely exceeds the elastic limit of the material. This is schematically shown in Figure 3 (a).
(a) (b)
Figure 3. Safety margin for stress based design (a) and safety margin for strain based design (b).
As mentioned above, pipelines linking remote locations often traverse harsh environments which can
cause large plastic deformations. Using the classic stress-based design in such cases is usually overly
conservative and brings unnecessary high costs. Because of this, interest has shifted towards strain-
based design
In strain-based design, the longitudinal strain capacity, in addition to transverse yield strength is used as
a measure for design safety [2]. This allows for more flexible design because the focus is on the pipe’s
strain capacity rather than the stress capacity. A main objective of this design is to ensure that the safety
and reliability is comparable to conventional stress-based designed pipelines. The strain-based design
approach is schematically displayed in Figure 3 (b).
1.3 Sub-scale testing
1.3.1 Tearing resistance
Although a certain pipe material may theoretically meet up to the design requirements, the production,
installation and connection (through welding) of the pipe segments may introduce certain flaws into the
material. Under loading conditions, these flaws can grow and eventually lead to failure. It is
economically impossible to repair all these flaws. It is furthermore impossible to even detect al the
flaws. Because of this, the materials need to be tested to determine to which extent certain flaws can be
allowed. Although full-scale tests on pipeline segments would be the most representative, they are far
too expensive and impractical. Therefore several sub-scale tests have been developed. These sub-scale
tests aim to test certain parameter of the material. For strain-based design, the most important
parameter is the so called tearing resistance (or crack growth resistance) [2]. The tearing resistance is a
material property that is influenced by type of loading and the material geometry. It gives a measure of
3
how well a certain material gives resistance to further tearing under load if an initial flaw is present. This
can be displayed by the use of resistance curves or R-curves, which are discussed in section 1.3.2.5.
1.3.2 Fracture mechanics
The field of fracture mechanics studies material behavior in the vicinity of cracks and notches. Linear
elastic fracture mechanics (LEFM) assumes that the stresses around the crack do not exceed the yield
stress of the material. Therefore, no plasticity is present around the crack. This is mainly the case for
brittle materials or fatigue problems. Tearing resistance tests for pipeline steel will have plasticity
around the crack tips. Therefore elastic plastic fracture mechanics (EPFM) needs to be applied. Within
this field, there are two important parameters which determine the material’s toughness: the crack tip
opening displacement (CTOD) and the J-integral [6]. Both are discussed below.
1.3.2.1 Crack tip opening displacement (CTOD)
The crack tip opening displacement can be used as a measure for a material’s toughness. The toughness
determines how a material behaves around a flaw. It is a measure for how well a material is resistant to
an unstable crack growth and thus to brittle tearing. CTOD is the distance between the two crack flanks
at the height of the crack tip. When there is plasticity around the crack tip during loading, the crack
flanks will move away from each other without an increase of the actual crack. How much these flanks
move without further tearing, is a measure for the material’s toughness. This plastic deformation
around the crack tip reduces the sharpness of the crack and the phenomenon is therefore called
blunting. This is displayed in Figure 4. Eventually, the CTOD will reach a critical value at which the crack
will start to grow. This is denoted as CTODini, because the crack growth initiates at this point.
Figure 4. Blunting of the crack tip [7].
There are different ways to measure the CTOD, depending on where the measurement is made. In this
thesis, only two methods are considered. The first one is the 90° intercept method. The CTOD or δ90 is
determined by two lines that are at an angle of 90° and originate from the initial crack tip. The distance
between the two intersection points of these lines with the crack flanks determines the CTOD length.
The second method is the so called δ5 method. The CTOD is measured by two points at 5mm distance
from each other across the crack tip. Both methods are displayed in Figure 5. Unless stated otherwise,
the 90° intercept method will always be used when referred to CTOD or δ.
Another definition to quantify the crack opening is the crack mouth opening displacement (CMOD). For
CMOD, the distance between the crack flanks is measured at the crack mouth.
4
Figure 5. Schematic representation of the δ90 and the δ5 definitions for CTOD [10].
1.3.2.2 J-integral
A second parameter to determine a material’s toughness is the J-integral. The J-integral is determined
by a contour integral with a random path around the crack tip. This is schematically shown in Figure 6.
The J-integral can be interpreted as the energy stored around the crack tip that is available for crack
propagation [7]. The J-integral is a more theoretical parameter and can only be calculated using
approximating formulas.
Figure 6. Example of a random contour integral around the crack tip to determine the J-integral [7].
1.3.2.3 CTOD vs J-integral
Shih et al. showed that there is a direct theoretical relation between the CTOD and the J-integral [8]:
𝐽 = 𝑚𝜎0𝐶𝑇𝑂𝐷 (1)
Here m stands for a constant that is dependent on the material and configuration and σ0 stands for the
material’s yield strength.
Since both parameters are linked, using one of them suffices to determine the material’s toughness.
CTOD has the big advantage that it is a physical parameter and can be easily measured. J-integral needs
the usage of approximating formulas and is therefore less hands-on. On top of that, the J-integral is
limited to zones of plasticity that don’t reach the specimen’s edges. Because of all the aforementioned
reasons, the J-integral will not be used in this thesis and thus the CTOD will be applied to determine the
material’s toughness.
1.3.2.4 Crack driving force curve
The crack driving force (CDF) is the force required for a crack to grow. The CDF curve can be expressed in
terms of CTOD or J. The curve is mostly in function of the crack growth Δa or the applied load (which is
5
the strain in this case). The CDF curve is obtained from finite element simulations or analytical formulas.
The curve is influenced by geometrical parameters, material parameters and loading conditions.
1.3.2.5 Resistance curve
To determine the fracture toughness of a material, resistance curves or R-curves can be evaluated. R-
curves are constructed by plotting the CTOD (or J) in function of the crack growth Δa. An example of
CTOD R-curves is given in Figure 7.
Figure 7. Schematic representation of different resistance curves.
R-curves are only useful if the material exhibits stable tearing behavior. In ductile materials, plasticity
occurs around the crack tip when the material is loaded. Further loading will increase the crack and so
will the plastic area. Therefore the R-curve shows a rising trend for these materials.
How a crack propagates in ductile materials is displayed in Figure 8. When a load is applied to the crack,
the crack tip will start to blunt to accommodate the high strains around it. At a certain value of the
strain around the crack, micro cracks or voids will appear near the crack tip. These will grow under
continued loading until a certain critical value is reached. At this point the voids will coalesce and the
crack tip will propagate. Further loading will create new voids around the newly formed crack tip and
will eventually coalesce again. This will continue and give the stable crack growth phenomenon [9].
6
Figure 8. Schematic representation of blunting and crack initiation [9].
In order to check if a certain material will exhibit stable crack propagation, the tearing resistance curve
and the crack driving force curves can be combined. This is called the tangency approach in literature.
The principle is shown in Figure 10. Here the crack driving force is expressed as CTOD in function of the
crack growth Δa. Different curves are displayed for fixed strain values. These are called iso-strain CDF
curves. The intersection between the CDF curve and the R-curve shows the amount of ductile tearing for
a given strain. The value of strain at which both curves are tangent determines the tensile strain
capacity εmax of the material. From this point on, unstable crack growth occurs. For strains higher than
εmax, stable crack growth is no longer possible [10].
Figure 9. Schematic representation of the tangency method [10].
7
1.3.3 Specimen configuration
As mentioned above, different sub-scale configurations have been developed in order to obtain the
resistance curve of different materials. All the tests consist of a specimen to which a notch has been
applied to simulate a surface defect. The most frequently used configurations are listed below, [11, 12].
Compact tension (CT): This is one of the older used tests. A relative small piece of material
has a notch at its center. Holes have been applied at both ends of the notch and the material
is put to a tensile load via pins through the holes. This is schematically shown in Figure 10.
Figure 10. Schematic representation of a compact tension (CT) specimen [11].
Single edge notch bending (SENB): In this test, the specimen has a notch applied to one of its
edges. The specimen is then subjected to a three point bending load where the notch should
be located above the bending point. This is displayed in
Figure 11.
Figure 11. Schematic representation of a single edge notch bending (SENB) specimen [11].
Single edge notch tension (SENT): SENT specimens have the same notch configuration as
SENB tests, but they are put under a tensional load instead of a bending load. There are two
different boundary conditions for this test. The rotation of the specimen can be free if it is
pin loaded or restricted if the specimen is clamped. Both configurations are displayed in
Figure 11.
8
Figure 12. Schematic representation of a single edge notch tension (SENT) specimen with pin loaded configuration (a) and clamped configuration (b) [11].
Double edge notch tension (DENT): Just like the SENT specimens, this notched specimen is
put under a tensile load. Instead of one notch, the specimen has two notches at opposite
sides of the specimen. Both the pin loaded boundary condition as the clamped condition are
possible. See Figure 13 for a schematic representation.
Figure 13. Schematic representation of a double edge notch tension (DENT) specimen [11].
Center cracked tension (CCT): A CCT specimen has a notch in the center of specimen rather
than one at the sides. This specimen is also subjected to a tensile load. This is shown in
Figure 14.
Figure 14. Schematic representation of a center cracked tension (CCT) specimen [11].
(a) (b)
9
Although all these different tests try to determine the tearing resistance of the material, they will all give
a different tearing resistance curve. For example, the difference in R-curves between a specimen under
a bending load and a tensile load is given in Figure 15.
Figure 15. Difference in resistance curve for identical specimens under a bending and a tensile load.
To determine which test most closely relates to reality, certain parameters need to be compared. This
can be done by looking at the constraint parameter of the specimens. This parameter determines how
the stress fields are distributed around the crack tip. One way to quantify the constraint is by looking at
the stress triaxiality h. Stress triaxiality is defined as the ratio of the hydrostatic stress σm to the
equivalent Von Mises stress σe [13]:
ℎ =𝜎𝑚
𝜎𝑒 (2)
The hydrostatic stress symbolizes the pressure that acts all around the specimen without inducing
deformation. The equivalent Von Mises stress characterizes a multidimensional stress by which a
specimen plastically deforms. The way these two stresses relate to each other determines the level of
constraint. A high constraint value (σm >> σe) means that there is little plasticity created during
deformation which indicates a more brittle behavior. This also gives rise to a lower R-curve. A high
constraint value (σm << σe) on the other hand, gives the appearance that the specimen has a higher
toughness and will yield higher R-curves [14].
For the specimen configurations mentioned here. CT and deeply notched SENB specimens show no
decrease in constraint when they are loaded. These specimens are therefore termed as high constraint
specimens. This lack of constraint decrease is in contrast with tests performed on full pipeline segments,
which show a decrease in constraint under load. These specimens should therefore not be used in the
assessment for strain based design. CCT specimens on the other hand show a much higher drop in
constraint under load compared to full pipelines, so these specimens should also be avoided within this
field. Shallow notched SENB do show a drop in constraint under load as opposed to their deeply notched
counterparts, although this is still less than for SENT specimens. Therefore SENB R-curves are over-
conservative (see Figure 15). SENT and DENT specimens approach the constraint evolution of real
10
pipelines the most, although they are a bit more conservative. DENT specimens however are less
favorable because there is need for the monitoring of two notches, which is less practical [15-19].
1.3.4 SENT test
In the previous paragraph it was discussed that SENT tests most closely relate to real pipeline tearing
behavior. Therefore, only SENT tests will be studied in this thesis and a more detailed overview of the
used SENT specimen configuration will be given. First off, Figure 16 shows how a SENT specimen is
obtained from a pipe.
Figure 16 Schematic representation of a SENT specimen obtained from a pipe.
The specimen is a rectangular bar with width W, thickness B and length L. The specimen has a machined
notch with depth a0, which is called the initial crack depth. The length between the gripping points is
called the daylight gripping length H. See Figure 17 for a schematic overview of the dimensional
parameters.
Figure 17. Dimensional parameters of a clamped SENT specimen [20].
As was discussed in the previous paragraph, there are two ways to fix the specimen: pin-loaded and
clamped. The pin-loaded method has the advantage that the specimen length is not influential on the
results because there is no restriction on the rotation of the specimen. For clamped specimens on the
other hand, the length of the specimen may have an influence on the measurements due to the rotation
of the specimen being hindered. More precisely, the daylight gripping length H of the specimen
11
determines the effect and should be taken long enough to eliminate the influence of the gripping. The
total length L should therefore be of sufficient length to present enough clamping area as well as an
adequate daylight gripping length. The pin-loaded method has the drawback that the boundary
conditions are harder to define. Also, the method is influenced by the pin locations. Given the fact that it
is rather difficult to make ideally pin-loaded specimens, the locations will vary from specimen to
specimen which is detrimental for consistency [20, 21]. Therefore the clamped specimen is preferred as
fixing method, of course given that the daylight gripping length is chosen long enough. In this thesis,
clamped specimens will therefore be used.
Shen et al. [22] showed that a daylight gripping length H=10W gives a good similarity towards a
circumferential crack in a pipe subjected to tensile loading. Therefore this is generally the used value for
H. This relation will also be applied in this thesis.
Literature reports the use of square (B=W) and over-square (B=2W) configurations with regard to the
cross section. Although they both yield the same results in terms of resistance curves, the smaller cross
section of the square configuration requires approximately half the test capacity compared to the over-
square configuration [24]. The square cross section configuration will be applied on all tests performed
here.
Another important parameter is the initial notch a0. This parameter is mostly taken relative to the width
W. DNV-RP-F108 [20] allows a range of 0.2 ≤ a0/W ≤ 0.5. Values for a0/W around 0.2 are referred to as
shallow notches and values around 0.5 are called deep notches. Moore et al. [25] showed that a relative
initial crack depth a0/W=0.3 gives accurate CTOD measurements. All specimens in this dissertation will
therefore have a relative initial crack depth a0/W=0.3.
The notch can be made as is described in the ASTM E1820 standard for SENB specimens [26]. The notch
is first saw-cut to a certain depth, which is less than the desired depth. The notch should then be fatigue
pre-cracked until the crack grows to the required depth a0. This will create an infinitely sharp crack tip. It
is however quite difficult to control this cracking. Therefore a very small blade (150µm) can be used to
cut the final part of the notch as is described by the ExxonMobil SENT test procedure [23]. Although this
does not give an infinitely sharp crack (but rather a 0.075mm root radius as described by the
procedure), this problem is overcome when the material is loaded, given that it is a ductile material. As
was mentioned earlier, the (fatigued) crack will start to blunt in the beginning of loading. The cut notch
will also blunt and eventually the same state will be achieved as the fatigue pre-cracked form. This is
schematically shown in Figure 18.
12
Figure 18. Schematic representation of a fatigued and a machined crack [27].
Lastly, the use of V-shaped side grooves is advised [28]. Without side grooves, so called crack tunneling
occurs. This tunneling effect gives a parabolic shaped crack front due to the fact that the mid-section of
the specimen is in a plane strain state and the side of the specimen is in a plane stress state [29].
Riemelmoser et al. [30] showed that the crack tunneling has a negative effect on the accuracy of the
crack growth measurements. By machining two V-shaped side grooves at both sides of the notch, the
side of the specimen will also be in a plane strain condition and there will be a more uniform crack
growth. Shen et al. [28] advises side groves with a depth of 0.075B, giving a total reduction of 15% to the
thickness B. The new thickness is called the net thickness BN=0.85B. These side grooves will be applied to
the specimens used in this thesis and will also have dimensions according to ASTM E1820 [26], i.e.
opening angle less than 90° and a root radius of 0.5 ± 0.2 mm. The side groove configuration is
schematically shown in Figure 19.
Figure 19. Detailed view of a side groove.
13
2 Test methods SENT testing has gained a lot of interest over the last couple of years because its testing results closely
relate to the full size pipelines. Much effort is being put in the making of a standard, like there exists for
the other mentioned specimens (for example the ASTM E1820 standard for SENB [26]). During the
making of this thesis, the British Standard Institution (BSI) published a first SENT testing standard,
namely the BS 8571:2014 [31]. Apart from that, many institutions and companies made efforts to create
procedures for SENT testing. Many of these procedures are based on the ASTM E1820 standard for SENB
specimens [26]. A lot of research is still being conducted to further improve these procedures. This
thesis makes use of a procedure developed by BMT fleet technology.
In order to obtain the resistance curves of materials, the ductile crack extension needs to be measured.
There are two common used methods to do this. The first is the Direct Current Potential Drop method
(DCPD). This method measures the voltage over the crack by sending a direct current through the
specimen. As the crack grows the remaining cross section of the specimen decreases and thus the
resistance increases. By monitoring the voltage over the crack, the crack growth can be estimated. A
second method is the Unloading Compliance method (UC). This method makes use of the compliance of
a specimen, which changes when the crack grows. Matthias et al. [11] showed that both methods are
quite equivalent as they similarly indicate the crack initiation and have a comparable accuracy. Because
the DCPD method works with very small voltages, it is very susceptible to noise. Therefore the UC
method has been chosen as the measuring method in this thesis. Next to the crack propagation
measurements, Digital Image Correlation (DIC) measurements will be performed to obtain and visualize
the strain fields around the crack. Finally a post mortem analysis will be done by means of final crack
measurements.
2.1 CTOD measurement The CTOD is a key parameter in the construction of the resistance curves. A method that has gained a lot
of interest for measuring CTOD is the so called double clip gauge method [25, 32]. Two clip gauges are
placed over the crack at different heights. These clip gauges are mounted onto specially designed knives
which are fixed to the specimen at each side of the notch. See Figure 20 for a schematic overview. The
first clip gauge is at height h1 from the surface and measures distance V1 across the notch. The second
clip gauge is at height h2 from the surface and measures distance V2 across the notch. The CTOD can
then be measured using the following formula:
𝛿90 = 𝑉1 −ℎ1+𝑎0
ℎ2−ℎ1(𝑉2 − 𝑉1) (3)
14
Figure 20. Schematic representation of the clip gauge blades for measuring the CTOD.
2.2 Unloading compliance
2.2.1 Principle
The unloading compliance method estimates the crack growth by using the material’s compliance
parameter C. The compliance is the inverse of the specimen’s stiffness and is expressed in mm/kN.
There is a direct relation assumed between the compliance and the crack size. An increase in the
compliance indicates crack growth. A calculation for the compliance is proposed by Shen et al. [33]. The
specimen is strain loaded and unloaded at predetermined intervals of CMOD (crack mouth opening
displacement, see paragraph 1.3.2.1). The compliance can then be measured by determining the slope
of a linear regression line through the unloading phase (see Figure 21). The formula for compliance
reads:
𝐶 =∆𝐶𝑀𝑂𝐷
∆𝐹 (4)
Here ΔCMOD is the change in CMOD for a given change in force ΔF. The force F is obtained from a load
cell on the test rig and the CMOD is obtained from the double clip gauge method as was described in the
previous paragraph. The CMOD is obtained from the clip gauges as follows:
𝐶𝑀𝑂𝐷 = 𝑉1 −ℎ1
ℎ2−ℎ1(𝑉2 − 𝑉1) (5)
(a) (b)
Figure 21. Example of a typical unloading compliance curve (a) and the determination of the compliance from the unloading slope
(b).
15
2.2.2 Crack size calculation
As was mentioned earlier, the procedure that is used here was developed by BMT. This procedure is
mainly based on common used methods. The main principles will be discussed here. The specimen is
first loaded (displacement controlled) within the elastic zone and then unloaded again. This is done five
times. After this, the specimen is loaded into the plastic zone. At predetermined CMOD increases, the
specimen is partially unloaded. At each unloading cycle, the compliance is measured. These loading and
unloading cycles are continued until the force of a loading cycle drops below 80% of the maximum
occurred force during the test.
The crack growth can be calculated from the compliance by using analytical formulas as proposed by
Shen et al. [33]. It is calculated from the following nth-order polynomial:
𝑎
𝑊= ∑ 𝑟𝑖𝑈
𝑖𝑛𝑖=0 (6)
In equation (6) a/W represents the relative crack depth, which is the crack length a normalized by the
specimen width W. The terms ri are predetermined constants and U is a dimensionless parameter
defined by the following equation:
𝑈 =1
1+√𝐵𝑒𝐶𝐸 (7)
Here C is the compliance, E is the specimens Young’s modulus and Be is the effective thickness as
described by ASTM E1820 [26]:
𝐵𝑒 = 𝐵 −(𝐵−𝐵𝑁)
𝐵
2 (8)
B is the specimen’s thickness and BN is the net thickness which is the thickness minus the side grooves:
𝐵𝑁 = 0.85𝐵 (9)
2.2.3 Polynomial determination
The relative crack depth is determined by three input parameters, i.e. a/W=f(B,C,E). Equation (6)
transforms the dimensionless parameter U to the relative crack depth by using an nth-order polynomial.
In literature, there are many polynomials defined for this purpose, each with differing orders and
constants ri. Wang et al. [34] studied the six most used CMOD based polynomials of the form of
equation (6). They concluded that the polynomials proposed by Shen et al. [21] and by Cravero et al.
[35] are the most accurate of the six. Although both authors use the same form of polynomial, they have
different assumptions. The most important assumptions are listed in Table 1.
The first difference between the two is that Cravero assumes a plane strain state of the specimen
whereas Shen assumes a plane stress state. As a result different definitions of the Young’s modulus E in
equation (7) are used. For plane stress, the conventional modulus E is used. For plane strain on the
other hand, the elastic modulus corresponding to plane strain E’ is used. E’ is the normal elastic modulus
E corrected by Poisson’s ratio ν. Although the material is at a plane strain condition at the center, the
sides are in a plane stress condition. Since the compliance is a function of displacements everywhere in
the specimen and not just around the crack tip, Tyson et al. [36] concluded that it is best to use the
16
plane stress condition, because the constraint for the major part of specimens in tension is closer to
plane stress.
Table 1. Assumptions of the CMOD based compliance equations.
Cravero (2007) Shen (2009)
H=10W H=10W
5th order polynomial 8th order polynomial
Clamped Clamped
Not specified W=B
Plane strain Plane stress
E' = E/(1-ν²) E
0.1≤a/W≤0.7 0.05≤a/W≤0.95
A second difference lies in the order of the polynomial and accompanying coefficients ri. The values for
the coefficients ri are given in Table 2. These values are independent of the width-to-thickness ratio
W/B, the relative crack depth a/W or the assumed state (plane stress or plane strain).
A third difference lies in the applicability area where the polynomial is accurate. The 5th order
polynomial is valid for values of a/W between 0.1 and 0.7. The 8th order polynomial, although more
cumbersome, does have a larger interval of validity (0.05≤a/W≤0.95). A comparative study performed
by Tyson et al. [36], determined a very small difference between the two definitions in an interval of
0.1≤a/W≤0.8. Wang et al. [34] also noted a small difference between the two methods, although they
concluded that the 5th order polynomial as proposed by Cravero was the most precise.
A comparative study between the two equations has been made for specimens used in this thesis. The
crack growth measurements for two specimens (for material details, see chapter 3) with different notch
locations (with respect to the pipe orientation) are calculated by using the different polynomials in
equation (6). Figure 22 shows the comparison between the obtained relative crack depths a/W for the
two polynomials. For the whole crack growth, the values obtained by Shen’s formula are lower than
those obtained by Cravero’s. This is the case for both notch locations (see Figure 22 (a) and (b)). The
difference between the two values, however, is very small (less than 1% difference between all obtained
values). Furthermore, the difference gets gradually smaller for higher a/W values.
17
(a) (b)
Figure 22. Calculated crack depth comparison for 7S-TT specimen (a) and 7S-ID specimen (b).
As can be seen above, the differences are very small. This is in accordance to the observations made by
Tyson et al. [36] and Wang et al. [34]. Since the difference is very small, choosing one polynomial or the
other will not affect the outcome of the calculations by much. Therefore either formula can be used.
Since the polynomial as proposed by Cravero is of lesser order, the use of it is less cumbersome. For this
reason, Cravero’s polynomial will be used for further calculations in this thesis.
2.2.4 Rotation effect and necking
As was mentioned earlier, a growing crack size is linked to an increase in the specimen’s compliance.
Therefore one would logically expect a constant rise in compliance when the CMOD is increasing.
However, in many cases the compliance dropped for low CMOD values. A drop in the compliance would
indicate a negative crack growth, which is of course not possible. This drop can be seen in Figure 23 (a).
(a) (b)
Figure 23. Negative crack growth in a compliance curve (a) and the influence of relative crack size on initial compliance decrease
[11].
The phenomenon was explained by Mathias et al. [11]. The drop in compliance for low CMOD levels is
caused by the rotation effect. This rotation is caused by an offset between the center line of the
remotely applied force F and the centroid of force in the remaining ligament of the specimen Fres. This
difference is illustrated in Figure 24. The applied force F is applied evenly at both ends of the specimen,
18
therefore its load line runs at the center of the specimen. The area at the notch is reduced. The stress
distribution at this location is not uniform, but rather asymptotical. The resulting reaction force Fres is
therefore shifted with respect to the center line. The offset dF of the lines creates a moment Mres in the
specimen. This moment causes a drop in compliance at low CMOD values.
This phenomenon is more outspoken for deeply notches specimens (e.g. a/w=0.5), because the offset
dF is quite large. As the initial notch depth is reduced, the lines will move closer to each other and the
resulting moment Mres will reduce. For specimens with a shallow notch (i.e. a/W=0.2) rotation is no
longer present. The effect of rotation on the compliance is displayed in Figure 23 (b) for different
relative notch depths.
Figure 24. Schematic representation of the force line offset causing the rotation effect [11].
A second phenomenon that is observed is necking. Necking causes a reduction of the remaining
ligament which in turn gives an increase in the compliance. Necking mostly occurs after the maximum
force has been reached, although for shallow notched specimens, necking already has an effect earlier
on. This is caused by the blunting of the crack tip which causes necking in the vicinity of the crack tip.
2.2.5 Crack initiation and blunting
The equations that were considered in paragraph 2.2.2 assume a monotonically increasing relationship
between CMOD and the compliance. Since the compliance often decreases for small CMOD values, the
formulas used to calculate the crack length (equation (6) and (7)) will yield a decreasing value for the
crack depth. As was mentioned before, this has no physical meaning. Cauwelier et al. [12] determined
that the moment at which the compliance curve rises again corresponds to the moment where the crack
starts to grow, i.e. crack initiation. Prior to this moment, blunting occurs. For deeply notched values,
crack initiation occurs when the compliance curve reaches its minimum (Figure 25 (a)). Shallow notched
specimens, however, do not have a decrease in compliance and therefore have no minimum. The
compliance does show a linear rise in the beginning. Crack initiation is believed to start when the curve
starts to deviate from this linear trend (see Figure 25 (b)).
19
(a) (b)
Figure 25. Crack point initiation for deeply notched specimens (a) and for shallow notched specimens (b) [27].
Shen et al. [21] formulated a correction factor that corrects the compliance for the rotation and necking
effect. The compliance corrected for rotation and necking Cc is obtained from the following equation:
𝐶𝑐(𝑖) =𝐶𝑖
𝐹𝑟 (10)
Here, Cc(i) is the corrected compliance for the ith cycle, Ci is the measured compliance for the ith cycle and
Fr is the rotation and necking correction factor. Fr is calculated using the following equation:
𝐹𝑟 = 1 − 0.165𝑎0
𝑊
𝑃(𝑖)
𝑃𝑌 (11)
a0/W is the initial relative crack depth, P(i) is the load at the ith cycle and PY is the limit load at initial
conditions:
𝑃𝑌 = 𝐵𝑁(𝑊 − 𝑎0)𝜎𝑌 (12)
Here BN is the net thickness (see equation (9)) and σY is the effective yield strength:
𝜎𝑌 =𝜎0+𝜎𝑇𝑆
2 (13)
In the above equations, B is the specimen’s thickness, σ0 is the material’s 0.2% offset yield strength and
σTS is the ultimate tensile strength. The corrected compliance Cc should then be used in equation (7) for
the calculation of the crack growth.
2.2.6 CTOD R-curve calculation
For the construction of the R-curves based on CTOD, the ASTM E1820 standard for SENB [26] can be
used. First off, an estimated initial crack size a0q needs to be calculated. This value is obtained from the
following equation:
𝑎(𝛿) = 𝑎0𝑞 +𝛿
1.4+ 𝐶1𝛿2 + 𝐶2𝛿3 (14)
In this equation, a is the crack size obtained from equation (6), δ is the CTOD value obtained from the
double clip gauge method and C1 and C2 are constants. Equation (14) needs to be fitted to each (ai, δi)
data pair using the method of least squares to determine a0q. The standard states that the difference
between the initial estimated crack size a0q and the optically measured initial crack size a0 (see
20
paragraph 2.4.1) shall not be higher than 0.01W or 0.5mm (the highest value of the two is taken). The
ductile crack extension Δa is taken as the difference between the crack length and the estimated initial
crack depth:
∆𝑎𝑖 = 𝑎𝑖 − 𝑎0𝑞 (15)
The resistance curve is then made from the crack extension and CTOD pairs (Δai, δi). To further improve
the data, a power law function needs to be fitted to the data using the method of least squares. This
power law function is defined by:
𝛿𝑓 = 𝛼𝛿(∆𝑎)𝜂𝛿 (16)
Here δf is the fitted CTOD and αδ and ηδ are fitting parameters. It needs to be noted that the curve fit
should only be applied from a certain onset. The BMT procedure takes this onset at 0.5mm. (Δai, δi)
pairs prior to this chosen onset should not be considered. This is to eliminate the blunting effect from
the fit.
2.3 Digital image correlation
2.3.1 Principle
Another test method that is going to be used is digital image correlation (DIC). Digital image correlation
allows the visualization of the deformations and accompanying strains of the test specimen’s surface by
analyzing consecutive images of the test. A speckle pattern needs to be applied to the specimen’s
surface. This pattern must be isotropic, needs to have a high contrast and must be non-repetitive [37].
As the specimens deforms, so does the applied pattern. Correlation software translates the image into a
2D matrix with a node distribution. For each node, a so called subset is applied. The subset is a square
that determines the field around the node that needs to be analyzed. The first image of the set, which
should represent the undeformed state, is taken as the reference image. For each subsequent image,
the subset is shifted until the pattern in the deformed image corresponds to the pattern of the
reference image as closely as possible. This is done by evaluating the grey-scale intensity as a weighted
average of each subset box [37]. The principle is shown in Figure 26.
Figure 26. Schematic representation of subset shifting to determine surface displacements [37].
21
2.3.2 Procedure
In this thesis, a two-camera system from Limess GmbH is used [38]. By using two cameras, 3D image
correlation is possible. The cameras are placed at approximately 1.5m from the specimen. Between the
two cameras, there is an angle of about 25°. The setup is schematically shown in Figure 27. The speckle
pattern has been applied by using a highly elastic white paint as base layer. On top of this, black speckles
are sprayed on top of the white layer to create the contrast.
Figure 27. Schematic representation of the two-camera setup.
The processing of the digital images is done by correlation software. The used software in this thesis is
the Vic-3D software [39]. The software allows visualizing the displacements and the strains in both 2D
and 3D. An example of such a calculation is given in Figure 28. Here the Von Mises strains were
calculated.
(a) (b)
Figure 28. Picture taken by DIC camera during test of 8U-DIC-TT specimen (a) and the calculated Von Mises strains displayed on the
surface using the correlation software (b).
2.3.3 δ5 measurement
Digital image correlation also allows the measurement different crack opening displacements. In this
thesis, the δ5 CTOD method will be obtained from the DIC measurements. The δ5 is obtained by
measuring the distance between two points, each at 2.5mm from the crack tip.
22
Figure 29. δ5 determination using DIC correlation software.
2.4 Post mortem crack measurement
2.4.1 Ductile crack extension
After each test, a post mortem analysis needs to be conducted on the specimens in order to obtain the
amount of ductile crack extension. Also, equation (11) and (12) require the initial crack depth a0, so this
needs to be obtained from the specimen. The measurement must be performed on the fracture
surfaces. In order to do that, the specimens need to be broken so the fracture surfaces can be accessed.
The breaking of the specimens is done by an internal procedure at Labo Soete.
First off, the specimens need to be heat tinted. This is done by putting the specimens in an oven at
200°C for about three hours. This treatment oxidizes the surface of the material, giving it a distinct color.
This color change is needed to be able to distinguish the difference between the ductile tear from the
test and the fracture surface from breaking the specimen in half. To be able to break the specimen, it
needs to be chilled to low temperatures. This is done by putting the material in liquid air which is at -
195°C. This makes the material brittle, which allows easier breaking. Every specimen yields two fracture
surfaces which are then put under the microscope to obtain detailed images of the fracture surfaces.
The ductile tearing of the fracture surfaces is measured by using the nine points average method as
proposed by the ASTM E1820 standard [26]. Both the initial notch a0 and the final crack length af are
measured nine times at equally spaced intervals. On Figure 30 the vertical lines show how the nine
measurements are distributed across the fracture surface. The initial notch depth and the final crack
extension are then determined by:
𝑎0,9𝑝 =𝑎0,1+𝑎0,9
2+∑ 𝑎0,𝑖
8𝑖=2
8 (17)
𝑎𝑓,9𝑝 =
𝑎𝑓,1+𝑎𝑓,9
2+∑ 𝑎𝑓,𝑖
8𝑖=2
8 (18)
In the equation above, a0,I and af,I are the ith measurement on the fracture surface for the initial crack
size and the final crack size respectively. The ductile crack extension is then determined as the
difference between the initial and final crack size:
∆𝑎9𝑝 = 𝑎𝑓,9𝑝 − 𝑎0,9𝑝 (19)
23
It needs to be noted that the crack extension due to blunting is also present in the ductile crack
extension value.
Figure 30. Microscope image of a fracture surface with nine points average measurement lines.
2.4.2 Blunting
In order to study the blunting, measurements of the blunting are also being made using the
aforementioned nine points average technique. Since blunting is assumed prior to crack initiation, the
blunting region is expected right in the beginning of the ductile tearing, i.e. right after the initial notch
depth. Figure 31 shows a detailed view of the region around the initial notch. Right after the edge of the
notch, plastic deformation can be observed. This plastically deformed region is assumed to be the
blunting region. The size of this region, which is the blunting Δab, is again measured using the nine
points average method:
∆𝑎𝑏,9𝑝 =∆𝑎𝑏,1+∆𝑎𝑏,9
2+∑ ∆𝑎𝑏,𝑖
8𝑖=2
8 (20)
Here Δab,I is the ith measurement taken from the top of the blunting region to the bottom of this region.
Figure 31. Detailed view of the blunting region on the fracture surface.
24
3 Test specimens
3.1 Pipe forming process In this thesis, the material of four different pipes is being analyzed. The pipes differ by either the way
they were constructed or by their strength level. There are two forming processes for the used pipeline
material: spiral welded pipes and UOE formed pipes. For each formed pipe, two pipes with different
strength levels will be studied giving a total of four different pipes.
3.1.1 Spiral welded pipes
Spiral welded pipes start from a flat sheet of metal. The sheet is bent at a certain angle so that a
spiraling tube is formed. The abutting edges of the spiraled metal sheet are continuously welded in line.
See Figure 32 for a schematic representation of the process. Spiral welding of pipes has the great
advantage that different pipe diameters can be formed from a single sheet. The diameter of the pipe is
determined by the approach angle of the sheet. The smaller the inlet angle, the larger the diameter of
the pipe will be for a given plate width. The range of diameter for spiral welded pipes usually lies
between 500 and 2500mm. The wall thickness is usually maximum 20mm, although higher values are
possible. The length of the pipe segments are not limited by the process, since the pipe can be cut at any
desired length [40].
There are two main categories for spiral pipe forming. The first and most conventional method consists
of a single process where the pipe welding is integrated with the pipe forming. This welding is done by
submerged arc welding (SAW) and is performed at the outside as well as the inside of the pipe. The
forming process is substantially faster than the SAW welding step. This has given rise to the
development of a second category of spiral pipe forming where forming and welding steps are
performed separately. The pipe is first formed and connected using fast (temporary) tack welds, which is
done by metal active gas (MAG) welding. The fast forming stage is then able to supply multiple (slower)
SAW welding stands, which increases the production speed [40].
Figure 32. Schematic representation of the spiral forming process [41].
3.1.2 UOE formed pipes
Most large diameter onshore pipes are made using the UOE process. This process starts from a flat plate
of metal, but instead of a spiral weld, a longitudinal weld is used. The name UOE originates from the
cold forming process steps that are involved in the forming process. An edge crimped plate is first bent
into a U-shape. In the following step the U-shaped plate is formed into an O-shape. The closed pipe is
25
then first tack welded using MAG welding and later fully welded using SAW welding. To achieve the
pipe’s final dimensions, the pipe is mechanically expanded (the E-stage of the process). This is achieved
by mechanical or hydraulic expanders. The plastic expansion of the pipe also ensures a better
roundness, because the welded O-shape has some degree of ovality. The process is schematically shown
in Figure 33. Diameters for these pipes usually range between 400 and 1620mm. Wall thicknesses of 6
to 40mm can be obtained. The pipe segments are made to a length of up to 18m [40].
Figure 33. Schematic representation of the UOE forming process [42].
3.2 Steel grades The strength level for pipe lines are often defined by the API-5L (2007) standard [43]. The standard is
made by the American Petroleum Institute (API) and is applied worldwide. The standard contains
general information that specifies attributes like dimensions, mechanical and chemical properties and
manufacturing techniques. Pipes manufactured according to this standard are named based on their
specified minimum yield strength (SMYS). The steel grade is named by the letter X followed by a number
which specifies the material’s minimum yield strength in kilopounds per square inch (ksi). For example,
an API-5L X80 steel pipe has a SMYS of 80ksi, which is approximately 555MPa. For simplification,
materials based on this standard will be abbreviated henceforth, e.g. ‘API-5L X80’ will be referred to as
just ‘X80’.
In this thesis, two steel grades are being analyzed. These grades are X70 and X80. These steels have,
according to the API-5L standard, a minimum yield strength of 485MPa and 555Mpa respectively.
3.3 SENT specimens
3.3.1 Full pipe
As was mentioned earlier, four different pipe lines will be studied. Two pipes were made by the spiral
weld process and the other two by the UOE process. For each forming process, two different steel
grades are taken (X70 and X80). The tests will be performed on base material, so welds are not taken
into account. General information about the full pipes is listed in Table 3.
Table 3. General properties of the different pipes.
API 5L Steels Pipe Dimensions Manufacturing Features
Grades Name O.D. ["]
O.D. [mm] W.T. [mm] Year of
production Process
X70 7S 48 1219.2 13.7 1993 Spiral
7U 48 1219.2 16.9 2001 UOE
X80 8U 44 1117.6 16.2 2001 UOE
8S 48 1219.2 23.7 2013 Spiral
26
As can be seen from the table above, the pipes have been manufactured over a different timespan. The
first spiral pipe can be labeled as an older pipe. The two UOE pipes are considered as contemporary
pipelines and the X80 spiral pipe is a modern one. The pipes mostly have similar outer diameters. The
wall thicknesses on the other hand are different for each pipe.
3.3.2 Notch locations
The specimens are obtained from a single slab which has been cut out of every pipe. The slab has been
machined to remove rust and to make sure a rectangular plate is obtained by removing the curvature of
the pipe. The specimens are then cut out of each plate. For the spiral pipes, the specimens were cut
longitudinal to the pipe axis. The UOE specimens were taken longitudinal to the rolling direction of the
plate from which the pipe was made. This way, the UOE specimens will have grains that are in line with
the specimen since it follows the rolling direction of the plate. The grains of the spiral pipes will be at an
angle with respect to the specimen’s longitudinal direction.
In order to analyze the effect of a material’s heterogeneity on the tearing resistance, two different notch
locations will be used. The first location is taken at the inner diameter (ID) of the pipe. The second notch
is cut in the through thickness (TT) direction of the pipe. The notch locations are schematically shown in
Figure 34.
(a) (b)
Figure 34. Schematic representation of the inner diameter (ID) notch (a) and the through thickness (TT) notch within a pipe (b).
The biggest influence on the difference in material properties is expected from the rolling of the plates.
In order to get representative results, three specimens for each notch location will be tested. This is also
specified in the ASTM E1820 standard [26]. This gives a total of twenty-four specimens that will be
analyzed using the unloading compliance method. All these specimens will also have side grooves as was
discussed in section 1.3.4.
For each material, one extra specimen will be tested for each notch location using the DIC method. This
gives eight extra test specimens. These specimens will not have side grooves. This is chosen to have a
smooth, uninterrupted surface around the notch. This is not detrimental for the measurements since
the interest of the DIC test is not on the crack growth, but rather on the strains on the surface.
3.3.3 Unloading compliance specimens
For each steel, tensile tests were performed on round bar specimens. These tests were performed
within the framework of an ongoing PhD. The most important results are listed in Table 4. From this it
can clearly be noted that steel 7U is, stress wise, the weakest material. It does have, on the other hand,
the highest strain hardening exponent n, which means it has better formability. When we look at the
27
yield-to-tensile ratio (Y/T), the two UOE pipes seem to have a lower ratio (around 0.8) than the two
spiral pipes (around 0.9). For strain based design, a lower yield-to-tensile ratio is preferred. The uniform
elongation em, which is the amount of strain a material has before necking initiates, is the highest for the
X70 UOE pipe. The other pipes have comparable values except for the 8S pipe, which has a considerable
lower value for the uniform elongation.
Table 4. Tensile properties for each steel.
Material σ0
[Mpa] σTS
[Mpa] E
[Mpa] n [-]
Y/T [-]
em [%]
7S 553 625 190928 0.108 0.88 9.3
7U 477 616 235291 0.120 0.77 10.2
8U 557 695 236319 0.095 0.80 8.8
8S 597 654 208276 0.068 0.91 5.8
All specimens have dimension in accordance with the proposed dimensions from section 1.3.4. The most
important dimension for steel 7S, 7U, 8U and 8S are in given in Table 5, Table 6, Table 7 and Table 8
respectively.
Table 5. Dimensions for steel 7S.
Notch orientation
Name a0
[mm] W
[mm] B
[mm] H
[mm] a0/W
[-]
Used symbol
TT
7S-TT1 3.75 12.45 12.47 125 0.3
7S-TT2 3.75 12.45 12.46 125 0.3
7S-TT3 3.75 12.45 12.47 125 0.3
ID
7S-ID1 3.75 12.47 12.44 125 0.3
7S-ID2 3.75 12.48 12.47 125 0.3
7S-ID3 3.75 12.49 12.46 125 0.3
Table 6. Dimensions for steel 7U.
Notch orientation
Name a0
[mm] W
[mm] B
[mm] H
[mm] a0/W
[-]
Used symbol
TT
7U-TT1 4.50 14.99 15.00 140 0.3
7U-TT2 4.50 15.00 15.00 140 0.3
7U-TT3 4.50 15.00 14.99 140 0.3
ID
7U-ID1 4.50 14.98 15.00 142 0.3
7U-ID2 4.50 14.94 14.98 140 0.3
7U-ID3 4.50 14.97 14.99 140 0.3
28
Table 7. Dimensions for steel 8U.
Notch orientation
Name a0
[mm] W
[mm] B
[mm] H
[mm] a0/W
[-]
Used symbol
TT
8U-TT1 4.50 14.97 14.96 150 0.3
8U-TT2 4.50 14.95 14.97 150 0.3
8U-TT3 4.50 14.96 14.96 150 0.3
ID
8U-ID1 4.50 14.96 14.95 150 0.3
8U-ID2 4.50 14.94 14.96 149 0.3
8U-ID3 4.50 14.97 14.95 150 0.3
Table 8. Dimensions for steel 8S.
Notch orientation
Name a0
[mm] W
[mm] B
[mm] H
[mm] a0/W
[-]
Used symbol
TT
8S-TT1 6.60 22.00 22.05 220 0.3
8S-TT2 6.60 22.00 21.97 220 0.3
8S-TT3 6.60 21.98 21.98 220 0.3
ID
8S-ID1 6.60 21.97 22.00 220 0.3
8S-ID2 6.60 22.00 22.00 220 0.3
8S-ID3 6.60 21.98 22.00 220 0.3
Certain symbols have been assigned to each configuration. This is done to avoid overloading the graphs
with information. In the following of this thesis, each symbol will correspond to the indicated
configuration.
3.3.4 DIC specimens
The DIC specimens have the same tensile properties as mentioned for the specimens above (see Table
4). As was mentioned earlier, the DIC specimens do not have side grooves. The dimensions of the
specimens are given in Table 9.
Table 9. Dimensions for the DIC specimens.
Steel Notch
orientation Name a0
[mm] W
[mm] B
[mm] H
[mm] a0/W
[-]
7S TT 7S-DIC-TT 3.75 12.46 12.48 125 0.3
ID 7S-DIC-ID 3.75 12.47 12.46 125 0.3
7U TT 7U-DIC-TT 4.50 14.97 14.99 150 0.3
ID 7U-DIC-ID 4.50 14.96 15.00 150 0.3
8U TT 8U-DIC-TT 4.50 14.96 14.96 150 0.3
ID 8U-DIC-ID 4.50 14.95 14.95 150 0.3
8S TT 8S-DIC-TT 6.60 22.00 22.00 220 0.3
ID 8S-DIC-ID 6.60 22.00 22.00 220 0.3
29
3.4 Test rigs The specimens are tested on tensile test rigs. Because the specimens from different pipes have different
dimensions, it was necessary to use two different test rigs. The smaller 7S, 7U and 8U specimens were
tested on a 150kN test rig, because the maximum forces needed for these specimens did not exceed
150kN. The maximum forces needed for the testing of the 8S specimens on the other hand did exceed
that number, so this test rig was not adequate. Therefore a larger tensile test rig with a maximum force
of 1000kN was used for these specimens.
The specimens used for the 150kN test rig were mounted by using mechanical clamps. The clamps were
fastened with a bolt using a large wrench. The specimens for the 1000kN rig were clamped using
hydraulic clamps. Both clamping mechanisms are shown in Figure 35.
(a) (b)
Figure 35. Specimen fixed by mechanical clamps in the 150kN test rig (a) and specimen clamped by hydraulic clamps in the 1000kN
test rig (b).
4 Evaluation of the test results
4.1 Compliance curves As was discussed earlier, an increasing relationship between the compliance C and the CMOD is
expected. Figure 36 shows the compliance curves for one TT specimen from each steel. The curves seem
to comply with the expected results. All curves show an increase in compliance for an increase in CMOD.
Also, there is no initial drop noticeable due to rotation, but rather an initial linear increase. This is in
accordance with what Verstraete et al. [11] reported. It was stated that the rotation effect mostly
influences deeply notched specimens (a0/W=0.5). Shallow notched specimens were not really affected
by the rotation effect and rather exhibit an initial linear increase. Since the specimens tested here have
a more shallow notch (a0/W=0.3), the latter is expected. This confirmed by the results shown in Figure
36, where each specimen clearly exhibits an initial linear increase. As was also mentioned before, the
moment where the compliance curve deviates from the linear line indicates the initiation of the crack.
30
The compliances shown here are the ones obtained from each unloading cycle. It is these compliances
that are used in equation (7) to estimate the crack growth. Although it needs to be noted, that these are
not directly used, since the values are influence by rotation and necking. Therefore these values need
first to be corrected using the correction factor stated in equations (10) and (11).
(a) (b)
(c) (d)
Figure 36. Compliance vs CMOD and initial linear increase for the 7S-TT1 specimen (a), the 7U-TT1 specimen (b), the 8U-TT1
specimen (c) and the 8S-TT1 specimen (d).
A thing that was noticeable about the results of the compliance curves is that the 8S steel shows a lot of
scatter. There seems to be a quite a difference between the three same configuration specimens (both
TT and ID) as well as variance in the individual curves. To illustrate this difference, the compliance
results are shown in Figure 37 next to the results for the 7S steel, where more consistent data can be
seen.
4.2 Initial crack estimation In order to validate the obtained results, a comparison is made between the estimated initial crack size
a0q and the measured initial crack size a0,9p (from the facture surfaces). The way these parameters were
obtained was discussed in paragraph 2.2 and 2.4. It needs to be noted that estimated value of the initial
crack size, a0q, is seen as the size of the crack just before crack initiation, so the distance covered by
blunting is included in this estimation. Therefore, in order to make a meaningful comparison, the
31
measured blunting Δb9p needs to be added to the measured value of the initial crack size a0,9p. This will
be defined as the total measured initial crack size a0,m.
𝑎0,𝑚 = 𝑎0,9𝑝 + ∆𝑏9𝑝
↔ 𝑎0𝑞 (21)
(a) (b)
Figure 37. Difference in scatter between the 7S TT specimens (a) and the 8S TT specimens (b).
According to the ASTM E1820 standard [26], this difference may not be higher than 0.5mm. Figure 38
shows the result for all specimens. It needs to be noted that for each notch configuration, the average
has been taken of the three corresponding specimens. A first thing that can be noted is that the data
points for TT and ID configuration lie close to each other for all steels except the 8S specimens. This
difference is caused by high scatter in the compliance measurements for each specimen of the 8S steel,
as was showed in Figure 37. Since the compliances are used to estimate the crack sizes, difference in this
data will give different crack size estimations.
Secondly, two materials fall outside of the 0.5mm limit and this for both the notch configurations.
Furthermore, all crack sizes are over estimated. This over estimation is also shown in Table 10. The table
shows that the 8S exceeds this limit by twice the allowed value. This result raises the question to the
validity of the data for the 8S specimens.
Figure 38. Comparison between estimated and measured initial crack size.
32
Table 10. Estimated vs. measured initial crack size for each configuration.
Material a0,m a0q a0q-a0,m
7S-TT 4.00 4.32 0.32
7S-ID 4.08 4.35 0.27
7U-TT 4.82 5.62 0.80
7U-ID 4.79 5.45 0.65
8U-TT 4.77 5.22 0.45
8U-ID 4.78 5.09 0.30
8S-TT 7.43 8.65 1.22
8S-ID 6.98 7.88 0.90
4.3 Total crack propagation The unloading compliance allows the calculation of the final crack propagation by subtracting the
estimated initial crack size a0q from the estimated crack size a (obtained from the compliances). In order
to evaluate the accuracy of the unloading compliance method, these values can then be compared with
the measured values of the crack propagation (obtained from the nine points average method). This
comparison is displayed in Figure 39. To evaluate this accuracy, a 95% confidence interval is displayed
around the data. This is obtained by taking an interval of ±1.96STD around the data, where STD is the
standard deviation, assuming a normal error distribution.
In contrast to the results of the initial crack size obtained in the previous paragraph, it can be noticed
from Figure 39 that the results are not all over-estimated. The results are both above and under the
identity line. The reason the total crack propagation is not over-estimated like the initial crack size a0q, is
that the crack length a calculated from the compliances is also over-estimated. This way, the subtraction
of the two eliminates this extra crack length, giving more comparable crack propagation results.
Furthermore, the results seem to all fit within the 95% confidence interval (except for S8-ID), which
shows that the crack growth estimation seems pretty accurate. It seems that for these specimens that
the unloading compliance method is better at estimating the total crack propagation than estimating
the initial (and final) crack lengths. Since resistance curves are constructed by comparing the CTOD with
the crack propagation, a correct crack propagation estimation is the most important.
Another thing that can be noticed is that the data points for two notch configuration of the same steel
are more apart from each other. This indicates that the notch location influences the total crack
propagation. For all materials except the 7S the ID specimens yield a longer total crack length.
33
Figure 39. Comparison between estimated total crack propagation obtained by the unloading compliance method and the total
crack propagation obtained by measurements on the crack surfaces.
4.4 Blunting As was mentioned earlier, the blunting was measured using the nine points average method on the
fracture surfaces. This was needed in order to compare the measured initial crack length with the
estimated initial crack length. The unloading compliance method estimates the initial crack length, but it
includes the blunting. It does not offer a way to obtain the blunting from this value. There is however
another way to estimate the blunting from the measurements. The blunting is based on the evaluation
of the CTOD before crack initiation [44].
∆𝑎𝑏 =𝐶𝑇𝑂𝐷
2 𝐶𝑇𝑂𝐷 ≤
𝐶𝑇𝑂𝐷𝑖𝑛𝑖
2 (22)
∆𝑎𝑏 =𝐶𝑇𝑂𝐷𝑖𝑛𝑖
2 𝐶𝑇𝑂𝐷 ≥
𝐶𝑇𝑂𝐷𝑖𝑛𝑖
2 (23)
CTODini is the CTOD value when crack initiates. The moment of initiation was described in section 2.2.5.
Since all specimens have shallow notches, the second method is used to determine the moment of crack
initiation. To illustrate this, an example of how to acquire the CTODini is given in Figure 40. The C vs.
CMOD curve is used to determine the unloading cycle at which the curve deviates from the linear trend.
Then the CTOD value corresponding to this cycle is taken as CTODini.
34
(a) (b)
Figure 40. Determination of crack initiation (a) and determination of CTODini (b).
The comparison between the measured blunting distances and the estimated distances is given in Figure
41. The figure shows that the formula based on CTOD at initiation estimates the blunting quite well.
Most values are within the 95% confidence interval, which is quite narrow (0.10mm). It needs to be
noted that the determination of crack initiation requires some exercise as it is not always as easily seen.
The measurement of the blunting on the crack surface is also not so straight forward since the blunting
is not always uniformly distributed across the crack surface.
Figure 41. Comparison between blunting measured by the nine points average method on the fracture surfaces and the estimated
blunting based on CTOD at crack initiation.
35
5 Evaluation of the resistance curves
5.1 Accuracy of the resistance curves In the first chapter it was discussed that the tearing resistance is an essential parameter within strain
based design. Resistance curves can be used to objectively quantify this parameter. The resistance
curves were determined by the compliance curves obtained in the previous chapter. The R-curves are
built up from the relationship between the crack tip opening and the crack propagation. The reliability
of these curves partially depends on the correct estimation of this crack growth. Although the results
showed an over-estimation of the actual crack size, the total distance of the propagation was estimated
correctly.
Scatter in the compliance curves will automatically be translated in scatter in the crack growth
estimation, since the two are related via mathematical equations. In the previous chapter, it was
mentioned that the 8S steel showed a lot of scatter. The results for the resistance curves were therefore
also incoherent. This can be clearly seen in Figure 42 (g) and (h).
(a) (b)
(c) (d)
36
(e) (f)
(g) (h)
Figure 42. Scatter of calculated R-curves for specimens 7S-TT (a), 7S-ID (b), 7U-TT (c), 7U-ID (d), 8U-TT (e), 8U-ID (f), 8S-TT (g) and
8S-ID (h).
Figure 42 shows the fitted resistance curve (solid line) for each notch configuration. This fitted line was
determined according to equation (16). Together with this curve fit, all data points obtained from all
three specimens of each configuration are displayed. A 95% confidence interval is also displayed on each
graph and serves as a scatter band. It needs to be noted that the curve fit is calculated from crack
initiation onwards, so the initial data points do not contribute to the curve fit and the accompanying
confidence interval. The scatter bands have a constant width of 2δs and are determined by:
𝐶𝑇𝑂𝐷 = 𝛼𝛿∆𝑎𝜂𝛿 ± 𝛿𝑠 (24)
Figure 42 (a) and (b) shows the data points for the R-curves of the X70 spiral pipe for the TT and ID
configuration respectively. For both configurations, the data points all seem to be in the vicinity of the
curve fit. The 95% interval is also quite narrow, showing that the data is quite consistent. The TT
configuration has an accuracy of ±0.12mm (6% of max CTOD value) and the ID configuration has an
accuracy of ±0.13mm (6%).
37
Figure 42 (c) shows the data for the TT configuration of the X70 UOE pipe. It needs to be noted that this
curve is based on the data of only two specimens since one specimen had a faulty test with unusable
data. Both data sets have a similar evolution, but differ in absolute values. Still the accuracy is around
±0.19mm (6%), which is similar to the results for the 7S steel. Figure 42 (d) displays the ID configuration
for the 7U material. The data points for this configuration seem to be more dispersed, which also gives a
larger scatter band. The accuracy here is ±0.27mm (10%), which is almost double.
The R-curves for the TT and ID configurations for the X80 UOE pipe are shown on Figure 42 (e) and (f)
respectively. For this material all data points seem to be close to the curve fit (both for TT and ID
configuration). The accuracy is ±0.16mm (5%) and ±0.14mm (5%) for TT and ID respectively.
Finally, Figure 42 (g) and (h) show the result for the X80 spiral pipe. As was mentioned earlier, it shows a
lot of scatter. For both configurations the data points are widely dispersed without really showing a
common trend. The size of the scatter band here is ±0.50mm (12%) and 0.67mm (16%), which is quite
large. Because of the large scatter and inconsistency of the data, the R-curves are deemed as non-
representative and will not be further investigated. Somehow the material does not seem adequate to
be tested by the unloading compliance method. This will be discussed later on.
Disregarding the results for the S8 specimens, all materials seem to have similar relative scatter band
size (except for 7U-ID), so the results appear to have the same accuracy.
A final thing that can be noted from the scatter plots is that all specimens show a high scatter in the
beginning of the curve. It is assumed that this variance in data is caused by the blunting of the crack tip.
This is also the reason why this phase is ignored in the calculation of the R-curves.
5.2 R-curve comparison
5.2.1 Comparison per material
Figure 43 shows the R-curves for the TT and ID configurations per material. A first thing that can be
noticed is that for both the UOE materials, the TT configuration has the best tearing resistance. For the
X70 spiral pipe however, the opposite is true.
The curves displayed here are fitted until the final crack extension of the tests. The values of these crack
extensions are displayed in Table 11. For the X70 steels, the crack extension between the two notch
configurations does not differ by much (±8%). For the X80 pipe however, the ID notch had a 15% longer
tear. Together with the fact that the TT configuration for this pipe is considerably better, it seems that a
notch at the inner diameter of the pipe is much less favorable.
Since the specimens do not all have the same dimensions, absolute values cannot really be compared.
Therefore Table 11 shows the amount of ductile tearing relative to the specimen width. It shows that
the tearing is around 14-16% for all specimens except the 8U-ID specimen which is at 18%. So except for
the last configuration, the relative amount of tearing seems to be similar for all three materials.
38
Table 11. Total crack extension for each notch configuration.
Steel Notch
orientation
Total crack extension (Δa)
[mm]
Width of specimen (W)
[mm]
Δa/W [%]
7S TT 1.88 12.50 15.04
ID 1.73 12.50 13.84
7U TT 2.20 15.00 14.67
ID 2.39 15.00 15.93
8U TT 2.28 15.00 15.20
ID 2.69 15.00 17.93
The difference between the R-curve for TT and ID configuration is not as outspoken for the X70
materials as it is for the X80 pipe. The difference around final crack extension is about 10% for the 7S
material and around 11% for the 7U material. The difference for the 8U however is about 20%, which is
twice as much.
(a)
39
(b)
(c)
Figure 43. R-curve comparison between TT and ID notch location for material 7S (a), 7U (b) and 8U (c).
5.2.2 Comparison per notch configuration
In this section, the R-curves from the different materials are compared to each other. This is done for
both the notch configurations (see Figure 44). The curves are plotted for a ductile crack extension of
3mm, which is further than the actual crack extension. This allows a better comparison between the
shapes of the curves, since the specimens have different geometries and therefore different absolute
crack extensions.
Tearing resistance wise, it is clearly seen that the X70 UOE pipe offers the best results. This is true for
both the TT and the ID notch configurations. From comparisons between configurations made in the
40
previous paragraph it was seen that the UOE pipes exhibit better tearing resistance for a notch in the TT
sense. Looking at the comparison of the TT configuration (Figure 44 (a)) it is noticeable that both the
UOE pipes score considerably better than the spiral pipe.
For the ID configurations (Figure 44 (b)) the 7S and the 8U material seem to have comparable results
(although the latter is a bit better). The 7U material is on average quite better. However the accuracy
seems to be the worst of all results.
(a)
(b)
Figure 44. R-curve comparison between materials for the same notch configuration.
41
5.3 Influences on the R-curves
5.3.1 Tensile properties
In the previous sections it was seen that there is quite some difference between the R-curves. A first
way to discuss the difference between the R-curves of the different materials is by looking at the tensile
data. From the previous curves not only a difference between different API-5L grade steels could be
seen, but also between same grade steels.
Figure 45 shows the engineering stress-strain curves for all four materials. The accompanying values
obtained from these curves were already given in Table 4 (paragraph 3.3.3). The two X70 steels have a
somewhat comparable stress-strain curve, except that the spiral pipe has a so called Lüders plateau. The
spiral pipe also has a slightly higher strength than the UOE pipe (14% higher σ0 and 2% higher σTS). The
curve difference between the two X80 steels is much more significant. Although the X80 spiral pipe has
a higher yield strength (7% higher), the ultimate tensile strength is higher for the UOE pipe (6% higher).
Compared to the X70 steels, the X80 UOE has significantly higher strength values. Still, previous results
have shown that the 7U material has the best tearing resistance. This shows the importance of the
strain based design. In the classic stress based design method, material with higher strength properties
are seen as the better materials. Within the stress based design the tearing resistance is a keyword and
from the results it seems that the material that appears as the weakest on paper yields the best results
in terms of tearing resistance. Therefore the focus cannot be solely on the strength of the material.
Figure 45. Stress-strain curves for all testing materials.
Two tensile parameters that need to be taken into account in strain based design are the yield-to-tensile
ratio (Y/T) and the uniform elongation (em). These two parameters are somewhat related in that an
increasing Y/T ratio will give a decrease in the uniform elongation and vice versa [45]. Materials with a
high yield-to-tensile ratio have less reserve to withstand loads above yield strength and are therefore
prone to localized deformations. Therefore a lower Y/T ratio is preferable for strain based design. Also,
0 5 10 15 20 25
0
200
400
600
800
Engi
nee
rin
g st
ress
[M
Pa]
Engineering strain [%]
0 5 10 15 20 25
0
200
400
600
800
0 5 10 15 20 25
0
200
400
600
800
7S
7U
8U
8S
0 5 10 15 20 25
0
200
400
600
800
42
materials with a high Y/T ratio will reach their maximum strength with less deformation, which will
cause the material to fail quicker under large deformations. The uniform elongation is important
because it determines how much the material can deform before it starts to neck. After necking, the
material strength drops relatively due to a high reduction in the cross section of the material.
The 7U material has the lowest Y/T ratio (0.77) and has the highest uniform elongation (10.2%). This
material scores best on these two parameters and has the best tearing resistance results, indicating that
these parameters indeed have a great influence on the tearing resistance. When comparing the 7S and
the 8U material, the 7S has a higher uniform elongation (9.3% against 8.8%), but also a higher yield-to-
tensile ratio (0.88 against 0.80). For the ID notch, both materials have comparable tearing resistance
results, but for the TT notch the 8U material performed much better. Based on this data, it’s hard to say
which parameter most influences the performance.
As was mentioned earlier, the two X70 steels have a somewhat comparable stress-strain curve. The
spiral pipe has a higher value for the Y/T ratio and this may be the reason why the UOE pipe scores
better in terms of tearing resistance. However, there is a second important difference between the two
materials. The UOE pipe has what is called continuous yielding behavior. This type of yielding is
characterized by a smooth course from the linear elastic region to the yielding region. The spiral pipe on
the other hand has a discontinuous yielding behavior. Discontinuous yielding shows a sharp yield point
followed by a Lüders plateau. The difference between the two is shown in Figure 46. The presence of
the Lüders plateau has a negative influence on the material’s buckling resistance and therefore lowers
the bending strain capacity of the material [46]. This discontinuous yielding of the spiral pipe probably
causes the lower tearing resistance of the material. Materials with discontinuous yielding are therefore
best avoided in strain based design applications. The X70 spiral pipe is the oldest pipe of the four
(production year 1993). The other three newer pipes do not show any discontinuous yielding, which
indicates that manufacturers indeed avoid this phenomenon in their materials.
The X80 spiral pipe shows the worst results in terms of yield-to-tensile ratio and uniform elongation. The
Y/T value is quite high (0.91) and the uniform elongation is considerably lower (5.8%). The shape of the
stress-strain curve also shows that the material strength quickly drops for increasing strains. The high
Y/T value causes localized deformations when the material exceeds the yield strength. This may be the
reason why the SENT tests of this material gave such largely scattered results and therefore unusable
tearing resistance curves. Since there will be much local deformation around the crack tip, the CMOD
and CTOD values will not be measured properly. Therefore the compliance will be calculated incorrectly,
rendering the unloading compliance method useless. This material would be good from a stress based
angle. It has a high ultimate tensile strength and because of its high Y/T ratio, the design stress can be
taken quite high even with the necessary safety margin. From a strain based point of view, this material
is not really suited.
43
Figure 46. Continuous and discontinuous yielding behavior [46].
5.3.2 Hardness properties
The R-curves for each material showed differences between the curves obtained from a TT notch and an
ID notch even though the material was identical. This proves that there is a certain level of
heterogeneity in the material properties. These heterogeneities can be caused by the material
composition, the production method, the applied treatments, the forming processes … A way to map
this heterogeneity is by performing a hardness test on the material.
In order to obtain this map, Vickers indentations with a load of 0.5 kg (HV0.5) are performed across the
thickness of each material. The results of these hardness measurements are displayed in Figure 47.
In the figure, ID indicates the inner diameter, OD the outer diameter and CL the centerline of the
specimen. A first thing that can be noted from the hardness maps is that all materials except the 7U
have a clear softer region at the center of the material. Especially the X80 materials exhibit this
phenomenon. Table 12 shows the average hardness values per region. How these regions are defined is
shown in Figure 48. The difference between the center region and the outer regions is about 3%, 1%, 4%
and 5% for the 7S, 7U, 8U and 8S specimens respectively.
Table 12. Average Vickers hardness values per region.
Material ID
[kgf/mm²] CL
[kgf/mm²] OD
[kgf/mm²] Global
[kgf/mm²]
7S 252 248 256 252
7U 174 175 179 176
8U 261 254 267 261
8S 220 213 227 220
(a) (b)
44
(c) (d)
Figure 47. Hardness map for material 7S (a), 7U (b), 8U (c) and 8S (d).
Figure 48. Schematic overview of the hardness regions and notch and crack extension distributions of the hardness regions for the
7S specimen.
Figure 48 also schematically shows how the notch is located on the specimen and how the crack extends
through the hardness region for the 7S material (the locations are analogous for the other materials).
For a TT notch the crack front lies across the three regions and therefore sees different hardness levels
along the front. As the crack propagates, a fixed point on the crack front will not see much difference in
hardness. However, since the hardness along the length of the crack front does change, a difference in
crack propagation is to be expected along this line. This was indeed observed when examining the crack
surfaces (see Figure 49), except for the 8S material. The surfaces show a curved crack front indicating
non-uniform crack propagation.
45
(a) (b)
(c) (d)
Figure 49. Crack front shape for TT notched specimens for material 7S (a), 7U (b), 8U (c) and 8S (d).
For an ID notch, the opposite is true. Along the crack front, the hardness is practically the same, but as
the crack propagates, the crack front will traverse into different hardness regions. From Figure 48 it can
be seen that the notch goes until the border of the softer center region. The crack thus propagates into
the softer region. Since the hardness is uniform along the crack front, a more uniform crack propagation
is expected. This was again observed on the fracture surfaces which showed straight final crack fronts
for the majority of the ID specimens.
For both the X70 materials, the difference between the R-curves for the TT and ID configurations was
rather small. For the 7U material this can be attributed to the fact that the hardness is more or less
uniform. For the 7S material, the softer region in the middle is quite small. As can be seen from Figure
48, the crack barely reaches the soft region. For the 8U specimen on the other hand, a much larger
difference between the ID and TT configuration was noticed. This material has a much larger soft central
zone and it’s easy to see that the ID crack propagates into this region. This was also noted when the
crack extension was examined. While the TT specimens showed a steady crack growth of 0.15-0.20mm
per unloading cycle, the ID specimens initially exhibit the same amount of crack extension, but this
increases to 0.30-0.40mm for later unloading cycles. This clearly shows that this softer region is
detrimental for the tearing resistance. This is however in contradiction with what Verstraete et al. [11]
reported. Here it was found that softer regions within a weld gave better tearing resistance. This shows
46
that the hardness itself does not really determine the tearing resistance, but rather the material
composition.
5.3.3 Metallurgical properties
In collaboration with an ongoing PhD, which mainly focuses on microstructural properties, metallurgical
properties have been analyzed. This thesis will not go deeply into this topic, since the main focus is on
the mechanical side of the materials. However, a very short discussion will be put here.
The four steels can be separated into two groups. The first group consists of steels having an old or
contemporary microstructural design. This group includes the 7S, 7U and the 8U materials. The second
group, encasing the 8S material, has a modern microstructural design. The first group mainly consists of
heavily banded and oriented ferrite-pearlite structures. The structure of the 8S material consists of a
non-traditional bainitic structure.
When comparing the 7U and the 8S (which are the best and the worst strain based material
respectively) the microstructure explains why these material behave differently. Figure 50 shows
microscopic images of both the materials. The X80 pipe has a structure consisting of different phases
(mainly carbon ferrite, acicular ferrite and martensite). A strong characteristic of this material is that it
has a high temperability. This means it can withstand high temperatures without undergoing phase
changes or creating secondary phases. The material has high toughness, but when the material starts
deforming, the whole matrix contracts and there is practically no more resistance to further
deformation (which proofs the high Y/T value and the low em). The 7U material on the other hand has
regions of pearlite (see the darker regions in Figure 50 (a)). These regions act as hard components which
impede further deformation of the matrix when deformation starts, giving it a better yield-to-tensile
ratio.
(a) (b)
Figure 50. Microscopic image obtained by optical microscopy (OM) for the 7U material (a) and the 8S material (b).
The 7S and 8U have slightly different microstructures due to heat treatments and forming processes,
but the general functioning of the matrix is similar to the one of the 7U material. This shows that
microstructural analysis of the material is important within strain based design. Also, the newer
materials which use the latest technologies to obtain high quality steels are not necessarily the better
choices within strain based design.
47
6 DIC results
6.1 CTOD measurement comparison As was mentioned earlier, DIC techniques can be used to measure the δ5 definition of the CTOD. By
placing two points at 2.5mm distance from the crack tip, the software is able to track the displacements
of these points and thus calculate δ5. Figure 51 (a) shows how δ5 was defined within the program.
(a) (b)
Figure 51. Definition of the δ5 measurement using DIC software (a) and results for obtained values of δ5 and δ90 per unloading cycle for specimen 7S-DIC-ID (b).
Figure 51 (b) shows the results from the obtained δ5 and δ90 per unloading cycle for the ID specimen for
the X70 spiral pipe. Both measurements seem to coincide initially, but then the 90° method yields higher
CTOD measurements. This observation was also seen for all the other specimens, where the δ90 always
gives higher values than the δ5.
Figure 52 shows the comparison between the two methods for the CTOD obtained from the last cycle of
each specimen. As can be seen, every final CTOD value is higher for the 90° intercept method. This is
explained by the fact that the δ5 method measures the crack tip opening at the height of the crack tip.
The 90° method on the other hand is measured slightly above the crack tip. When the crack opens, the
crack fronts will rotate a bit and thus cause the distances between the flanks to increase when you go
away from the crack tip. This way the 90° method will be a bit higher than the δ5 method [11]. Apart
from this slight difference for higher CTOD values, the δ5 measurements show that the CTOD values
obtained by the double clip gauge method are fairly accurate.
48
Figure 52. Comparison between CTOD values obtained by δ5 and δ90 for the final cycle of each specimen.
6.2 Strain field analysis
6.2.1 Strain fields at the blunting phase
The DIC software is not only used to measure the displacement between two points, but it is also used
to determine the displacements across the whole surface. From this the strains present on the surface
can be calculated and visualized.
In the previous chapters it was discussed that a blunting phase was present prior to crack initiation. How
this phase influences the tearing resistance of the material is not really known. In an attempt to study
the correlation between the blunting and the strain fields of the materials, different moments during
the blunting phase are compared. In the following figures a few evenly spread moments during the
assumed blunting stage are displayed. The TT and ID configurations are paired together in order to make
a better comparison. Two other points are indicated in the figures, namely the limit load Py, which
serves as the yield load for the notched ligament and the maximum load Pm, which is the maximum
occurring load during the test.
Previous works noted that the strain fields spread at 45° angles from the crack tip if the material is
uniform (no welds, mismatches…) [11, 12, 27]. Since only base material is considered in this thesis, two
uniform strain fields are expected. This is indeed observed for all specimens. Figure 53 displays the
aforementioned moments for the X70 spiral pipe.
Comparing the two notch orientations for the 7S material, practically no difference is noticeable. The
strain fields have the same shape, size and the scale is also the same. Table 13 gives the values for the
measured blunting obtained from the nine points average method discussed in paragraph 2.4.2. The
table shows that the difference between the blunting in the two configurations is indeed small, but still
49
there is a difference. Comparing the pictures at the end of blunting, the ID configuration seems to have
a larger spread of the strain bands (especially when the difference in zoom of the pictures is taken into
account). The larger blunting in this area may help spread the strains, causing less crack extension.
TT
PY + start blunting
Blunting Blunting End of blunting Pm
ID
Start of blunting
PY + Blunting Blunting End of blunting Pm
Figure 53. Strain fields for different cycle moments for TT and ID specimens for the 7S material.
Table 13. Measured blunting distances, measured crack extensions and relative blunting distances for all specimens.
Steel Notch Δb
[mm] Δa
[mm] Δb/Δa
[%]
X70-S TT 0.18 1.88 9.7
ID 0.20 1.73 11.8
X70-M TT 0.26 2.20 12.0
ID 0.26 2.39 10.8
X80-M TT 0.21 2.28 9.4
ID 0.19 2.69 7.0
X80-K TT 0.56 2.66 20.9
ID 0.34 2.72 12.6
Figure 54 shows the strain fields for the 7U material. There is quite a big difference in strain fields when
the limit load is reached. The ID specimen has a more pronounced strain field which has higher values
50
(the scale is higher). However, the total blunting distance seems to be unaffected by this higher strain.
The crack extension on the other hand is longer for the ID specimen. The strain bands seem to be
shorter and broader for the TT specimen, which seems to give a better distribution.
TT
PY Start of blunting
Blunting End of blunting Pm
ID
PY Start of blunting
Blunting End of blunting Pm
Figure 54. Strain fields for different cycle moments for TT and ID specimens for the 7U material.
Figure 55 shows the results for the 8U material. The same phenomenon as the 7U material is observed.
The TT orientation has broader and shorter strain bands with a lower strain scale compared to the ID
configuration. The blunting is in absolute value longer for the TT specimens. The larger distribution of
the strain may cause more blunting around the crack tip. This spread also causes lower strain values,
which in turn gives lower crack extensions.
Figure 56 gives the strain fields for the spiral X80 pipe. Looking at Table 13 it can be seen that the 8S
material has a high relative blunting distance, especially for the TT configuration (20% of the crack
extension). Although the relative blunting of the ID configuration is much smaller (12%), there is
practically no outspoken difference to be noted in the strain fields of the two configurations apart from
the difference in maximum strains.
From the previous observations, it seems that looking for effects from/on blunting by means of strain
fields is not quite clear. The blunting is a very small and localized phenomenon and is not noticeable on
the strain patterns.
51
TT
PY Start of blunting
Blunting Blunting Pm + end of
blunting
ID
PY + start of blunting
Blunting Blunting End of blunting Pm
Figure 55. Strain fields for different cycle moments for TT and ID specimens for the 8U material.
TT
PY Start of blunting
Blunting Pm + Blunting End of blunting
ID
PY Start of blunting
Blunting Pm + Blunting End of blunting
Figure 56. Strain fields for different cycle moments for TT and ID specimens for the 8S material.
52
7 Conclusions
7.1 Unloading compliance method SENT tests have already proven their great use within the strain based design, however the lack of a
standardized methodology brings with it that there are many different procedures available. The
unloading compliance (UC) and the direct current potential drop (DCPD) methods are the two used
methods for estimating the crack growth. Both methods have already been researched in previous
studies at the UGent. The results showed that both methods are reliable and yield similar accuracy.
Because of this the unloading compliance method was used in this thesis for its more hands-on
approach. A procedure developed by BMT fleet, which is often used at UGent, was used to perform the
UC tests. The procedure is mainly based on the ASTM standard for SENB specimens and has quite strict
requirements in terms of usable results.
Together with the unloading compliance, the double clip gauge method was applied to obtain the
required crack tip opening displacements (CTOD) and crack mouth opening displacements (CMOD). The
double clip gauge method allows accurate measurements of these parameters with little specimen
preparation needed. The accurate measurement of the CMOD is needed in order to calculate the
material’s compliance from the unloading cycles. From this compliance the crack growth is calculated.
After each test a post mortem analysis was done. The specimens were heat tinted, cooled and broken in
a brittle manner in order to obtain distinct fracture surfaces. The broken specimens were put under a
microscope to obtain detailed images of the fracture surfaces. On these surfaces nine points average
measuring methods were used to obtain the initial and final crack depths. The difference between these
two gives the ductile crack extension which is needed to construct the resistance curve. An attempt was
also made to measure the blunting on these surfaces. The blunting region was determined as the
plastically deformed zone bordering the machined notch and its length was then measured using the
nine points average method.
The measured results were then compared to the estimated values from the UC method. The measured
initial crack size was compared to the estimated initial crack size, obtained by the compliance and CTOD
values. The estimated value always over-estimated the initial crack size. The same was true for the final
crack size values, which were also over-estimated by the UC method. The subtraction of the final and
initial estimated values however, cancelled out this over-estimation giving a good correspondence with
the measured crack extension.
The measured blunting was compared to a method that obtains the blunting from CTOD values at crack
initiation. Although the used formula in this method seems rather simplistic, the measured values
corresponded quite well with the results from the CTOD method, showing that the blunting can be
estimated using either method.
The accuracy of the UC method seems to be strongly material dependent. Four different materials were
tested and for each material two different notch locations were used. In accordance with the ASTM
standard, three specimens of each configuration were tested. Some materials showed more scatter
between the same specimens. This was especially true for the X80 spiral pipe material. For some
materials, only higher scatter is observed for a certain notch configuration. This was the case for the X70
53
UOE pipe, which had a lot of scatter for only one notch configuration. For the rest of the specimens the
results were quite satisfactory.
7.2 Obtained resistance curves The combination of the UC and double clip gauge method allow easy construction of the R-curves. The
accuracy of these R-curves is strongly dependent on the scatter of the obtained parameters from the
two methods. Because of the high scatter of the 8S material, the obtained R-curves were rendered
unreliable and non-representative and were therefore omitted. The other materials yielded quite
accurate R-curves which showed relative small scatter bands.
The R-curves allow a direct and easy comparison in terms of tearing resistance between the different
materials and notch locations. The results showed that the weakest material in terms of strength had
the best tearing resistance. It was concluded that the yield-to-tensile ratio and the uniform elongation
were more determining parameters than strength in terms of tearing resistance. A low yield-to-tensile
ratio together with a high uniform elongation is preferred.
Furthermore, continuous yielding behavior is a desired trait of the material. The comparison between
the two X70 steels was a good example of that. Both steels have a comparable stress-strain curve
(mostly post yield), except that the spiral pipe showed discontinuous yielding, whilst the UOE pipe did
not. The R-curves for the former were much worse because of that.
In order to map the material heterogeneity, hardness measurements have been conducted. The 7S and
8U materials showed a pronounced softer region in the center of the material. This is most likely caused
by the rolling process of steel plates from which the pipes are formed. This heterogeneity had a clear
influence on the crack front (and thus crack growth). The softer central region seemed to have a lower
tearing resistance. This was seen for the 8U material for which the crack growing into this region (ID
notch) yielded worse R-curves.
A shallow analysis of the microstructure of the specimens showed how the 7U material obtained its
superior qualities in terms of strain based design. Its matrix containing harder particles that obstruct
further deformation showed to be preferable to the one of the 8S steel which lacked such deformation
impeding components once yielding commences.
7.3 DIC analysis Digital image correlation was also used on one specimen for each configuration. The UGent has an
internal procedure that was developed by previous studies. The procedure fully describes the process
and yields excellent results.
The DIC method allows quite accurate calculations of the deformation on the specimen’s surface. From
these deformations the 5mm CTOD method was used to measure the CTOD. This was then compared to
the obtained values from the double clip gauges. Both measurements were quite in accordance with
each other. The CTOD however estimates the crack tip opening a bit higher for larger deformations, but
this an inherent trait to the manner of measuring.
54
An attempt was made to analyze the strain fields during the blunting phase. The analysis did not give
conclusive results. The blunting seems to be too small scaled to be able to detect in a strain field
analysis.
7.4 Future work and recommendations Although the used unloading compliance method yielded satisfying results, there are still some flaws
present. The over-estimation of the initial and final crack is one of those flaws. Half of the specimens
had a difference between measured and estimated value that was higher than 0.5mm. If the ASTM
standard is strictly followed, these results would be deemed unusable. Because of this, the usage of the
estimated value a0q needs to be improved or another method needs to be applied.
A second problem that occurred with the UC method was the high scatter for the 8S material. The cause
seemed to be the high yield-to-tensile ratio of the material. This causes high local deformations which
negatively influence the accuracy of the measurements. Using the direct current potential drop method
may be better suited to estimate the crack length, since the current will not be influenced by this
material behavior.
Lastly, the material microstructure seemed to be very determinative for the tearing resistance. A closer
study to the relation between the different microstructures and their influence on the tearing resistance
might be useful.
55
References [1] Institute for Energy Research, “Statistical review of World Engergy 2013”, source:
http://instituteforenergyresearch.org/analysis/statistical-review-of-world-energy-2013-viva-la-shale-revolucion/, accessed on: 28-04-2015.
[2] Macia, M.L., Kibey, S.A., Arslan, H., Bardi, F., Ford, S.J., Kan, W.C., Cook, M.F. and Newbury, B., 2010, “Approaches to Qualify Strain-Based Designed Pipelines”, International Pipeline Conference, Calgary, Alberta, Canada, paper n° IPC2010-31662.
[4] McLamb, M., Hopkins, P., Marley, M., and Nessim, M., 2002, "A Justification for Designing and Operating Pipelines up to Stresses of 80% Smys", International Pipeline Conference, Calgary, Alberta, Canada, paper n° IPC02-27007.
[5] Liessem, A., Knauf, G. and Zimmermann, S., 2007, “Strain Based Design – What the Contribution of a Pipe Manufacturer Can Be”, International Offshore and Polar Engineering Conference, Lisbon, Portugal, paper n° ISOPE-2007-SBD14.
[6] Rice, J.R., 1968, “A Path Independent Integral and the Approximate Analysis of Strain Concentration by Notches and Cracks”, Journal of Applied Mechanics, 35, pp379-386.
[8] Shih, C. F., 1981, "Relationships between the J-Integral and the Crack Opening Displacement for Stationary and Extending Cracks", Journal of Mechanics and Physics of Solids, 29, pp. 305-326.
[9] Larson, L.H., 1983, “Introduction to the Advanced Seminar ASFM4 on Elastic-Plastic Fracture Mechanics”, Proceedings of the 4th Advanced Seminar on Fracture Mechanics, Ispra, Italy, pp. 1-12.
[10] Wang, Y.Y., Liu, M. and Song, Y., 2011, "Second generation models for strain-based design," Pipeline Research Council International (PRCI), Houston, Texas, USA
[11] Verstraete, M., 2014, “Experimental-Numerical Evaluation of Ductile Tearing Resistance and Tensile Strain Capacity of Biaxially Loaded Pipelines”, Ph.D. thesis, Ghent University, Ghent, Belgium.
[12] Cauwelier, M. and Maelfait, J., 2012, "Evaluatie van Ductiele Scheuruitbreiding in Pijpleidingstaal," Unpublished master's thesis, Ghent University, Ghent, Belgium.
[13] Henry, B. S., and Luxmoore, A. R., 1997, "The Stress Triaxiality Constraint and the Q-Value as a Ductile Fracture Parameter", Engineering Fracture Mechanics, 57, pp. 375-390.
[14] Zhu, X.K. and Joyce, J.A., 2012, "Review of fracture toughness (G, K, J, CTOD, CTOA) testing and standardization" Engineering Fracture Mechanics, vol. 85, pp. 1-46.
[15] Thaulow, C., Ostby, E., Nyhus, B., Zhang, Z. L., and Skallerud, B., 2004, "Constraint Correction of High Strength Steel - Selection of Test Specimens and Application of Direct Calculations", Engineering Fracture Mechanics, 71, pp. 2417-2433.
[16] Cravero, S., Bravo, R. E., and Ernst, H. A., 2008, "Constraint Evaluation and Effects on J-R Resistancecurves for Pipes under Combined Load Conditions", International Offshore and Polar Engineering Conference, Vancouver, British Columbia, Canada, pp. 149-156.
[17] Nevalainen, M., and Dodds, R. H., 1996, "Numerical Investigation of 3-D Constraint Effects on Brittle Fracture in SE(B) and C(T) Specimens", International Journal of Fracture, 74, pp. 131-161.
56
[18] Verstraete, M., Hertelé, S., Van Minnebruggen, K., Denys, R., and De Waele, W., 2013, "Considerations in Selecting Laboratory Scale Test Specimens for Evaluation of Fracture Toughness", Sustainable Construction and Design, 4, pp. 1-9.
[19] Zhang, J. X., and Shi, Y. W., 1997, "The Effect of Welding Mechanical Heterogeneity on Fracture Toughness Feature of Base Metal", International Journal of Pressure Vessels and Piping, 72, pp. 199-202.
[20] Det Norske Veritas, 2006, RP-F108: Fracture Control for Pipeline Installation Methods Introducing Cyclic Plastic Strain.
[21] Shen, G. W., Gianetto, J. A., and Tyson, W. R., 2009, “Measurement of J-R curves using single-specimen technique on clamped SE(T) specimen”, Proceedings of 19th International Offshore and Polar Engineering Conference, Osaka, Japan, pp92-99.
[22] Shen, G, Bouchard, R, Gianetto, JA, and Tyson, WR, “Fracture Toughness Evaluation of High Strength Steel Pipe”, Proceedings of PVP2008, ASME Pressure Vessel and Piping Division Conference, Chicago, Illinois, USA, paper n° PVP2008-61100.
[23] ExxonMobil, 2010, "Measurement of Crack Tip Opening Displacement (CTOD) - Fracture Resistance Curves Using Single-Edge Notched Tension (SENT) Specimens", ExxonMobil Upstream Research Company, Houston
[24] Fairchild, D. P., Kibey, S., Tang, H., Krishnan, V. R., Macia, M. L., Cheng, W., and Wang, X., 2012, "Continued Advancements Regarding Capacity Prediction of Strain-Based Pipelines", International Pipeline Conference, Calgary, Alberta, Canada, paper n° IPC2012-90471.
[25] Moore, P. L., and Pisarski, H. G., 2012, "Validation of Methods to Determine CTOD from SENT Specimens", International Offshore and Polar Engineering Conference, Rhodes, Greece, pp. 577-582
[26] American Society of Testing and Materials, 2011, E1820 - Standard Test Method for Measurement of Fracture Toughness.
[27] Geldhof, N., Van Iseghem, J., 2013, “Evaluatie van scheurgroei en rekcapaciteit in gelaste constructies”, Unpublished master's thesis, Ghent University, Ghent, Belgium.
[28] Shen, G., Tyson, W. R., Gianetto, J. A., and Park, D.-Y., 2010, "Effect of Side Grooves on Compliance, J-Integral and Constraint of Clamped SE(T) Specimen", Pressure Vessels and Piping Conference, Bellevue, Washington, USA, paper n° PVP2010-25164.
[29] Lan, W., Deng, X., and Sutton, M. A., 2010, "Investigation of Crack Tunneling in Ductile Materials", Engineering Fracture Mechanics, 77, pp. 2800- 2812.
[30] Riemelmoser, F. O., Pippan, R., Weinhandl, H., and Kolednik, O., 1999, "The Influence of Irregularities in the Crack Shape on the Crack Extension Measurement by Means of the Direct-Current-Potential-Drop Method", Journal of Testing and Evaluation, 27, pp. 42-46.
[31] British Standard Institution, 2014, BS 8571:2014 - Method of Test for Determination of Fracture Toughness in Metallic Materials Using Single Edge Notched Tension (SENT) Specimens.
[32] Tang, H., Minnaar, K., Kibey, S., Macia, M. L., Gioielli, P., and Fairchild, D. P., 2010, "Development of the SENT Test for Strain-Based Design of Welded Pipelines", International Pipeline Conference, Calgary, Alberta, Canada, paper n° IPC2010-31590.
[33] Shen, G., and Tyson, W. R., 2009, "Crack Size Evaluation Using Unloading Compliance in Single-Specimen Single-Edge-Notched Tension Fracture Toughness Testing", Journal of Testing and Evaluation, 37, pp. 347-357.
57
[34] Wang, E., Omiya, M., 2014, “Finite Element Analysis of Crack Mouth Opening Displacement Compliance in Crack Length Evaluation for Clamped Single Edge Tension Specimens”, Fatigue & Fracture of Engineering Materials & Structures, 38, pp.315-329.
[35] Cravero, S., Ruggieri, C., 2007, “Estimation procedure of J-resistance curves for SE(T) fracture specimens using unloading compliance”, Engineering Fracture Mechanics, 74, pp. 2735-2757.
[36] Tyson, B., Ding, P., Wang, X., 2014, “Elastic compliance of single-edge-notched tension SE(T) (or SENT) specimens”, Frattura ed Integrità Strutturale, 30, pp. 95-100.
[40] Brensing, K.H., Sommer, B., “Steel Tube and Pipe Manufacturing Processes”, source: www.smrw.de/files/steel_tube_and_pipe.pdf, accessed on: 21-11-2014.
[41] Steelpipe, “Spiral Welding Pipes”, source: http://www.steelpipe.co.nz/about-steelpipe/process/mills/, accessed on: 21-11-2014.
[42] Nagai, K., Shinohara, Y., Sakamoto, S., Tsuru, E., Asahi, H. and Hara, T., 2009, “Anisotropic strain aging behavior of high strength UOE linepipe”, Proceedings of 19th International Offshore and Polar Engineering Conference, Osaka, Japan, pp56-60.
[43] American Petroleum Institute, 2007, API-5L, Specification for Line Pipe, 44th edition.
[44] Landes, J. D., 1995, "The Blunting Line in Elastic-Plastic Fracture", Fatigue and Fracture Of Engineering Materials and Structures, 18, pp. 1289-1297.
[45] Brockenbrough, R. L., 1995, "Effect of Yield-Tensile Ratio on Structural Behavior - High Performance Steels for Bridge Construction", Final Report Naval Surface Warfare Center and Federal Highway Administration.
[46] Hertelé, S., 2012, "Coupled Experimental-Numerical Framework for the Assessment of Strain Capacity of Flawed Girth Welds in Pipelines", Ph.D. thesis, Ghent University, Ghent, Belgium.
58
List of figures Figure 1. World energy consumption by source in 2012 [1]. ........................................................................ 1
Figure 2. Trans-Alaska pipe line traversing an arctic region [3]. ................................................................... 1
Figure 3. Safety margin for stress based design (a) and safety margin for strain based design (b). ............ 2
Figure 4. Blunting of the crack tip [7]. ........................................................................................................... 3
Figure 5. Schematic representation of the δ90 and the δ5 definitions for CTOD [10]. .................................. 4
Figure 6. Example of a random contour integral around the crack tip to determine the J-integral [7]. ...... 4
Figure 7. Schematic representation of different resistance curves. ............................................................. 5
Figure 8. Schematic representation of blunting and crack initiation [9]. ..................................................... 6
Figure 9. Schematic representation of the tangency method [10]. .............................................................. 6
Figure 10. Schematic representation of a compact tension (CT) specimen [11]. ......................................... 7
Figure 11. Schematic representation of a single edge notch bending (SENB) specimen [11]. ..................... 7
Figure 12. Schematic representation of a single edge notch tension (SENT) specimen with pin loaded
configuration (a) and clamped configuration (b) [11]. .................................................................................. 8
Figure 13. Schematic representation of a double edge notch tension (DENT) specimen [11]. .................... 8
Figure 14. Schematic representation of a center cracked tension (CCT) specimen [11]. ............................. 8
Figure 15. Difference in resistance curve for identical specimens under a bending and a tensile load. ..... 9
Figure 16 Schematic representation of a SENT specimen obtained from a pipe. ...................................... 10
Figure 17. Dimensional parameters of a clamped SENT specimen [20]. .................................................... 10
Figure 18. Schematic representation of a fatigued and a machined crack [27]. ........................................ 12
Figure 19. Detailed view of a side groove. .................................................................................................. 12
Figure 20. Schematic representation of the clip gauge blades for measuring the CTOD. .......................... 14
Figure 21. Example of a typical unloading compliance curve (a) and the determination of the compliance
from the unloading slope (b). ...................................................................................................................... 14
Figure 22. Calculated crack depth comparison for 7S-TT specimen (a) and 7S-ID specimen (b). .............. 17
Figure 23. Negative crack growth in a compliance curve (a) and the influence of relative crack size on
Figure 26. Schematic representation of subset shifting to determine surface displacements [37]. .......... 20
Figure 27. Schematic representation of the two-camera setup. ................................................................ 21
Figure 28. Picture taken by DIC camera during test of 8U-DIC-TT specimen (a) and the calculated Von
Mises strains displayed on the surface using the correlation software (b). ............................................... 21
Figure 29. δ5 determination using DIC correlation software. ..................................................................... 22
Figure 30. Microscope image of a fracture surface with nine points average measurement lines. .......... 23
Figure 31. Detailed view of the blunting region on the fracture surface.................................................... 23
Figure 32. Schematic representation of the spiral forming process [41]. .................................................. 24
Figure 33. Schematic representation of the UOE forming process [42]. .................................................... 25
Figure 34. Schematic representation of the inner diameter (ID) notch (a) and the through thickness (TT)
notch within a pipe (b). ............................................................................................................................... 26
Figure 35. Specimen fixed by mechanical clamps in the 150kN test rig (a) and specimen clamped by
hydraulic clamps in the 1000kN test rig (b). ............................................................................................... 29
59
Figure 36. Compliance vs CMOD and initial linear increase for the 7S-TT1 specimen (a), the 7U-TT1
specimen (b), the 8U-TT1 specimen (c) and the 8S-TT1 specimen (d). ...................................................... 30
Figure 37. Difference in scatter between the 7S TT specimens (a) and the 8S TT specimens (b). ............. 31
Figure 38. Comparison between estimated and measured initial crack size. ............................................ 31
Figure 39. Comparison between estimated total crack propagation obtained by the unloading
compliance method and the total crack propagation obtained by measurements on the crack surfaces.
Figure 40. Determination of crack initiation (a) and determination of CTODini (b). ................................... 34
Figure 41. Comparison between blunting measured by the nine points average method on the fracture
surfaces and the estimated blunting based on CTOD at crack initiation. ................................................... 34
Figure 42. Scatter of calculated R-curves for specimens 7S-TT (a), 7S-ID (b), 7U-TT (c), 7U-ID (d), 8U-TT
(e), 8U-ID (f), 8S-TT (g) and 8S-ID (h). .......................................................................................................... 36
Figure 43. R-curve comparison between TT and ID notch location for material 7S (a), 7U (b) and 8U (c). 39
Figure 44. R-curve comparison between materials for the same notch configuration. ............................. 40
Figure 45. Stress-strain curves for all testing materials. ............................................................................. 41
Figure 46. Continuous and discontinuous yielding behavior [46]. .............................................................. 43
Figure 47. Hardness map for material 7S (a), 7U (b), 8U (c) and 8S (d). ..................................................... 44
Figure 48. Schematic overview of the hardness regions and notch and crack extension distributions of
the hardness regions for the 7S specimen. ................................................................................................. 44
Figure 49. Crack front shape for TT notched specimens for material 7S (a), 7U (b), 8U (c) and 8S (d). ..... 45
Figure 50. Microscopic image obtained by optical microscopy (OM) for the 7U material (a) and the 8S
material (b). ................................................................................................................................................. 46
Figure 51. Definition of the δ5 measurement using DIC software (a) and results for obtained values of δ5
and δ90 per unloading cycle for specimen 7S-DIC-ID (b). ............................................................................ 47
Figure 52. Comparison between CTOD values obtained by δ5 and δ90 for the final cycle of each specimen.