Professional Documents
Culture Documents
Over the past several years, a number of specific Engineering Critical Assessment (ECA)
approaches have been developed to evaluate the fitness-for-purpose of pipeline girth weld
containing fabrication flaws [1-9]. To date, these methodologies can be applied when the
applied stresses in the axial direction remain within their elastic stress-strain response limits.
The primary input parameters for a conventional ECA analysis are: flaw type and size, level of
applied stress, and toughness of the material containing the flaw. Here, it should be
emphasized that an ECA is critically dependent on the ability to reliably find and accurately
determine the size and location of any flaw that is present in the weld area.
Current ECA methodologies do not provide the desired guidance for the evaluation of the
interdependent input factors affecting girth weld integrity. Fortunately, the critical (failure)
flaw sizes are normally larger than those predicted by the mathematical model. This is
because:
• The prediction assume that the material is homogeneous while the weld metal is
normally stronger than the pipe metal (weld metal mis-match effect)
• flaws are represented by their containment rectangle and, this assumption increases
the level of conservatism for the possible and very likely instance of irregular-shaped
flaws.
A better understanding of the preceding factors would make the assessment more accurate.
However, these factors still do not take into account all aspects of the problem. Fig A1 gives
a general overview of the complexities associated with an ECA. Although existing analyses do
not address the many complex interactions, the challenge remains to understand and to find
the right balance between the many factors affecting the integrity of pipeline girth welds.
3
Allowable
flaw size limits
Minimum
toughness and tearing
Surface-breaking vs buried flaws behaviour ?
SMAW, GMAW, P-GMAW, FCAW ?
Accuracy in height sizing ? (Type of flaws)
Interaction rules ?
How to measure ?
Charpy V or CTOD ?
Finally, it might be noted that the authorities do not always accept ECA based allowable flaw
sizes. Beyond the psychological impact, this problem can only be overcome if the
mathematical model used is supplemented with experimental evidence demonstrating the
validity of the predicted flaw sizes. For the case where failure is controlled by yield and plastic
collapse, such evidence is now available as a result of comprehensive experimental
investigations [10-18].
A standard ECA approach ensures that the calculated flaw size is acceptable both from a
(brittle/unstable) fracture and a stable (plastic collapse or ductile failure) failure viewpoint:
• For the fracture (or toughness dominated) assessment, a fracture mechanics CTOD
based design curve is used to ensure that failure by unstable fracture will not occur for
known flaw dimensions, applied stress and CTOD toughness.
• The plastic collapse (or flaw size dominated) assessment prevents failure by local or
overall yielding of the flawed cross section. For that purpose, flaw size and applied
stress are compared with a notional flow stress, derived from the tensile properties of
the flawed region.
The fracture assessment finds it origin in the early 1970’s. At that time, achieving adequate
weld metal toughness properties was a real challenge. With advances in welding technology,
brittle fracture is nowadays only a concern for very low temperature applications. Research
over the last several years has shown that the primary failure mode of flawed girth welds is
by plastic collapse. Moreover, large-scale tests show that flawed girth welds can exceed the
pipe metal yield strength and fail at strain levels many times larger than the strain at the
yield point [11].
The quantitative nature of an ECA analyses implies a need for quantitative input data. Safe
ECA predictions are obtained when the driving force, which causes a flaw to extend, and
toughness, which is the resistance of the material containing the flaw(s) to this extension,
and the flaw size can be adequately determined. This implies that any assessment procedure
modelling the relationship between material properties, applied load, flaw size and toughness
represents the actual situation. In deterministic terms, this means that:
• The most onerous applied stresses or strains acting on the flawed weld are
conservatively estimated,
• The tensile properties of the girth weld meets those of the pipe metal (Weld metal yield
strength mis-match),
• The actual flaw dimensions and, in particular, the height of a crack-like flaw as well as
its position can be accurately measured (Flaw sizing).
The above observations are simply meant to say that uncertainties surrounding the values of
the input parameters will be transmitted through to the final result. Furthermore, an ECA
explicitly assumes that the provisions/assumptions used in the derivation of the
calculated/predicted flaw size limit are adhered to during construction while the mechanical
and toughness properties across and along the weld are consistent. Therefore, flaw
acceptance criteria must be based on “achievable” material properties and proven inspection
techniques.
Aside from the above requirements, the equation(s) used to calculate the allowable flaw size
should be:
It should also be emphasized that more accurate solutions can be obtained if realistic flaw
interaction criteria for assessing the effect of adjacent flaws and suitable recategorisation
procedures of near surface breaking flaws should be used.
Current ECA methodologies have limitations when addressing higher strength pipes or higher
strain applications. In particular, with the extension of pipeline applications to high strength
pipeline steels (X80 and above), more severe service design conditions, and the introduction
of limit state design requirements, current ECA approaches are not equipped to handle this
type of problems. For example, for non-elastic design conditions, the level of weld metal yield
strength mismatch is a factor that requires due consideration in an ECA analysis. An
undermatched girth weld may be a potential problem that limits the overall deformation
capacity. However, existing ECA procedures do not provide adequate guidance for this design
situation.
2 FAILURE MODES
Girth weld failure caused by a pre-existing flaw can occur in the loading range from brittle
fracture (applied stresses below yield - brittle facture) to plastic collapse (overload of
remaining ligament or remote pipe section - ductile failure). Between these two extremes
failure occurs by contained yielding, commonly indicated as elastic-plastic fracture
behaviour).
Brittle fracture (elastic or elastic-plastic behaviour) occurs when the yielding at the flaw tip is
contained. To prevent brittle fracture the pipeline industry includes in design codes toughness
requirements and weld quality inspection criteria. Failure by plastic collapse requires that a
lower bound limit on toughness. The point is that flawed girth welds can be made sufficiently
tough so that a relaxation of the triaxial state of stress ahead of the flaw tip is easily
achieved. Thus, the failure condition of a flawed pipeline girth welds can be described by the
tensile (flow stress) properties of the materials in the weld region.
In a girth weld containing a part-wall flaw, plastic collapse or ductile failure can occur by
Ligament (local) yielding, Global yielding (or Net Section Yielding, NSY) or Overall (or Gross,
pipe metal, Section Yielding, GSY), Fig. A2.
• Local (or ligament) collapse occurs when yielding is confined to the ligament ahead of the
flaw tip and the opposite back face (Fig. A2 – left figure).
• Global collapse (or Net Section Yielding - NSY) occurs when the complete cross sectional
area containing the flaw becomes plastic. In this case, yielding is uncontained. Yielding of
the entire cross section is only possible when the weld is subjected to tensile loading (Fig.
A2 – middle).
• Overall collapse (or Gross Section Yielding – GSY) occurs when the applied stress in the
remote cross sections exceeds the yield strength of the pipe. GSY is only possible if the
flawed material has strain hardening capabilities. Strain hardening acts to spread the
extent of the plasticity away from the flawed cross section (Fig. A2 – right figure).
Local
Collapse
Surface
Flaw
A Plastic
A
Deformation Global Overall
(NSY) (GSY)
Collapse Collapse
Cross Section AA
For strain hardening materials, the three plastic collapse modes can merge from one mode
into the other with increasing load. Local collapse is normally not the limiting deformation
mode since the materials used in modern pipelines can easily meet the minimum toughness
for collapse by NSY.
The occurrence of NSY or GSY collapse depends upon the level of applied stress, the flaw size
and location, the level of weld metal yield strength mismatch and the strain hardening
capability (or Y/T ratio) of the material containing the flaw. The transition from NSY to GSY
depends essentially on flaw size and strain hardening (or Y/T ratio).
The assessment of flaw severity depends essentially on the toughness properties of the
flawed material whereas the procedures of elastic (KIc-approach) or elastic-plastic (CTOD-
approach) fracture mechanics can be applied to assess flaw criticality. The allowable flaw size
can be derived from the following generic equation:
Above a certain level of toughness the (critical) values of stress and flaw size are insensitive
to toughness. On the other hand, for new constructions, codes or fabrication documents
require the pipe/weld metal toughness levels warranting plastic collapse as failure mode.
Thus, flawed girth welds in properly designed pipelines have the ability to deform plastically.
The point is that girth welds can be made sufficiently tough so that the failure conditions of
flawed girth welds are governed by the flow properties of the material.
For applied remote stresses below yield, plastic collapse predictions use the flaw dimensions
and the remote (nominal) applied stress to calculate the stress level developed in the flawed
(net) cross section. This stress level is compared with a notional flow stress derived from the
tensile properties of the material containing the flaw:
Several plastic collapse assessment models have been proposed which can be grouped into
three categories [19-25]. The first group embraces the models based on a theoretical analysis
of flat plates. The second group involves those based on a theoretical analysis of actual pipe
geometries. The final group of models is those derived from experimental results [24-25].
Standard plastic collapse assessments assume that the plasticity is confined to the flawed
cross section (Net Section Yielding). For design conditions requiring that the applied loads
cause plastic deformations in the remote pipe sections, existing plastic solutions cannot be
used.
Conventional ECA approaches do not exclude failure by brittle fracture. This assumption is not
very relevant for new pipeline constructions. Flawed girth welds in properly designed pipelines
fail by plastic collapse because:
• the selection of the weld metal is based on the assumption that undetected
(workmanship-type) weld flaws may be present. This assumption requires that the
toughness of the weld metal and HAZ regions should be sufficiently high to exclude the
possibility of a brittle fracture.
• it is now common practice to specify filler metals with a Charpy test requirement to ensure
adequate resistance to brittle fracture.
Having excluded the occurrence of brittle fracture, the assessment of a flawed girth welds
can be concentrated on plastic collapse. In particular, an ECA for pipeline girth welds can be
simplified since the failure conditions are controlled by geometrical factors (wall thickness
and pipe diameter) and material properties (pipe and weld metal strengths) rather than by
toughness. Stated in another way, beyond a toughness threshold, increased toughness does
not increase the flaw size limit. This is because plastic collapse solutions relating the applied
stress to the flow stress and flaw size do not include a toughness parameter.
EPRG (European Pipe Research Group) has used experimental correlations to determine the
lower bound Charpy V notch (CVN) impact values for plastic collapse by pipe metal yielding
(performance requirement = 0,5% pipe metal strain, GSY). The required CVN impact
energies for collapse controlled failure of matched / overmatched welds at minimum
operating temperature are [26]:
The toughness requirements are based on correlations between the results of curved wide
plate tests incorporating 3 mm deep surface breaking root cracks and Charpy V impact tests.
The Universiteit Gent database that was used for the EPRG study included girth welds made
in X52, X60, X65 and X70 pipeline steels in thickness ranging from 6,6 mm to 25,4 mm. The
correlations have also demonstrated that for a wall thickness less than 12,7 mm, the impact
requirement can be reduced to 27 J (minimum individual value: 20 J) [27]
For practical reasons, the CTOD fracture toughness cannot always be measured. For
example, it is virtually impossible to determine a representative CTOD value of girth welds in
older pipelines. For this and other similar situation, the Charpy V impact values are normally
available. However, to satisfy the possible need for CTOD data and to overcome the practical
difficulties, a number of attempts have been undertaken to correlate Charpy V impact energy
and CTOD toughness. Contrary to some opinions, reasonable correlations can be found
provided that care is taken that both tests sample the same region of the weld deposit [28].
Such a correlation is not yet available for pipeline girth welds.
To provide a workable basis for the prediction of CTOD using CVN test data for girth welds, a
substantial volume of pipeline specific CVN and CTOD results produced by Labo Soete - RUG
were compared with the aim to identify a lower bound correlation, Fig. A3. The plots in Fig.
A3 contain comparisons between both minimum (open diamonds) and mean values (solid
diamonds). The test results were taken from a database containing a broad range of weld
0.30
Average values
0.25 Minimum values
UM Welds
0.20
CTOD (mm)
0.15
0.05
0.00
0 40 80 120
CVN (J)
La bo S oe t e - G e nt
0.8
0.6
CVN = 40 J
CTOD (mm)
0.4
0.2
CVN = 40 J (EPRG)
CTOD = 0,0473 mm
0
0 50 100 150 200 250 300
CVN (J)
Labo Soete - Gent
Fig. A3 – Correlation between mean and minimum CVN and CTOD properties
metal strength and toughness levels (see also Part B) [29]. The CVN and CTOD values were
measured at the same temperature while the mean values were determined for at least three
specimens. The Charpy specimens were taken from the subsurface weld root region.
At first sight, and as expected, a straightforward correlation between CVN and CTOD cannot
be made. The large scatter is due to fact that the data was not grouped by welding process
and wall thickness. The other reasons for the poor correlations are discussed in [29].
However, one might conclude that no universal CVN-CTOD correlations can be found.
By fitting a lower bound envelope (straight line) to the experimental data, the Charpy V
impact energy can be used to estimate the lower bound CTOD toughness of girth welds. The
following equations give a conservative description of the correlation between the CVN and
CTOD properties of pipeline girth welds:
Eq. (A1) was derived for Charpy impact properties lower than 40 J (Fig. A3a). For Charpy
values exceeding 40 J, Eq. (A2) can be used (Fig. A3b) It should be noted that Eqs. (A1) and
(A2) gives the same value if the CVN impact energy is equal to 40 J.
In using Eqs. (A1) and (A2) the fracture toughness, CVN, or the notch toughness, CTOD, are
to be determined at the same test temperature. Thus, where the CTOD test is impractical,
the lower bound weld metal CTOD toughness may conveniently be estimated from Charpy
impact data using an empirical formulation provided by Eqs. (A1) and (A2). It should be
noted here that the equations should not be used for wall thickness exceeding 25.4 mm.
Another expression developed by the Japanese Welding Society yields similar predictions. The
Japanese fitness-for-purpose document WES 2805-1980, provides an expression that allows
the CTOD to be derived from CVN data [31]. The correlation is based on the results of
experiments conducted on plate, weld metal and HAZ tests of various kinds of steels:
CTOD = 0 . 01 C VN (A3)
t
T + 112 − YS − 5 t ( in kgm )
where t = plate thickness, YS = yield strength and T = design temperature or CTOD testing
temperature (°C).
For material with a yield strength of 550 MPa (X80) and a wall thickness of 25 mm to be used
at 0 °C, Eq. (A3) gives for a CVN value of 40 J to be obtained at +31 °C a CTOD value of 0.04
mm at 0 °C. This requirement compares well with that derived from Esq. (A1) or (A2).
For a Charpy V value of 40 J, Eq. (A1) allows concluding that the assessment of flaw in a
girth weld with a CTOD toughness of 0,0473 mm can be based on plastic collapse. Using full-
scale test results on circumferential part-wall flaws Hopkins proposed a CTOD of 0,05 mm
(without safety factor) as the “threshold’ toughness above which failure can be based on
plastic collapse [32-33]. A supplementary PD 6493:1991 fracture mechanics calculation by
Hopkins for buried flaws of any length in pipeline girth welds having a CTOD toughness of
0,05 mm led to the conclusion that brittle fracture at stresses up to yield strength can be
excluded provide their height does not exceed 3 mm [26].
Coote [twi p21] arrived at a similar conclusion at the time he assessed the Canadian full scale
test results which were used to validate the CSA Z662 Appendix K ECA approach. In
particular he concluded, “…high strength pipe with CTOD toughness exceeding 0.10 mm,
maximum allowable imperfection sizes will be limited by the analysis to prevent plastic
collapse…”.
The previous observations allow concluding that the inexpensive Charpy V test and the CTOD
test are equivalent test methods. In addition, the CVN impact requirements of 30 J / 40 J can
be used as a simple criterion to define the boundary between fracture and plastic collapse
controlled fracture. In particular, the CTOD test is not needed to determine the failure mode
if, as demonstrated, an acceptable empirical correlation between CTOD and Charpy can be
applied. On the other hand, the possibility of expressing CTOD as a function of CVN is
attractive for situation where CTOD data is not available. However, note that CTOD values
derived from CVN-CTOD correlations provide lower bound data, and thus lower bound ECA
based allowable flaw sizes. Therefore, such correlations cannot always be a substitute for the
CTOD test if failure behaviour is toughness dependent
3 PLASTIC COLLAPSE
3.1 Background
The design of pipelines under conditions normally encountered in the oil and gas industry is
based on pressure containment, bending and thermally induced stresses. These loads cause
hoop and axial/longitudinal stresses. Codes limit the allowable stresses to prevent failure.
The criteria are stress-based and limit the maximum allowable hoop stress (or design factor)
to 72 or 80% of the specified minimum yield strength (SMYS). The maximum axial stress
level is normally restricted to 50 % of the SMYS. Using stress-based design criteria means
that the pipeline is not exposed to plastic strains since the pipeline industry uses a strain limit
of 0.5%. This limit coincides with the strain at which the value of the SMYS is defined.
The failure characteristics of a flawed girth weld subject to bending can be derived from the
failure characteristics of a tensile loaded curved pipe segment. This simplification is based on
large scale test results which have demonstrated that tensile loaded curved wide plate (CWP)
specimens fail at lower strain and stress levels than full scale bend test pieces do, Fig. A4
[10,18].
0.75
0.5
0.25
0
-70 -50 -30 -10 10 30
Fig. A4 - Comparison of full-scale bend and curved wide plate (CWP) test results
Conventional, stress-based, plastic collapse solutions presume that the limit state occurs
when the average net section stress or the limit stress for failure reaches some notional flow
stress, FS. This assumption is reasonable for large flaws, but it can be unrealistic, i.e.
conservative, for “small” flaws in strain hardening materials. The effect of strain hardening,
which can be characterised in a simplified way by the yield-to-tensile ratio, Y/T, is counted on
to delay the localisation of strain in the net section until remote (pipe metal) yielding is
achieved.
As shown in Fig. A4, the actual limit stress can exceed the predicted stress for failure. It is
further worth of note that the level of weld metal yield strength mis-match also interferes in
this deformation process. Currently, little practical guidance can be given to determine the
combinations of Y/T ratio and weld metal yield strength mis-match on the limit stress for
failure.
Conventional plastic collapse solutions use the flaw dimensions and the remote (nominal)
applied stress to calculate the stress in the flawed (net) cross section. This stress is then
compared with a notional flow stress, σf or FS, derived from the stress-strain tensile
behaviour of the material containing the flaw. For strain hardening materials, the value of the
flow stress exceeds the uniaxial yield strength.
When the net section stress acting on a flawed pipe segment of arc length, s (or plate width,
W), is equal to the flow stress, plastic collapse is obtained for an applied remote (gross)
tensile stress given by:
lh
σ pc = σ applied ( remote ) = FS [1 − ] (A4)
st
In this generic equation, σpc is the applied gross tensile stress at plastic collapse, FS is the
flow stress, l is the flaw length, h is the flaw height, s (or W) is the arc length and t is the
wall thickness.
The flow stress is an artificial material property for calculating the strength at plastic collapse
of a flawed ductile strain hardening material. The flow stress is at least equal to the yield
strength and its value is situated between the yield and ultimate tensile strength values (see
also next Section).
Eq. A4 shows that the reduction in limit load for collapse is simply proportional to the
reduction in cross sectional area caused by the flaw. However, Fig. A5 illustrates that the
actual relationship is not linear.
In Fig A5, the predicted failure stresses, using Eq. (A4) (dashed line), is compared with the
measured failure stresses (solid and open data points) for a 10 mm thick pipeline steel. As
can be seen, the predictions (dashed line) provide a comfortable margin of safety (as
expressed by the difference between the predicted and measured failure stresses – data
points above the dashed line are safely predicted) for plates failing by GSY (solid circles).
Fig. A5 also illustrates that a simple plastic collapse assessment can greatly underestimate
the limit load in the case of GSY. When the applied stress reaches the flow stress, the pipe
metal has, depending on the flaw size additional load carrying capacity. The margin of safety
decreases for NSY (open circles). On the other hand, the predicted failure stresses can be
unconservative for materials having a high propensity to stable ductile tearing in the NSY
deformation mode [34].
700
FS
600
GSY
YS NSY
550
Predicted
Actual
failure curve
failure curve
500
0 100 200
Defect length (mm)
Several numerical definitions of flow stress have been developed. Some flow stress equations
are a function of Yield Strength (YS) and Ultimate Tensile Strength (UTS), while the others
depend on yield or tensile strength. The published expressions are based either on theoretical
considerations and or on empirical correlations [35]. For general fitness-for-purpose and, in
particular, for pipeline girth weld flaw assessments, the flow stress is usually defined as the
average of the yield and ultimate tensile strengths, Eq. (A5) or 1,2.YS whichever is less (BS-
PD 6493):
YS + TS
FS = (A5)
2
With R = YS/TS, the above flow stress equation can be expressed and characterized in terms
of yield strength and the yield to tensile ratio, R:
YS (1 + R)
FS = (A6)
2R
Depending on the stress-strain properties of the material, and in particular of the Y/T ratio,
the flow stress given by Eq. (A6) will vary in its magnitude.
The differences in yield strength between the pipe metal and weld metal (yield strength mis-
nmatch) influences the plastic deformation behaviour of a flawed girth weld. However, yield
strength mis-match in the weld region complicates the prediction of the conditions of plastic
collapse.
BS 7910-2000, Section 7.2.8, states that in welded joints the tensile properties of the region
in which the flaw is located should be used. For HAZ zone regions it should be assumed that
the material has the lower of the adjacent weld metal or parent metal properties for
application to assessment of flaws in structures. The accuracy of this recommendation can be
investigated by comparing the predictions against experimental results.
Fig. A6 compares the experimental data with the predictions. In calculating the failure stress
for collapse, the flow stresses of both the pipe and weld metals have been used. The
experimental data were taken from a curved wide plate (CWP) test database containing 382
results. The test specimens contained a single surface breaking flaw in either an
undermatched or an overmatch girth weld (the range of weld metal yield strength mismatch
varied from –25 % to 54 %) while the base metal consisted of X52 up to X80 pipeline steel
grades [30].
The data points represent, as indicated, the fracture or NSY (failure strain < 0.5 %) and GSY
(failure strain > 0.5 %) failure mode. Figs A6a and A6b do not make a distinction between
under and overmatched welds. However, these data are re-plotted for matched/overmatched
and undermatched welds in Figs. A6a1, a2 (M/OM) and Fig.A6b1 and b2 (UM).
The comparison of the test results (“Measured”) with the calculated (“Predicted”) failure
stresses using Eq. (A4), illustrates that the failure stresses are conservatively predicted if the
flow stress of the weakest material is used. More specifically, the comparisons show quite
clearly that the collapse behaviour of undermatched welds should be assessed on the basis of
the flow stress of the girth weld, Fig. A6a. For the situation of flawed overmatched welds, the
flow stress of the pipe metal must be used to obtain conservative predictions. In other words,
the use of the weld metal flow stress could lead to non-conservative predictions. This also
implies that BS7910-2000 provides inaccurate information for the assessment of flaws
containted in overmatched weld metal.
Various formulae have been proposed to predict the collapse or ‘theoretical’ limit load of a
flawed material [20,22,23,36]. In an attempt to provide guidance on the selection of on the
most appropriate formula for application to flaws in ductile pipeline girth welds, Wang et al
[23] have assessed a number of possible solutions that have been developed for various
loading conditions, or a combination thereof. These solutions are not considered here for the
simple reason that they have not been developed for HSLA/TMCP pipeline steels while
experimental validation of their usefulness is rather limited.
800 800
Unconservative
Unconservative
700 700
Predicted stress (MPa)
500 500
Conservative
Conservative
800 800
FS of Weld metal FS of Pipe metal
OK
Predicted stress (MPa)
700 700
600 600
500 500
OK !
700
Predicted stress (MPa)
Predicted stress (MPa)
700
600 600
500 500
The Net Section Collapse (NSC) approach is also excluded in the present review because the
NSC analysis assumes that failure occurs once the entire cross section containing the flaw
becomes plastic. For pipeline girth welds under bending the NSC approach will give on-
conservative predictions because the stresses occurring in pipe vary, depending on the
circumferential position, from tensile to compression.
For all these reasons, the discussions below are concentrated on experimentally validated
collapse solutions. These solutions are the DEN model (used in ERPG-Tier 2) and the
modified-ligament instability model (adopted by CSA Z662 Appendix K). As will be discussed
in the next Sections, these solutions are simple but conservative since the loading condition
considered is uniform tension; i.e. the models assume that a uniform tensile load equal to the
greatest tensile stress occurs within a well-defined fraction of the pipe circumference that
contains the flaw. This is a conservative simplification of the actual loading that girth weld
flaws might experience since the most critical loading for a circumferentially oriented flaw in a
pipeline girth weld is bending.
DEN collapse model - The model is a “curved” plate solution that is extensively used by
Laboratorium Soete, Universiteit Gent, to analyse the deformation behaviour of curved wide
plate tests. This model combines Eqs. (A4) and (A6). That is, Eq. (A4) can be converted to
the following equation:
YS(1 + R) lh
σ applied = σ = [1 - ] (A7)
pc 2R st
This equation has been used in the derivation of the EPRG-Tier 2 flaw acceptance levels (see
Section 4.3). Note further that Eq. (A7) enables the quantification of the effect of YS/TS ratio,
wall thickness, level of applied stress (less than or equal to yield strength), weld-mismatch,
flaw height (or length) and flaw length (or height) to be assessed.
Failure of flaw free pipe welds under axial loading requires plastic tensile strains. As
discussed, provided the strength of the weld matches/overmatches that of the pipe, girth
welds containing (small) flaws can yield (strain > 0.5 %) in sections remote from the flaw
(Fig. A5).
Performance criterion - For the situation where pipe yielding is a credible event it is
practical to require GSY (applied strain ≥ 0.5 %) as a demonstration of satisfactory per-
formance. This requirement is conceptually different from a fracture mechanics or NSY based
plastic collapse based assessment since the GSY failure criterion can be used for fitness-for-
purpose assessments without some form of safety (see Section 5.4).
Back to the basics - The idea of using the GSY requirement as failure criterion was first
explored in the 1960’s. At that time, wide plate test results were compared to a semi-
arbitrary pass/fail strain level of 0,5 % (four times yield strain). The 0.5% yield criterion was
used to provide an answer for a particular problem related to local strain embrittlement in
“old type pressure vessel steels” [38].
The necessity for a consistent but simple and practical engineering means for assessing either
or both elastic/plastic or plastic material behaviour in wide plate test performance led Gent
University in the early 70’s to the development of the Gross Section Yielding (GSY) concept
[39-41]. The concept reflects the idea that when the material at the crack tip can strain
harden enough to compensate for the missing cross sectional area in the plane of the flaw,
the applied strain can be (uniformly) distributed all along the welded joint prior to failure.
GSY and welds - For flawed girth welds, GSY is achieved when the gross/remote section
fracture stress exceeds the pipe metal yield strength. It is easier to obtain GSY for
overmatched welds than for undermatched ones. For overmatched welds containing large
flaws it is not to be excluded that the plastic strains will be confined to the weld metal region.
For undermatched welds, the GSY requirement can be satisfied when the weld metal
possesses adequate strain hardening capability and provided the flaw size is rather small.
Benchmark - The fact that the GSY plastic collapse model has been validated by testing
some 800 large-scale specimens, the GSY criterion can be used as the benchmark in the
ensuing discussions.
The girth weld specific ECA methodologies most frequently used are detailed in API 1104
Appendix A, CSA Z662 Appendix K and EPRG-Tier 2 [3-4,8]. Before discussing the
approaches, the following points are worth mentioning:
• the methodologies generate an “allowable” flaw size and not a critical flaw size,
• detailed guidelines on NDE methods are not given (flaw sizing issue)
• none of the methodologies contain procedures for assessing the effect of weld metal
mismatch on allowable flaw size (material’s issue).
The mentioned ECA approaches are discussed in the following Sections. The effects of flaw
sizing and weld metal yield strength mismatch on allowable flaw size are discussed in
Sections 6.
The API 1104 Appendix A approach is based on the original COD design-curve concept.
However, an adjustment is made to the flaw size to account for the potential for fatigue-crack
growth of the flaw during the life of the pipeline. The amount of fatigue crack growth is
calculated by integrating the Paris law expression. API 1104 Appendix A assumes failure by
brittle fracture and, therefore, does not include a plastic-collapse check on the allowable flaw
size.
The total applied strain in the API 1104 Appendix A approach is the sum of the applied strain,
as determined from a stress analysis, and the residual strain. The allowance for residual
strain is assumed to be equal to 0.2 % strain for all materials.
API 1104 Appendix A requires that the CTOD toughness be measured at the minimum
operating temperature of the pipeline. In addition, the weld procedure must be qualified to
one of two CTOD fracture-toughness values, 0.127 mm (0.005 inch) or 0.254 mm (0.010
inch) CTOD. It is necessary for the measured CTOD to meet the minimum of 0.127 mm
(0.005 inch) in order to use the Appendix A procedures.
API 1104 Appendix A further stipulates that the flaw length be less than 40 percent of the
pipe diameter for flaws less than or equal to 25 percent of the pipe wall thickness in height,
and less than four times the pipe wall thickness for flaws between 25 and 50 percent of the
pipe wall thickness in height. Flaws exceeding 50 percent of the wall thickness are not
allowed.
The Canadian Standard CSA Z662-Appendix K includes a fracture and a plastic collapse
analysis. The allowable flaw length is the lesser of the maximum allowable flaw length to
prevent brittle fracture and the maximum allowable flaw length to prevent plastic collapse.
The plastic collapse assessment used in CSA Z662 Appendix K is based on a modified
ligament instability model in which failure is assumed to occur once the plastic zone size in
the section containing the flaw extends past approximately 10 % of the pipe circumference.
To achieve a factor of safety on the assessment of flaw size a factor of 2.0 is applied on flaw
height and 2.5 on flaw length. By applying the above approach to the results of full-scale pipe
bend tests, the following linear equation for plastic collapse K was derived, Eq. (A8):
18 l h
σpc = SMYS [1,03 - ] (A8)
πDt
where l = allowable flaw length, h = flaw height, D = pipe diameter, t = wall thickness, SMYS
= Specified Minimum Yield Strength of the pipe metal and σpc is the applied stress at plastic
collapse.
In addition, CSA Z662 Appendix K stipulates that the maximum height of the flaw must be
restricted to one-half the pipe wall thickness. These flaw size restrictions are imposed as the
modified ligament instability model has only been checked experimentally against the results
of full-scale bend tests, which had flaws within these limits.
Eq. (A8) shows that plastic collapse is related to the SMYS (specified minimum yield strength)
and thus neglects the effect of Y/T ratio. The factor of 1.03 is an empirical value that reflects
the curve fit effects. The use of the SMYS avoids ambiguity in defining the flow stress while
the SMYS is a guaranteed value that is not subject to variability. This choice is partly
influenced by the fact that the prediction of the allowable flaw length, l (or L2max in terms of
Appendix K), using a flow stress of 1,03.YS rather than the actual yield strength, provides a
minimum safety factor of 1,5 on stress for the full scale test data which were used to verify
this solution.
The CSA approach specifies that the weld toughness shall be determined by CTOD testing at
the minimum design temperature. Note that, unlike the API approach, the CSA approach
does not require a minimum value of toughness (CTOD). Furthermore, it is stipulated that
two weld-metal tensile-test specimens shall be tested and that the yield strength of each
specimen shall be equal to or greater than the specified minimum yield strength of the parent
pipe material, i.e., the weld metal strength shall match/overmatch the parent material
strength.
Finally, unlike its American counterpart, API 1104 Appendix A, a fatigue analysis is not
considered necessary for the CSA approach since the CSA Z662 Appendix K approach is only
applicable to gas pipelines where the cyclic loading on girth welds in gas pipelines is
extremely low1.
1
CSA approach for the liquid-pipeline, the maximum height of a flaw is restricted to 25% percent of the pipe-wall thickness
compared to the 50 % restriction imposed in the gas pipeline standard. This additional restriction on the flaw depth for liquid
lines is a guard against fatigue-crack growth. If the operator of the liquid pipeline wishes to allow flaws greater than 25 percent
of the wall thickness in depth but less than 50 percent of the wall thickness, then they must show by analysis that the subject
flaw will not grow by fatigue.
4.3 EPRG
4.3.1 Background
Prior to the development of the European Pipeline Research Group (EPRG) ECA guidelines on
allowable flaw sizes in pipeline girth welds, the discrepancies and limitations between the
above standards were carefully assessed [43]. As a result of this study, a novel philosophy for
setting allowable fabrication flaw size limits was adopted.
EPRG offers the opportunity for an ECA analysis two levels of conservatism. The first level
(Tier 2) is a simple assessment method with a built in safety factor. The second level (Tier 3)
allows using “User Specific Flaw Acceptance Levels”. Note that the discussion of the Tier 3
approach is beyond the present discussion. Also note that Tier 1 specifies the Workmanship
flaw acceptance levels.
The EPRG approach is somewhat different than the two previously discussed approaches
because flaw assessment according to ERPG-Tier 2 is uniquely based on plastic collapse while
API 1104 and CSA Z662 do not exclude brittle fracture. The EPRG-Tier 2 guidelines require a
girth weld to meet a minimum toughness (30J minimum / 40J mean), and provided this
toughness requirement is met, allowable flaw sizes are derived from pipe yielding as limit
load.
4.3.2 EPRG-Tier 2
The basis for EPRG-Tier 2 flaw acceptance is the simple DEN plastic collapse model, Eq. (A7),
and the performance criterion described in Section 3.3.4. For pipe yielding (GSY), σpc = YS,
Eq. (A7) can be converted to the following equation:
YS (1 + R) lh
YS = [1 - ] (A9)
2R st
The maximum allowable flaw length for GSY is obtained by inverting Eq. (A9):
(1 − R ) s t
l= (A10)
1+ R h
Eq. (A10) assumes that collapse occurs if the pipe yields over a fixed arc length of “s” mm.
Curved wide plate test results have shown that conservative predictions are obtained for an
arc length, s, of 300 mm [24]. This length corresponds to approximately 10 % of the
circumference of a large diameter pipe.
Regardless the NDE method used, EPRG-Tier 2 specifies that no flaw shall have a through
thickness height in excess of 3.0 mm. That is, it is assumed that the average flaw height of
natural flaws is less than 3.0 mm.
For a YS/TS ratio (R) of 0.87 and an arc length of 300 mm, the maximum allowable length
for a fixed flaw height of 3.0 mm as a function of pipe wall thickness, t, using Eq. (A10), is:
l = 6,95t (A11)
The flaw geometry used in the above calculations is assumed to be rectangular. This
simplifying assumption, which is in line with the 'containment rectangle' approach used in
current flaw assessment, is conservative because natural flaws are irregularly shaped.
Therefore, EPRG have decided to set the allowable flaw length at 7t. This length gives a flaw
area limit of 7 % per 300 mm length of weld.
By using the maximum allowable flaw size, 7t x 3.0 mm2, it is assumed that a Charpy
toughness of 40 J average, 30 J minimum is achieved at the design temperature, and the
weld metal is matching or overmatching. Note further that EPRG-Tier 2 does not require
CTOD toughness testing since the minimum toughness for plastic collapse can be derived
from Charpy V impact tests.
The above ECA approaches are not the only codified approaches for assessing the
serviceability of flaws in girth weld. For example, the Failure Assessment Diagram (FAD)
approach, BS 7910-2000 (revised PD6491-1991), API RP579, etc…, can also be used in a
fitness-for-service assessment. Those approaches are included in this review because:
• they are general purpose approaches which assess the likelihood of fracture and
plastic overload simultaneously
• the procedures are not restricted to the specific case of girth welds.
• they have not yet been, as far as the present authors know, extensively validated for
flaws in pipeline girth welds.
However, the major concern is that the assessment for plastic collapse does not represent the
particular configuration of flawed girth welds in large diameter pipelines.
Appendix A of API 1104 (further denoted as API) and CSA Z662 Appendix K (CSA) are now, in
terms of length of weld assessed, the most widely used methods of performing fitness-for-
service assessments. The limited application of the ERPG approach is most likely due the
imposed restriction on allowable flaw height (3 mm).
Prior to their use, the primary input parameters to be considered in an ECA are: toughness of
the region (weld metal or HAZ) containing the flaw, material properties of the weld region,
applied stress, and flaw type and size. Since current ECA methodologies are deterministic
based, the input parameters are chosen on the basis of lower bound values [44-46].
The last but not the least point of note is that the dimensions of flaw (length and height),
and its through-wall position must be established by an appropriate non-destructive-
inspection technique before an ECA can be used. The use of conventional radiography is
adequate for measurement of flaw length but is insufficient for determining flaw height. The
use of ultrasonics for determining flaw height is acceptable provided the accuracy of the
inspection procedure has been previously established and any potential inaccuracy is included
in the analysis.
As discussed, API, CSA and EPRG-Tier 2 (further denoted as EPRG) provide guidelines for
alternative flaw acceptance criteria for pipeline girth welds, together with requirements of
greater stringency for welding. These girth weld specific ECA’s cannot directly be compared
to their qualities. Each of these approaches use different failure models and require different
input requirements. For discussion purposes, the most important provisions and limitations
are compared in Table A1.
In examining Table A1, one can easily conclude that each code will generate different flaw
size limits depending on the incidental combination of pipe diameter, wall thickness, tough-
ness and applied stress. The striking differences are briefly outlined hereinafter. Section 5.2.
illustrates the differences by means of by a sample application in which the allowable flaw
sizes are determined using the procedures defined in each of the standards. Table A1
illustrates that:
• The approaches use different applied stresses/strains inputs. EPRG is based on remote
yielding at a fixed applied strain of 0.5 %. API and CSA require the user to determine the
applied strain (stress). The calculated applied strain is usually elastic (< 0.5 %).
• API includes a residual stress correction that is already embodied in the assessment
curve. Both CSA and EPRG ignore the effect of residual stress in the analysis. The neglect
of residual stresses is consistent with the experimental findings that residual stresses do
not affect ductile failure.
• The CSA and EPRG approaches, because of their derivation, provide a factor of large
safety. The authors of CSA introduced a margin of safety on flaw size. The conservatism
of EPRG is based on the GSY performance criterion.
• The failure criterion of API does not require a plastic collapse assessment. This limitation
is not, as illustrated in Section 2.5, very logical because the API toughness requirements
(0.127 mm or 0.254 mm) ensure plastic collapse while the failure model assumes “brittle
fracture”. A related point of note is that API ignores the imposed limitation incorporated
in the application of the CTOD design curve [47]. Thus, by ignoring a plastic collapse
FAILURE CRITERION
FRACTURE ASSESSMENT
TOUGHNESS TESTING
- CVN testing No 40 J 40 J / 30 J
TOUGHNESS REQUIREMENTS
EXPERIMENTAL VALIDATION
Can be
“RATING” Very safe Conservative
unconservative
assessment, API could be non-conservative if applied to the case of tough welds. In such
a case, the resistance of a flawed weld to failure is not related to toughness, but instead,
depends solely on the flow stress of the “weakest” material (see Section 3.3.3).
• Each of the ECA approaches is applicable for applied strain up to the onset remote
yielding (GSY). Although strain hardening (or the Y/T ratio) plays a crucial role in
obtaining remote yielding, neither API nor CSA address this effect. EPRG solves this issue
by using a strain hardening dependent flow stress.
• Both API and CSA require that the welding procedure must be qualified for CTOD
toughness. API specifies a minimum toughness of 0.127 mm (the toughness level of
0.254 mm generates larger allowable flaw size). CSA does not establish a minimum CTOD
level. However, the CSA Charpy requirement of 40 J implies that an effective minimum
CTOD must exist. EPRG does require CTOD testing. Toughness testing for EPRG (Tier 2)
is confined to Charpy testing.
• API assumes that the weld is “matching” if the tensile strength of the composite, as
measured in the cross (transverse weld) tensile test, meets the minimum specified tensile
strength. CSA stipulates that the yield strength of the cross-tensile specimen (without the
reinforcement in place) is at least equal to the SMYS of the pipe. EPRG requires
information on the weld metal yield strength. In addition, ERPG requires matching
(overmatching) weld metal is demonstrated by all-weld metal tests.
• API requires the user to establish the accuracy of the NDE technique to be used and thus
to apply a safety factor to the flaw height to account for potential flaw sizing
inaccuracies.
• API and CSA give maximum flaw lengths proportional to pipe diameter. The maximum
allowable flaw length of EPRG is proportional to wall thickness
• Flaws with a height greater than one half the wall thickness are not allowed for API and
CSA. On the other hand, flaw height in EPRG is restricted to 3.0 mm. The height
limitation specified by EPRG is a deterrent to its use if the user has to account for
potential flaw sizing errors. These errors are indirectly accounted for in the CSA approach
since the allowable flaw size curve embodies safety factors on length and height.
• The reliability of the CSA and EPRG assessment procedures has been established by
extensive series of large-scale tests, and for CSA, by more than 10 years of ‘field’
experience.
The above observations demonstrate that an ECA based flaw assessment can only be applied
if the “relevant” input parameters are well defined. Furthermore, it should be emphasized
that the welding procedures(s) qualified should ensure that good workmanship could be
maintained.
Another potential limitation common to any of the ECA approaches is that, if the production
welders fail to strictly apply the welding procedures previously qualified, the end result could
be production welds that have lower toughness values than qualified. This observation
emphasizes the fact that setting ‘high’ minimum toughness levels should be avoided.
The other concern is associated with the requirement that the weld-metal yield strength
matches (overmatches) the pipe metal strength. In practical weldments the tensile
characteristics of the weld deposit differ from those of the base metal. The level of weld-
metal strength matching is a significant factor for flaw behaviour. If the yield strength of the
weld zone does not exceed the yield strength of the pipe material, then the applied strain
might be concentrated in the weld region. This important issue is discussed in Section 6.
As a further aid in showing the differences between the ECA approaches discussed above,
graphical comparisons are presented for a series of sample cases, Fig. A7. Each plots shows
the allowable flaw length(s) versus the allowable flaw length(s) for a typical range of pipe
diameter and wall thickness combinations. The comparisons are restricted to 0,5% applied
strain and relate to the three approaches as applicable. Consequently, the comparisons are
focussed on failure by plastic collapse as it is implicitly assumed that the weld metal
toughness is adequate to ensure collapse by GSY.
Allowable flaw size curves have been generated for the cases of 40” diameter x 20 mm wall
and 20” diameter x 12.7 mm (and x 7.5 mm) wall X80 pipe. The resulting plots presented in
Fig. A7 show that the predicted maximum allowable flaw sizes are not consistent. Note also
that, both CSA and EPRG include a ‘default’ limitation on flaw length. It should also be
recalled that the API and CSA predictions are independent of the Y/T ratio. On the other
hand, EPRG limits the Y/T ratio, that is, the EPRG allowable flaw size cannot be applied for
Y/T ratios exceeding 0.90. CSA and API do not stipulate such a restriction.
API gives the largest allowable flaw sizes because the CTOD design, used to derive the API
curves, does not account for the plastic collapse and allows that high CTOD values give
proportionally large allowable flaw sizes. In order words, without additional plastic collapse
constraints, API may predict non-conservative estimates of allowable flaw sizes. CWP tests
have recently demonstrated this fact [18].
Both CSA and EPRG give conservative predictions for each of the cases considered here. The
smallest allowable flaw sizes are predicted by the CSA approach. The difference between API
and CSA is very significant for “thick” wall pipe while the difference becomes smaller for
“thin” wall pipe. For 7.5 mm wall pipe, API becomes restrictive on allowable length for flaw
heights exceeding the average height (3 mm) of a weld bead. In contrast, the limitation on
flaw length for EPRG is proportional to wall thickness. Of further note is the fact that CSA is
the least sensitive to the effect of wall thickness. Another way of viewing the comparisons is
that the CSA approach, as applicable, cannot be used for applied strains of yield magnitude.
In particular, it cannot be excluded that the maximum flaw size allowed by CSA could be less
than that allowed by workmanship standards.
The maximum flaw sizes preventing plastic collapse using the CSA approach are derived from
the collapse model given by Eq. (A8). As discussed, Eq. (A8) incorporates a safety factor on
flaw length of 2.5 and a safety factor on flaw depth of 2.0. Stripping Eq. (A8) of these safety
factors gives Eq. (A12):
l h
18
σ pc = SMYS [1,03 - 2.5 2 ]
πD t
lh
σ pc = 1.03 SMYS [1 - ] (A12)
0.847 D t
where l = maximum flaw length, h = maximum flaw height, R = Y/T ratio, D = pipe diameter
(D = 2πr, r = pipe radius), t = wall thickness, SMYS = (Specified Minimum) Yield Strength.
By comparing Eq (A8) - CSA approach with safety factor - to Eq. (A12) - CSA approach
without safety factors - for an applied strain of yield magnitude, it can be appreciated from
Fig. A8a that the safety margins on the allowable flaw sizes (CSA Z662 Appendix K approach)
are quite large for the case of high stress.
Fig. A8b compares the allowable flaw size curves for collapse by GSY using the DEN model,
Eq (A10), and the CSA approach without safety factors (Eq. (A12)). The DEN predictions are
shown for two Y/T ratios. Note again that the comparisons are for an applied stress of yield
magnitude (failure criterion: GSY). Examination of Fig. A8b shows that the differences
between the CSA and the DEN predictions become smaller if one removes the safety factors
contained in CSA Z662 Appendix K approach. Depending on the Y/T ratio, the DEN collapse
criterion generates either less restrictive (Y/T = 0.80) or more conservative (Y/T = 0.90) flaw
sizes than those generated by CSA without safety factors. In this connection, it should be
recalled that in deriving the EPRG allowable flaw sizes the Y/T ratio was set at 0.90.
Summarised, accepting the criticism of repetition, it must be stated once more that, the
mathematical models used to derive the CSA Z662 Appendix K and EPRG allowable flaw sizes,
have been ‘calibrated’ by using results of large-scale tests. However, it must also be
emphasized that the large majority of these large-scale tests were performed on large
diameter pipeline. This means that there is no direct comparison possible for small diameter
pipelines. The question whether of similar comparisons for ‘small’ diameter pipe will yield
similar conclusions is clearly of great importance and it is hoped that further research will
enable this to be clarified.
Y/T independent
9
Flaw height (mm)
6
API 1104
3
EPRG
CSA Z662
0
0 100 200 300 400 500
6
Y/T independent
Flaw height (mm)
3
API 1104
CSA Z662
EPRG
0
0 100 200 300
6
Y/T independent
Flaw height (mm)
CSA Z662
0
0 100 200 300
Fig. A7- Comparison allowable flaw sizes for three ECA approaches
5.3 Discussion
5.3.1 Performance criterion
The preceding comparisons have been focussed on the ‘base line’ case of GSY or pipe metal
yielding. The comparisons have to be placed in another perspective for axial the stresses
below yield. Using the DEN model, Fig. A8 shows the effect of applied stress on allowable
flaw size for GSY and an elastically strained girth weld. An axial stress of 0.80% SMYS
exceeds the actual stress. The actual stress is generally less than half the hoop stress. As
indicated, elastically stressed girth welds are assumed to fail by NSY.
9
Flaw height (mm)
0
0 100 200 300 400 500
9
Flaw height (mm)
DEN-MODEL
Y/T = 0,80
0
0 100 200 300 400 500
One will note from that Fig. A9 that the GSY failure criterion (applied remote axial stress =
SMYS) includes a substantial margin of safety when compared to the case of an elastically
stressed girth weld. Furthermore, it is worthwhile noting here the flaw length limit (50 mm)
allowed by workmanship acceptance criteria ensures failure by GSY whereas the flaw-size
limit for GSY is smaller than for NSY. Fig. A9 can also be used to explain why the incidence of
girth weld failure is very low.
Despite the observation that the GSY criterion be restrictive in terms of allowable flaw size,
the use of the GSY failure criterion is recommended. The GSY requirement eliminates a
number of possible uncertainties, namely:
6
Applied stress = SMYS
NSY
2
GSY
Workmanship
limit
(max. 50 mm)
0
0 100 200 300 400 500
Fig. A9 - Effect of the applied stress on the calculated flaw size limit
• the GSY failure criterion eliminates the need for detailed analyses to determine the
applied stress in the axial direction.
• GSY is a conservative and safe assumption for conventional (elastic) pipeline designs
because the criterion incorporates a quantifiable safety margin for conventional
designs while a degree of safety is also provided for stresses arising from sources
other than those used in design.
• possible flaw size measurement inaccuracies become less critical with GSY. In
contrast, for NSY flaw sizing might become a point of discussion since the safety
margin, or the ratio between the predicted and the critical flaw size, can be small. In
passing, it may also be pointed out that any deviation of the qualified properties
might reduce the assumed safety margin.
The collapse solution used by CSA embodies the effect of pipe diameter and was essentially
developed for failure by NSY. Moreover, the plastic collapse analysis in Appendix K is very
restrictive at stresses of yield magnitude. However, the CSA approach avoids the flaw sizing
uncertainties by using a safety factor on flaw height and flaw length in the collapse equation
while the flow stress only depends on the pipe metal yield strength. Except for very high
values of Y/T ratios, the value of the CSA flow stress is also smaller than that used by EPRG.
These differences generate different allowable flaw sizes, but the sample applications shown
in Fig. A8 have illustrated that CSA and EPRG provide a comparable margin of safety for large
diameter pipelines.
The conservatism incorporated into the original CSA collapse solution (Eq. (A8)) can be
reduced by replacing the ‘virtual’ value of the CSA flow stress, 1.03 SMYS, by a Y/T depended
flow stress and by eliminating the factor of safety on flaw length. For example, substituting
Eq. (A6) into Eq. (A8), the modified CSA collapse solution (further denoted MLM-collapse
criterion) is given by Eq. (A13):
18 l h
σpc = SMYS [1.03 - ] (A8)
πDt
SMYS(1 + R) lh
σ pc = [1 - ] (A13)
2R 0.169 D t
In Figs. A10 and A11, the original CSA and EPRG maximum allowable flaw size curves are
compared to the DEN (Eq. (A9) and the MLM collapse model with and without (Eq. (A13) the
safety factor of 2.5 on length flaw. The comparisons are made for a representative set of
pipe diameter, wall thickness and Y/T ratio combinations. All plots have been generated for
failure by GSY. The comparisons of the various curves lead to the following observations:
• the relative position of the curves depends on the Y/T ratio, pipe diameter and wall
thickness.
• as Figs. 10 and A11 indicate, the MLM model with safety factor on flaw size (curve
decorated with open circles) gives for the low Y/T ratio case (Y/T = 0,80) a substantial
increase in flaw size. For thin wall and high Y/T ratio (Y/T = 0.90) pipe, the increase in
flaw size is less spectacular.
• for small diameter pipe, Fig. A11, the differences between the MLM and the DEN-
predictions are due to the difference in the treatment of the extent of yielding. The
DEN assessment gives the least conservative predictions.
• the comparison of the relative positions of the flaw curve predicted by the MLM model
without the safety factor on flaw length (curve decorated with solid circles) and the
DEN analyses shows that the DEN model gives conservative results for large diameter
pipe. For small diameter pipe, the maximum flaw sizes derived form the MLM model
without a length correction are more restrictive than the DEN prediction. This is not
surprising since the MLM model is based on the assumption that the extent of yielding
at plastic failure is proportional to the pipe diameter. The DEN model uses a fixed arc
length of 300 mm irrespective the pipe diameter. Except for the case of large diameter
pipe, this assumption could be non-conservative for small diameter pipe.
CSA
3 EPRG
0
0 50 100 150 200
DEN-model
EPRG - Tier 2
9 MLM
MLM with length correction
Flaw height (mm)
3 EPRG
0
0 50 100 150 200
Fig. A10 - Comparison allowable flaw sizes and collapse models for large diameter pipe.
Summarised, the above observations would justify the use of the MLM collapse model without
length correction, Eq. (A13), as it offers improvements over the existing EPRG and CSA
collapse assessment procedures. Although a number of practical aspects must be resolved,
Eq. (A13) could also be used to reduce the safety margin embodied in CSA Z662 Appendix K
approach and to formulate an EPRG solution for predicting the collapse conditions of small
diameter pipe. However, the latter contention needs to be validated since appropriate large-
scale test data is not yet available.
9 DEN-model
EPRG - Tier 2
MLM
MLM with length correction
Flaw height (mm)
3
EPRG
CSA
0
0 50 100 150 200
9 DEN-model
EPRG - Tier 2
MLM
MLM with length correction
Flaw height (mm)
CSA
CSA Z662 Appendix K
6
3 EPRG
0
0 100 200 300
Fig. A11 - Comparison allowable flaw sizes and collapse models for small diameter pipe.
6 GENERAL REQUIREMENTS
Prior to the use the ERPG-Tier 2 based girth weld flaw assessment must be met. A stress
analysis is required to determine the maximum applied tensile stress acting on the girth weld.
However, since the GSY criterion is the performance requirement one has only to verify that
the design stress in the axial direction does not exceed the pipe metal yield strength. The
Weld qualification testing includes tensile testing to determine the yield strength of both pipe
and weld metal (see Section 6.1), and toughness testing of the weldment to determine the
minimum fracture toughness of the weld region. The Charpy V notch toughness tests are to
be determined at the minimum design temperature. A related, but very important
consideration in applying an ECA is the control of the welding procedure. The essential
welding variables have to be carefully monitored with the intention of minimizing the
variation in mechanical properties of the welds. Thus, the weld filler metals and weld
procedures must be carefully selected in order to produce girth welds with an optimum
balance of strength and toughness.
A NDT method must be used to be capable of determining the maximum flaw height in order
to determine the maximum allowable length of the detected flaw. However, EPRG assumes
that the height of individual flaws has a height equal to the maximum depth of the
corresponding weld pass. Thus, as presently developed, EPRG cannot be applied to flaw
heights exceeding 3.0 mm [48] (see Part B if this report).
It is undesirable for any region of the weld to contain low strength material. Low strength or
undermatched welds are prone to concentrate applied deformations [49-52]. Therefore, weld
metal testing is required to verify that the actual yield strength of the weld is at least equal to
the specified minimum yield strength of the pipe [53-55].
Specifications assume that the weld metal has the same properties as the pipe metal. This
assumption is not correct [56]. On the other hand, current plastic collapse assessments do
not specifically address the effect of weld metal undermatching or overmatching on flaw
tolerance. A standard ECA approach is focussed on flaws, which are assumed to be located in
a matched (or overmatched) weld. This assumption is rarely a problem for the common lower
grades of pipeline steels. However, in the event that the axial strains of yield magnitude
occur, this assumption might lead to unsafe predictions since undermatched welds tend to
localise plastic strains within the weld zone so that failure might occur at strain levels lower
predicted strain.
If the weld metal yield strength exceeds the applied stress, elastic designs do not exclude the
use of undermatching weld. In this case, however, the user should be aware of the
consequences since undermatched welds have a lower flaw tolerance than overmatched
welds, Fig. A12. The ‘lower’ flaw tolerance of undermatched welds implies that greater
attention needs to be paid to flaw sizing. This implies that undermatched welds require
precise, high accuracy NDT inspection. The issue of flaw sizing is discussed in the next
Section.
Conversely, overmatching protects weld flaws from large strains, as plasticity will occur in the
pipe material. The shielding effect provided by an overmatched weld allows, as compared to
undermatched weld, larger flaws for similar levels of applied remote strain. This effect applies
especially for elastic designs. Equally, overmatched welds permit less rigorous flaw sizing
requirements. For these reasons, it is attractive to require weld metal yield strength
overmatching2. On the other hand, in order to prevent failure at strain levels below yield (<
0,5 %), moderately undermatched welds exposed to plastic straining require a high strain
hardening capability (or a low YS/TS ratio).
Wall thickness = 20 mm
Y/T = 0.85
9
Undermatching
'Matching'
Flaw height = 3 mm
6
Flaw height (mm)
0
0 100 200 300
Flaw length (mm)
Fig. A12 – Effect of weld metal yield strength mismatch on allowable flaw size (schematic)
Even when an overmatched weld metal is selected it is still possible that some girth welds are
undermatched [58-61]. This is because the distributions representing the scatter in tensile
properties of the base and weld metal can overlap, Fig. 13.
Fig. A13 shows that it is easier to obtain overmatching in mechanised GMAW welds while the
spread of the yield strength properties is narrower than for SMAW welds. Furthermore, Fig.
A13 illustrates that a variable degree of pipe to weld mismatch may be observed in SMAW
girth welds. Moreover, for pipe grades X70 and above, the use of cellulosic electrodes does
not automatically exclude weld metal yield strength undermatching.
Since girth welds in high strength pipelines may unknowingly be undermatched because the
pipe material has much higher yield strengths than the SMYS, specific guidelines on achieving
matched/overmatched girth welds are needed.
2
Overmatching can be ensured when the minimum weld metal yield strength exceeds the specified minimum yield strength
of the pipe by 60 MPa (SMYS + 60 MPa) [57] and by a proper control of the welding parameters.
0.03
GMAW
CEN X70 PIPE METAL SMAW (85 tests - 4 projects)
Ave: 545 MPa (127 tests - 15 projects) Ave: 635 MPa
Max/m in. = 480-600 Ave: 562 MPa Max/m in. = 580-680
Max/m in. = 474 - 638
0.02
0.01
UM
0
470 520 570 620 670
Yield strength (MPa)
Fig. A13 - Effect of SMAW and GMAW weld metal variation on weld metal mismatch
(Hatched area represents the undermatching density)
Of most interest, when considering mismatch of girth welds, are the properties in the
longitudinal (pipe axis) direction. The tensile properties in the axial direction were and are not
often specified. Therefore, historical data is not commonly available from pipe manufacturers.
Correlation to hoop direction properties is difficult due to the varying effects described above,
and because it depends on other factors such as pipe diameter, whether the pipe is
longitudinal or spiral formed, cold expansion, test specimen geometry (round bar vs. full
section bar), etc. It is therefore advisable to either specify, or independently measure, the
pipe tensile properties in the axial direction, if an accurate measure of weld mismatch is to be
determined.
The range of yield strength, in any particular order of pipe, can be as high as 100 MPa. This
range can increase for thinner wall pipe. A further complication is that Y/T ratio tends to
increase as plate thickness decreases []. Moreover, as common large diameter pipe grades
are now higher, pipe manufacturers tend to produce similar steel for many pipe grades. The
result for “low grade’ pipe is that the range of yield strengths might exceed the standard
strength distribution.
Weld metal strength is affected by process factors such as heat input, travel speed, and
welding current [62-65]. These factors will vary from weld to weld due to differences in
technique between welders. As well, the width of the weld affects the amount of dilution
from the parent metal; the properties will therefore vary in the through thickness direction,
and according to the o'clock position around the pipe [59].
Studies have shown that the lowest weld metal strength will be measured at the bottom of
the pipe (6 o'clock). The levels of yield strength mismatch are normally quite variable in
manual SMAW girth welds. Mechanized GMAW girth welds have an advantage over manual
welds because they are more repeatable. The spread of properties is narrower than for
manual welds [60].
Weld metal yield strength testing is usually done on 6.4 mm diameter all-weld metal
specimens. This provides only a sample from near the middle of the weld, and may not be
representative of the strength of the entire weld. A rough estimate of weld metal strength
can be obtained from micro hardness measurements [7] and is particularly useful if a lower
strength electrode is used for the root pass. Another approach consists of comparing the
results of edge notched pipe metal and edge notched transverse weld tensile specimens [8].
Flaw acceptance based on an ECA analysis requires information on both flaw length and
height of the detected flaw, and on its location (depth) within the weld. In other words, it is
essential to consider NDT as an integral part of an ECA.
Since the actual dimensions and the through-thickness position of a girth weld flaw (surface
breaking vs. embedded) affect its significance, it is necessary to consider how flaw sizing,
recategorization of an embedded flaw and flaw interaction can be dealt with (Fig. A14).
Since girth weld flaws might have complex geometries, there are three areas where the
designer should have answers:
such measurements. Radiography can be used to measure flaw length. However, radiography
is of little use in assessing flaw height from conventional radiographs under field conditions.
DETERMINE
Surface breaking INSPECTION
ACCURACY
Embedded Type ?
FLAW
Height
Size/dimensions ? Characterisation
Length
Interaction
Near surface (Multiple flaws)
breaking flaw
Recategorisation
Depth (near surface
breaking flaws)
The hurdle that stands in an ECA is the quest for certainty in UT flaw sizing accuracy. This
accuracy of flaw sizing cannot always be quantified. It is not possible to say with complete
certainty that a flaw will be larger or smaller than its estimated (measured) size. That is, a
‘unique’ value for flaw height (or depth) sizing inaccuracy cannot be given as this value
depends on the welding process, flaw shape and flaw through-wall position, and ultimately on
the calculated/predicted flaw size limit. Also, the sizing error might have a larger effect on the
accuracy of the predictions than the choice of the collapse model.
The required precision may not be possible or necessary [70]. This observation is not meant
to place a disproportionate emphasis on flaw height-sizing accuracy because, there are other
factors influencing the required accuracy. For example, the level of weld yield strength
overmatch can offset the error in flaw sizing.
Significant errors in the prediction of the flaw size limits can be made if the flaw size is
underestimated, in this case the conservatism of the ECA analysis is eroded and the margin
of safety is reduced. An overestimate of the flaw size will increase the degree of conservatism
and may result in an unnecessary repair of flaws that would have been innocuous in reality.
Moreover, an ECA analysis based on conservative assumptions can predict allowable flaw
sizes smaller than the workmanship acceptance levels. For example, this problem might occur
for girth weld flaws in thin wall pipes.
The user must be address these issues and safety factors should be applied to account for
potential sizing inaccuracies. As it is potentially unsafe to assume that a flaw has the
dimensions predicted by UT, the inaccuracy in height sizing has to be accounted for by
adding a correction factor equal to the inspection error, ∆h, to the calculated flaw
dimensions. The direct implication is that the dimensions of the allowable flaw should be
smaller than its calculated limit dimensions.
The API approach requires such an approach in that it requires the user to establish in
advance the accuracy of the NDE technique to be used, and to apply a safety factor to the
flaw height to account for potential inaccuracies. For CSA, the uncertainty in flaw height
sizing is directly accounted for in the applied collapse model.
6.2.2 Illustration
The effects of a flaw height correction on the calculated flaw size are shown in Fig. A15. The
calculated flaw length-height curves (see thick solid lines) apply to surface breaking flaws in
20.0 mm thick pipes (Eq. (A10)). The calculations have been performed for two Y/T ratio
values (0.80 and 0.90). These curves delineate the boundary between GSY (flaw length-
height combinations below the curve) and NSY (flaw dimensions above the curve) failure
behaviour.
Furthermore, this simple assessment illustrates that the sizing error is minor problem for low
Y/T ratio pipeline steels because the condition for plastic collapse of an overmatched girth
weld depends on the flow stress of the weakest material (see Section 3.3.3 – Fig. A6)
The sizing problem can also be approached from another perspective. Historical data and
experimental evidence available demonstrates that very large planar flaws/cold cracks in
older (X-ray inspected) pipelines have been missed and thus have unwillingly been accepted.
Even so, with the use of the best NDT techniques, or a combination thereof, some ‘large’
flaws will be missed. Judged from these observations and the evidence of long service
without failure, one can use the rigorous concept of probability or incorporate some elements
of the reliability theory in an ECA analysis. However, a serious warning should be given
against an indiscriminate use of such methods. Currently, few data exist in sufficient quantity
to permit a statistical analysis of the many factors (wall thickness, flaw location, flaw type,
flaw profile, welding process, weld bevel preparation, inspection procedure, etc..) affecting
flaw-sizing accuracy.
The flaw interaction and flaw recategorisation criteria currently being used are very, if not
extremely, conservative and do certainly not reflect the fact that flawed girth welds, made
following standard procedures, fail by plastic collapse. Flaw characterisation rules are based
on elastic fracture behaviour. Beyond this limitation, the current interaction and
recategorisation procedures do not take account of the favourable effects of weld metal yield
strength overmatching and weld reinforcement on weld performance.
DEN-MODEL
Y/T = 0,80
h - 2 mm
3 EPRG
Accepatble
flaw sizes
0
0 100 200 300
6
DEN-MODEL
Y/T = 0,90
h - 1 mm
3 EPRG
h - 2 mm
Accepatble
flaw sizes
0
0 100 200 300
Fig. A15 – Sample calculations illustrating the effects of Y/T ratio and flaw sizing error on
allowable flaw height (performance requirement: GSY).
Flaw interaction - Existing interaction rules for multiple flaws are highly conservative for
girth welds subjected to plastic strains [71]. An associated issue is that UT inspection might
find a larger number of planar ’neighbouring’ flaws. On the other hand, since UT is ‘subject’
to inaccuracies in flaw sizing, the application of current interaction rules may cause higher
rejection and repair rates. Note that the same comment applies for ‘workmanship based’ flaw
accumulation criteria.
‘harmful’ surface-breaking flaw. In other words, current practices lead to a significant number
of unnecessary weld repairs, considering the fact that tearing only occurs after extensive
yielding.
As current flaw characterisation practices lead to a significant number of weld repairs that
are, considering the ad-hoc testing experience, not needed, it is recommended to investigate
these problems.
It should be underlined that a probabilistic approach is only one amongst several solutions to
established reliable allowable flaw sizes. For example, the interaction between the
uncertainties related to flaw height sizing and acceptable flaw size can directly be simulated
by Curved Wide Plate (CWP) testing. This is a very attractive option for the study of girth
welds in pipelines for which little previous experience exists. By testing CWP specimens
containing a surface breaking flaw with a height equal to its assumed value (normally 3 mm)
plus a 2,0 mm allowance for error in height sizing, the actual failure characteristics can be
determined. Although the option of CWP testing has a number of advantages, it must be
emphasized that CWP testing is only an intermediate step in the development of a generally
accepted ECA solution consistent with the capabilities and limitations of the inspection
technique.
7 RECOMMENDATIONS
Throughout this review, reference has been made to various aspects of an ECA that are not
completely understood. However, the current state of technology is sufficiently advanced to
make a major step in the development of reliable allowable sizes for flawed girth welds. This
possibility might be of concern or beneficial depending on the viewpoint taken.
Emphasis for the foreseeable future should be placed on steady improvements in the ECA
approaches by introducing issues not currently addressed. Some of the more important issues
are listed below. The non-exhaustive list of recommendations is presented with the
expectation that others will have opinions to offer.
SSpecify materials, welding processes and procedures ensuring ductile failure behaviour.
SAvoid materials susceptible to scatter in properties. In particular, limit the upper supply
range of pipe yield strength in the axial direction.
SEnsure weld metal yield strength match/overmatch, if possible (the welding process used
might be the limiting factor).
SEstablish correlations between specification requirements and the actual toughness and
tensile properties. Lack of reliable material data hampers the efficient use of reliability
models to optimise inspection and design.
SUse an experimentally proven ECA methodology specific to the application in mind. Explore
the material properties available, if needed.
SDetermine and document the effects of the input parameters and their relative effects on
the calculated defect size limit(s). Incorporate this information into the material and
welding specifications.
STeach engineers how to avoid potentially unsafe ECA assumptions which actually may
reduce girth-weld integrity. This recommendation reflects the fact that many users are not
familiar with all details involved in an ECA analysis.
SFor critical applications, plastic designs and for situations involving high-strength pipe
establish the critical defect size limits by type (CWP or full-sale bend) testing.
SRevise the testing requirements. In particular, design and specify material testing
requirements that provide data relevant to the ERPG-Tier 2 approach. This action must be
regarded as a prime need in developing a cost-effective ECA methodology.
SDetermine the weld metal's yield strength properties and their variability.
SReduce human and procedural errors by applying simple and transparent testing and
inspection procedures.
SDetermine what defect size limit(s) the inspection should be designed to find.
STailor the inspection procedures to the location of the defect (surface-breaking vs.
embedded).
SCollect appropriate weld and pipe metal yield and tensile strength distributions to:
- Determine the level of weld-metal vs. pipe-metal yield strength mismatch as a
function of pipe grade and welding process/procedure
- Optimise the reliability models for future use in defect assessments of ‘old’ pipelines
SDevelop a multi-level ECA methodology that connects both the pipe-to-weld metal
mismatch effects and the inspection capabilities. In particular, create "tailor-made"
tolerable defect length-height curves as a function of the performance requirement
(stress-based vs. strain-based), weld properties, defect type, well thickness and inspection
capabilities.
SDetermine the arc length for plastic collapse of small diameter pipelines
SFor (very) high strength pipes, quantify the interaction between yield strength, Y/T ratio
and uniform elongation of the pipe metal because the quantification of the deformation
and failure behaviour of a high strength steel pipe in terms of Y/T ratio alone may be an
over simplification.
SDevelop workmanship requirements for AUT inspected girth welds and strain-based
designs. These requirements should be based on simple and transparent fitness-for-
purpose principles.
SDevelop a simple and representative small-scale test method that simulates the constraint
conditions occurring at defects in pipeline girth welds. If possible, the test should allow
one to determine the extent of crack-tip ductile tearing in terms of applied load and yield
strength/tensile strength ratio.
SExperimentally verify the limit levels of weld-metal yield strength undermatch and
overmatch for stress-based and strain-based designs.
SEstablish a database that provides detailed information on the type and the shape of weld
defects occurring in field welds. This information would allow engineers to:
- Place the relation between defect profile and defect sizing accuracy in a proper
perspective
- Select the most relevant and economic non-destructive testing technique
SDevelop non-destructive testing requirements that take into account the service
requirements (stress-based or strain-based design), defect location (surface breaking vs.
embedded) and material toughness properties (undermatching vs. overmatching weld
metal strength to pipe strength).
REFERENCES
1. BSI-PD 6493-1991, "Guidance on Some Methods for the Derivation of Acceptance Levels for
Defects in Fusion Welded Joints," British Standards Institution, London, 1991.
2. BS 7910-2000 Guide on methods for assessing the acceptability of flaws in metallic structures,
London, BSI, 2000.
3. API Standard 1104, "Welding of Pipelines and Related Facilities," 17th Edition, September 1999.
4. Canadian Standards Association, CSA-Z662-99, "Oil and Gas Pipeline Systems," Ontario,
December 1996.
5. Australian Standard, AS 2885.2-1995, "Pipelines - Gas and Liquid Petroleum, Part 2: Welding," 1995.
7. CEN 12732-2000, “Gas Supply Systems – Welding Steel Pipework – Functional Requirements”
8. Knauf, G. and Hopkins, P., "The EPRG Guidelines on the Assessment of Defects in Transmission
Pipeline Girth Welds," 3 R International, 35 Jahrgang, Heft 10-11/1996, S. 620-624.
9. SINTAP: Structural Assessment Procedures for European Industry. Final Report, July1999,
Contract No BRPR-CT95-0024, Project No. BE95-1426, Report No. BE95-1426/FR/7.
10. Hopkins, P., Demofonti, G., Knauf. G., and Denys, R., “An Experimental Appraisal of the
Significance of Defects in Pipeline Girth Welds”, Paper 20, Proc. EPRG/NG18 Eight Biennal Joint
Technical Meeting on Line Pipe Research, Paris, May 14 -17, 1991
11. Denys, R.M., "Fracture Behaviour of Large Diameter Pipeline Girth Welds: Effect of Weld Metal
Yield Strength in 25 mm Thick Pipe", AGA Catalog No 202-011, Vol. I, Febr, 9-94
12. Denys, R.M., "Effect of Weld Metal Matching on Girth Weld Performance, Vol. II," AGA Catalog
No. L51 851-IN1, p. 137, January 1993.
13. Denys, R., 'Fracture Behaviour of Large Diameter Pipeline Girth Welds, Effect of Weld Metal Yield
Strength and Defect Interaction', AGA Report, PRC-Line Pipe Welding Supervisory Committee,
Contract Number PR 202-011, May, 24, 1994
14. Denys, R.M., Lefevre, A.A., Glover, A..A. and Horseley, D.J., "An Approach to Defect Acceptance
Criteria for GMAW Girth Welds in High Pressure Pipelines," Conf. Proceedings PRCI 9th
Symposium on Pipeline Research, Houston, Texas, 1996,
15. Denys, R.M., Lefevre, A.A., "PRCI report PR 202-9328: Alternative Acceptance Criteria of Girth
Weld Defects in Cross Country Pipelines," American Gas Association, Arlington, Virginia, June
1996.
16. Demofonti G, Fersini M, Knauf G and Schipaanboord W: Defects in girth welds. New test results,
NDT experience, EPRG Guidelines. PRC/EPRG Eleventh Biennial Joint Technical Meeting,
Arlington, Virginia, April 7-11, 1997. Paper 17.
17. Schipaanboord, W., Denys, R., Lefevre, A., and Roovers., P “Validation tests to Support the EPRG
– Tier 2 Guidelines for Misaligned gGrth Welds in Large Diameter Onshore Pipes through Full
Scale Bend and Wide Plate Tensile Testing”, Conference Proceedings 12th Biennal Joint Technical
Meeting PRC/EPRG on Pipe line Research, Groningen, Mei 17-21, 1999
18. Denys RM, Lefevre AA, De Jaeger C and Claessens S: Failure characteristics and defect tolerance
levels of girth welds in large diameter X65/X70 steel pipelines: experimental verification through
wide plate testing and comparison with ECA prediction models. In: Denys RM, editor. Pipeline
Technology Proceedings of the third international pipeline technology conference Volume I,
Brugge, Belgium, May 21-24 2000. Amsterdam: Elsevier Science, 2000, pp.151-67.
19. Kastner, W., Roehrich E, Schmitt W and Steinbuch R, “Critical Crack Sizes in Ductile Piping”,
International Journal of Pressure Vessels and Piping 9 (1981), pp 197-219.
20. Willoughby, A.A., “A Survey of Plastic Collapse Solutions Used In Failure Assessment”, Welding
Institute Report 191/1982, 1982.
21. Milne I, Ainsworth RA, Dowling AR and Stewart AT: Assessment of the Integrity Of Structures
Containing Defects”,CEGB Report R/H/R6 Rev 3, 1986.
22. Miller, A. G., "Review of Limit Codes of Structure Containing Defects," International Journal of
Pressure Vessels and Piping, Vol. 32, 1988, pp.191-327.
23. Wang Y-Y, Wilkowski GM and Horsley DJ: Plastic collapse analysis of pipelines containing surface
breaking circumferential defects, Denys R, editor. Pipeline Technology Proceedings of the third
international pipeline technology conference, Volume I, Brugge, Belgium, May 21-24 2000.
Amsterdam: Elsevier Science, 2000, pp.191-209.
24. Denys, R., A Plastic Collapse Based Procedure for Girth Weld Defect Acceptance, Proc. Int. Conf.
on Pipeline Reliability, Canmet, Pergamon, Calgary Canada, Pergamon Press, 1992, Vol. II, paper
VIII.1-VIII.11
25. Coote, R. I., Glover, A.G., Pick, R.J. and Burns, D.J., “Alternative Girth Weld Acceptance
Standards in the Canadian Gas Pipeline Code”, “Investigation of Plastic Collapse Criteria for
Defects in Line Pipe Girth Welds During Bending” , Proc. 3rd. Int. Conf. on Welding and
Performance of Pipeline, T.W.I., London, Nov. 1986, paper 21.
26. Hopkins, P., and Denys, R.M., "The Background to the European Pipeline Research Group's Girth
Weld Limits for Transmission Pipelines," EPRG/NG-18, 9th Biennial Joint Technical Meeting on
Line Pipe Research, Houston, Texas, May 11-14, 1993, Paper 33.
27. Denys, R.M., "Toughness Requirements for Pipeline Integrity," Proc. of 13th OMAE Conf., Vol.
III, Part A, Copenhagen, Denmark, June 18-24, 1995.
28. Dolby, R.E., “Charpy V and COD – Correlations Between Test Data for Ferritic Weld Metal”, Metal
Construction, Vol. 13, No. 1, Januari 1981, pp. 43-51.
29. Denys, R., “Lowerbound Charpy V Notch Toughness for Plastic Collapse”, To be published.
30. NN. Gegevens bestanden van Laboratorium Soete m.b.t. kleine en grootschalige proeven op
rondnaden in pijpleidingen, April 2001.
31. NN, Method of assessment for defects in fusion welded joints with respect to brittle fracture,
Japan Welding Engineering Society Standards WES 2805-1980, The Japan Welding Engineering
Society, Tokyo (1980)
32. Hopkins, P., Pistone, V.and Clyne, A., “A Study of the Behaviour of Defects in Pipeline Girth
Welds, The Work for EPRG”, Conf. Proc. Int. Conf. on Pipeline Reliability, Canmet, Pergamon,
Calgary Canada, Pergamon Press, 1992,
33. Hopkins, P., “The Application of Fitness-for-Purpose Methods to Defects Detected in Offshore
Transmission Pipelines”, Proc. Conf. On Welding and Weld Performance in the Process Industry,
IBC, London, 27-28 April, 1992.
34. Denys RM, Lefevre AA, De Baets P and Degrieck J: “Effects of stable ductile crack growth on
plastic collapse defect assessments”. Pipeline Technology Proceedings of the third international
pipeline technology conference Volume I, Brugge, Belgium, May 21-24 2000. Amsterdam:
Elsevier Science, 2000, 69-89.
35. Denys, R., 'Girth Welds In Modern Pipelines, Quality Control and Safety Aspects', Pipeline
Pigging and Integrity Monitoring Conference, Amsterdam, 11-14 april, 1994.
36. Clyne, A. J. and Jones, D. G., "The Integrity of Transmission Pipelines Containing Circumferential
Girth Weld Defects," Pipeline Technology, Eds. R. Denys, Vol. II, 1995, pp. 299-314.
37. Worswick M.J., Coote R.I., and Burns D.J., “Investigation of Plastic Collapse Criteria for Defects
in Line Pipe Girth Welds During Bending” , Proc. 3rd. Int. Conf. on Welding and Performance of
Pipeline, T.W.I., London, Nov. 1986, paper 30.
38. Woodley, C.C., Burdekin, F.M. and Wells, A.A., “Mild Steel for Pressure Equipment at Sub-Zero
Temperatures,” Brit. Weld J. 1964, 11, 3, pp.123-136, 1964.
39. Soete, W., and Denys, R.M., “Fracture Toughness Testing of Welds”, Proc. of Conf. on Welding of
HSLA (Micro-alloyed) Structural Steels, Roma, Nov. 1976, ASM, ISBN 087170 0050, pp.63-84
40. Denys, R.M., “Wide Plate Testing of Weldments, Parts I, II and III”, “Fatigue and Fracture
Testing of Weldments”, ASTM STP 1058, Eds. H. McHenry and J. Potter, ASTM, Philadelphia,
1990, pp.157-228.
41. Denys R.M., “Toughness Requirements In Transversely Load Welded Joints - An Evaluation
Based On Wide Plate Testing”, The Fracture Mechanics of Welds, EGF Pub. 2 (Edited by J.G.
Blauel and K.H. Schwalbe, 1987), Mechanical Engineering Publications, London, pp. 155-189.
42. Mayville, R.A. and Hilton, P. D., “A Study of Plasticity Effecst on the Crack Geometry Correction
Factors Used with the COD-Design Curve”, Fracture Mechanics, 18th Symposium, ASTM STP 945,
1988, pp. 686-698.
43. Roodbergen, A.H., and Denys, R., Limitations of Fitness for purpose Assessments of Pipeline
Girth Welds, Proc. Int. Conf. on Pipeline Technolgy, AIM and Commission of European
Communities, Rome, Nov. 1987.
44. Denys, R.M., "Effect of Weld Metal Matching on Girth Weld Performance, Vol. I," AGA Catalog No.
L51 651-IN1, p. 96, February 1992.
45. Scott, P.M., “Interpretative Study of Published and Recent Research on the applicability and
limitations of Current Fracture Prediction Models for Girth Welds,” Canadian Metallurgical
Quarterly, Vol. 32, No. 3, pp. 223-237, 1993
46. Wilkowski GM, Olson RJ and Scott PM: State of the art report on piping fracture mechanics
NUREG/CR-6540. Columbus, Ohio, Battelle, 1998.
47. Dawes, M.G., 1985, The CTOD Design Curve Approach: Limitations and Finite Size and
Application, Welding Institute Report 278/1985.
48. Andrews, R.M., Denys, R.M., Muesch. K, “Recent studies to Enlarge the Limits of the EPRG
Guidelines on the Assessment of Defects in Transmission Pipeline Girth Welds”. PRC/EPRG Tenth
Biennial Joint Technical Meeting, New Orleans, Louisiana, USA, April 30-May 4,, 2001. Paper 3
49. Denys, R., Difference Between Small and Large Scale Testing of Weldments, An Evaluation of the
Weld Metal Overmatching Effects in Relation to Weld Joint Performance”, The Welding Journal,
Research Supplement, N2, Febr. 89, p 33s-43s.
50. Lian, B., Denys, R and L. Van de walle, An experimental assessment on the effect of weld metal
yield strength overmatching in pipeline girth welds , Proc. of the Third Conference on Welding
and Performance of Pipeline, TWI, London, November 1986.
51. Denys, R., Lefevre, A.A., Martin, J.T., Glover, A.G., Yield Strength Mis-Match Effects on Fitness
for Purpose Assessments of Girth Welds, PRC/EPRG Tenth Biennial Joint Technical Meeting on
Line Pipe Research, American Gas Association, Paper 3, Cambridge, UK, April 18-21, 1995, pp.
1-15
52. Denys R., “Methods for the Assessment of Girth Weld Strength”, Welding Technology Institute of
Australia, Proc. Conference on Welding of High Strength Thin-Walled Pipelines”, Wollongong,
October 26, 1995, Paper 7, 1-15.
53. Denys, R., and Lefevre, A.A., "Weld Metal Mismatch, Challenges and Opportunities” Proc. ICAWT
’99 - Pipeline Welding and Technology Conference, Galveston, TX, USA, EWI-AWS, October 26-
26, 1999.
54. Teale, R.A., Smoot W.T., and Trotter, J.J. Pipeline Girth Welding Consumables for High Strength
Steels, Proc. 5th. Int. OMAE Conf. Houston, 1986, pp. 145 -148.
55. Teale, B. and Price J.C., Welding Requirements for Major North Sea Gas Lines, Proc. Pipeline
Technology Conference, Oostende, Oct. 15-18, 1990.
56. Denys, R., “Is the Transverse Weld Tensile Test a Reliable Test?”, “Proceedings Pipeline
Technology Conference”, September 11 - 14, 1995, ISBN Number: 0 444 82197, Vol II - pp. 581
– 591
57. Denys R., “How Much Weld Metal Yield Strength Do We Need?”, Second Pipeline Technology
Conference Ostend, September 11-14, 1995, Vol II - pp. 555 - 560.
58. Cousin V., and Baguley R., “Enhancing Pipeline Reliability: A Constractors Perspective”, Canmet,
Int. Conf. on Pipeline Reliability, June 2 5, 1992, Calgary, Canada, Paper II-9.
59. Denys, R.M., Lefevre A.A., A.G. Glover, "Weld Metal Yield Strength Variability in Pipeline Girth
Welds", Proc. of the Second International Conference on Pipeline Technology, Elsevier Publishers,
Ostend, 11-14 september 1995, p.591-598.
60. Horsley F.J. and Glover, A.G., “Girth weld Strength Under-Matching in High Strength Natural Gas
Pipelines”, Second Symposium on Mis-Matching of Interfaces and Welds”, Proceeding GKSS
Geesthacht, April 24-26,1996, pp. 547-560.
61. Bannister, A.C. and Harrison, P.L., “A Comparative Study of Wide Plate Behaviour of a Range of
Structural Steels Using the Failure Assessment Diagram”, OMAE, 1995, Vol III, Materials
Engineering, pp. 65-75.
62. C. Dallam, M. Quintana, “Welding X-80 And Beyond: A Consumable Manufacturers Perspective”,
Pipeline Technology Proceedings of the third international pipeline technology conference Volume
II, Brugge, Belgium, May 21-24 2000. Amsterdam: Elsevier Science, 2000, pp.483-498.
63. D.J. Widgery, “Welding High Strength Steel Pipelines - A Consumable Manufacturer's View”,
Pipeline Technology Proceedings of the third international pipeline technology conference Volume
II, Brugge, Belgium, May 21-24 2000. Amsterdam: Elsevier Science, 2000, pp.499-508
65. Blackman SA and Dorling DV: Advanced welding processes for transmission pipelines. In: Denys
RM, editor. Pipeline Technology Proceedings of the third international pipeline technology
conference Volume II, Brugge, Belgium, May 21-24 2000, Amsterdam: Elsevier Science, 2000,
371-387.
66. Gross B., O’Beirne J., Delanty B., Comparison of Radiographic and Ultrasonic Inspection Methods
on Mechanized Girth Welds, Proc. Pipeline Technology Conference, Oostende, Oct. 15-18, Paper
14.17, 1990
67. Van Merrienboer H.A.M, Mechanised Ultrasonic Inpsection of Manually Welded Girth Welds
Compared with Radiography, Canmet, Int. Conf. on Pipeline Reliability, June 2-5, 1992, Calgary,
Canada, Paper IV14.
68. Coors P., Denys R., ‘Effect of Automated Ultrasonic Inspection Capabilities on the EPRG
Guidelines’, Paper 16, PRCI/EPRG 12th Biennial Joint Technical Meeting on Line Pipe Research,
1999, Groningen.
69. F.H. Dijkstra, J. van der Ent, T. Bouma, “Defect Sizing and ECA Criteria: State of the Art in AUT”,
Pipeline Technology Proceedings of the third international pipeline technology conference Volume
I, Brugge, Belgium, May 21-24 2000. Amsterdam: Elsevier Science, 2000, 69-89, pp.483-498
70. Denys, R.M., Lefevre, A.A., De Jaegher, C. and Claessens, S., “Weld Acceptance Criteria”, Final
Report of Study Commissioned by the International Pipe Line & Offshore Contractors Association
(IPLOCA), 2000, Gent, Belgium
71. Denys, R., Lefevre. A.A. and De Jaeger, C., “Defect Interaction for Ductile Material Behaviour”,
OMAE98-2058, ”Proceedings of 17th OMAE Conference, Materials Sessions, CD-rom, p. 8, July 5-
9,1998, Lisbon, ISBN No 0-7918-1952-3.