Engineering Failure Analysis 14 (2007) 1245–1265

www.elsevier.com/locate/engfailanal

Best practice for the assessment of defects in

pipelines – Corrosion
A. Cosham a, P. Hopkins a, K.A. Macdonald b,*

a
Penspen Integrity, Newcastle Business Park, Newcastle Upon Tyne NE4 7YL, UK
b
University of Stavanger, 4036 Stavanger, Norway

Received 6 July 2006; accepted 20 November 2006

Available online 5 February 2007

Abstract

Corrosion is a common form of degradation in pipelines that reduces both the static and cyclic strength of a pipeline. In
this paper, the best practices for the assessment of corrosion in pipelines are presented. Small scale and full scale tests, the-
oretical analyses and assessment methods are discussed and best practice in terms of assessment methods are described.
The work stems from the results of a joint industry project, documenting the best techniques currently available for the
assessment of pipeline defects (such as corrosion, dents, gouges, weld defects, etc.) in a simple and easy-to-use manual,
and gives guidance on their use.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Pipeline; Defect assessment; Corrosion

1. Introduction

Oil and gas transmission pipelines have a good safety record. This is due to a combination of good design,
materials and operating practices; however, like any engineering structure, pipelines do occasionally fail. The
most common causes of damage and failures in onshore and offshore, oil and gas transmission pipelines in
Western Europe and North America are external interference (mechanical damage) and corrosion [1–3].
Assessment methods are needed to determine the severity of such defects when they are detected in pipelines.
Defects occurring during the fabrication of a pipeline are usually assessed against recognised and proven
quality control (workmanship) limits. However, a pipeline will invariably contain larger defects during its life,
and these will require fitness-for-purpose assessment to determine whether or not to repair the pipeline. Con-
sequently, the past 40 years has seen the development of a number of methods for assessing the significance of
defects. Some of these methods have been incorporated into industry guidance; others are to be found in the
published literature. However, there is no definitive guidance that contains all of the assessment techniques, or

*
Corresponding author.
E-mail addresses: [email protected] (A. Cosham), [email protected] (P. Hopkins), [email protected] (K.A.
Macdonald).

1350-6307/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2006.11.035
1246 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

assesses each method against independently published test data, or indeed recommends best practice in their
application.

1.1. Fitness-for-purpose

Fitness-for-purpose (FFP), as discussed here, means that a particular structure is considered to be adequate
for its purpose, provided the conditions to reach failure are not reached [4].1 Fitness-for-purpose is based on a
detailed technical assessment of the significance of the defect. Local and national legislation and regulations
may not permit certain types of defects to be assessed by fitness-for-purpose methods or may mandate specific
limits. Such issues should always be considered prior to an assessment.
Safety must always be the prime consideration in any fitness-for-purpose assessment and it is always nec-
essary to appreciate the consequences of a failure. These will influence the necessary safety margin to be
applied to the calculations.

1.2. Pipeline integrity management

Pipeline failures are usually related to a breakdown in a system, e.g. the corrosion protection ‘system’ has
become faulty, and a combination of ageing coating, aggressive environment, and rapid corrosion growth may
lead to a corrosion failure. This type of failure is not simply a ‘corrosion’ failure, but a ‘corrosion control system’
failure. Similar observations can be drawn for failures due to external interference, stress corrosion cracking, etc.
Pipeline integrity management is the general term given to all efforts (design, construction, operation, main-
tenance, etc.) directed towards ensuring continuing pipeline integrity. The American Petroleum Institute (API)
has developed an industry consensus standard that gives guidance on developing integrity management pro-
grammes (API 1160) [5]. The American Society of Mechanical Engineers (ASME) has developed a similar
integrity management guidelines for a supplement to ASME B31.8 [6].
To address this industry need, a Joint Industry Project has developed a Pipeline Defect Assessment Manual
(PDAM). PDAM presents the ‘best’ currently available methods for the assessment of pipeline defects (such as
corrosion, dents, gouges, weld defects, etc.), in a simple and easy-to-use manual, and gives guidance in their
use. It is based on an extensive critical review of published ‘fitness-for-purpose’ methods and test data. The
best methods for assessing a variety of different types of defect are summarised (see Table 1) but this paper
focuses on the assessment of corrosion with reference to previous reviews of corrosion assessment methods
and published full scale test data.

2. Assessment methods in the pipeline defect assessment manual

A summary of all of the methods recommended in the pipeline defect assessment manual for predicting the
burst strength of a defect subject to internal pressure is given in Table 1 [7–16]. Longitudinally and circumfer-
entially-orientated defects are considered. The primary methods are plastic collapse (flow stress dependent or
limit state) failure criteria, and are only appropriate if a minimum toughness is attained [18]. The secondary
methods (indicated in italic font) are the alternative methods recommended when a minimum toughness is
not attained. Upper shelf behaviour is assumed throughout. The general procedures for assessing flaws in struc-
tures, based on fracture mechanics, given in BS 7910 [4] (and API 579 [17]) can be applied in general (irrespective
of upper or lower shelf behaviour), but will normally be conservative compared to the pipeline specific methods.2
Having introduced the subject and it’s background, the remainder of this paper:

(i) describes in general terms the various methods for the assessment of corrosion,
(ii) summarises the available full scale test data,

1
Note that fitness-for-purpose may also have a legal and contractual meaning in different countries.
2
PAFFC incorporates correlations between the fracture toughness and the upper shelf Charpy impact energy; therefore, PAFFC is not
applicable to lower shelf conditions (although the underlying theoretical model is applicable if the fracture toughness (K, J or d) is
measured).
Table 1
Recommended methods for assessing the burst strength of defects subject to internal pressure
Internal pressure (static) longitudinally orientated Internal pressure (static) circumferentially orientated
Corrosion DNV-RP-F101 [13] Kastner local collapse solution [15]
Modified B31G [14]
RSTRENG [14]

A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

Gouges NG-18 equations [7] Kastner local collapse solution
PAFFC [8,9] BS 7910 (or API 579)
BS 7910 [4] (or API 579 [17])
Plain dents Empirical limits
Kinked dents No methoda
Smooth dents on welds No method
Smooth dents and gouges Dent-gouge fracture model [11,12] No method
Smooth dents and other types of defect Dent-gouge fracture model No method
b
Manufacturing defects in the pipe body NG-18 equations Kastner local collapse solution
BS 7910 (or API 579) BS 7910 (or API 579)
Girth weld defects – Workmanship, EPRG [10]
BS 7910 (or API 579)
Seam weld defects Workmanship –
BS 7910 (or API 579)
Cracking BS 7910 (or API 579)
PAFFC
Environmental crackingc BS 7910 (or API 579)
PAFFC
Leak and rupture NG-18 equations, PAFFC Schulze global collapse solution [16]
a
‘No method’ indicates limitations in existing knowledge, and circumstances where the available methods are too complex for inclusion in a document such as PDAM.
b
The term ‘manufacturing defect’ covers a wide range of pipe body defect (laminations, inclusions, seams, gouges, pits, rolled-in slugs, etc.). Consequently, it may not be possible to
characterise a manufacturing defect in the pipe body as a metal-loss or crack-like defect. In these circumstances it is necessary to rely on workmanship limits and industry experience.
c
Environmental cracking (stress corrosion cracking, hydrogen blisters, hydrogen stress cracking, etc.) can be very difficult to measure and assess.

1247
1248 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

(iii) considers the role of fracture toughness,

(iv) identifies what are generally recognised as the ‘best’ methods for assessing corrosion, and
(v) presents the key considerations when assessing a corrosion defect in a pipeline.

3. Corrosion in pipelines

Corrosion is an electrochemical process. It is a time dependent mechanism and depends on the local envi-
ronment within or adjacent to the pipeline. Corrosion usual appears as either general corrosion or localised
(pitting) corrosion. There are many different types of corrosion, including galvanic corrosion, microbiologi-
cally induced corrosion, AC corrosion, differential soils, differential aeration and cracking. Corrosion causes
metal loss. It can occur on the internal or external surfaces of the pipe, in the base material, the seam weld, the
girth weld, and/or the associated heat affected zone (HAZ).
Internal and external corrosion are together one of the major causes of pipeline failures. Data for onshore
gas transmission pipelines in Western Europe for the period from 1970 to 1997 indicates that 17% of all inci-
dents resulting in a loss of gas were due to corrosion [1]. Incident data from the Office of Pipeline Safety in the
USA for the year 2001 attributes 29% of incidents in liquid pipelines, and 19% of incidents in gas pipelines, to
corrosion [3].
Environmentally assisted cracking, such as stress corrosion cracking (low pH and high pH SCC), hydrogen
induced cracking (HIC), etc., must be assessed using different methods to those describe here, because the deg-
radation mechanism causes cracking, blistering, etc., rather than blunt metal loss features.
Corrosion in a pipeline may be difficult to characterise. Typically, it will have an irregular depth profile and
extend in irregular pattern in both longitudinal and circumferential directions (as illustrated in Fig. 1). It may
occur as a single defect or as a cluster of adjacent defects separated by full thickness (uncorroded) material.
There are no clear definitions of different types of corrosion defects. The simplest and perhaps most widely
recognised definitions are as follows:

(i) pitting corrosion, defined as corrosion with a length and width less than or equal to three times the
uncorroded wall thickness, and
(ii) general corrosion, defined as corrosion with a length and width greater than three times the uncorroded
wall thickness.

The pipeline operators forum (POF) has developed a set of specifications and requirements for the inspec-
tion of pipelines by intelligent pigs, including definitions of types of metal loss features (pinhole, pitting,

Fig. 1. The irregular length, width and depth of a typical corrosion defect.
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1249

slotting, grooving and general) [19]. ‘Blunt’ has been defined in the literature as defects whose minimum radius
equals or exceeds half of the pipe wall thickness [20], and defects with a width greater than their local depth
[21].
A considerable amount of time and effort has been devoted to the study of the static strength of corrosion
defects in pipelines. Initially, research concentrated on the behaviour of sharp defects (machined V-shaped
notches and slits), but subsequently the work was extended to consider artificial and real corrosion defects.
The primary focus of research into the significance of corrosion defects has been towards longitudinally-ori-
entated defects subject to internal pressure loading. Several recently published papers have discussed the back-
ground to the various methods for assessing corrosion that exist in the literature (B31G, modified B31G,
RSTRENG, DNV-RP-F101, etc.) [22–24], so such information is not repeated here.

4. Full scale burst tests of real and artificial corrosion defects

Full scale vessel burst tests of real and artificial corrosion defects in line pipe, orientated predominantly lon-
gitudinal and subject to internal pressure, have been carried out by a number of different organisations. Arti-
ficial corrosion defects are machined pits, grooves and patches, blunt, flat-bottomed defects with a uniform
profile. A smaller number of tests have been conducted on longitudinal corrosion subject to axial and/or bend-
ing loads in addition to internal pressure, and on circumferential and helical defects; these tests are not con-
sidered here.
The total number of published burst tests of ‘corrosion’ defects considered here is 343 (including 215 tests in
the AGA/PRCI Database of Corroded Pipe Burst Tests), although only 157 tests are considered to be reliable
tests of longitudinally orientated corrosion subject to internal pressure (see below). The number and type of
tests conducted by each organisation is given below.

Barkow (1972) [25] 4 vessel tests (real corrosion)

Texas Eastern (1972) [26] 15 vessel tests (real corrosion)
Battelle (1973) [27] 47 vessel tests (real corrosion)
British Gas (1974) [28] 22 vessel tests (real corrosion)
University of Waterloo (1990, 1991) [29–32] (Mok et al.) 20 burst tests (machined grooves)
British Gas (1992) [33] 23 vessel tests (machined)
Vieth and Kiefner (1994) [34] (AGA/PRCI) 124 vessel tests (most real
corrosion)
Kiefner et al. (1995) [35] (AGA/PRCI) 91 vessel tests (some real
corrosion)
Battelle (1995) [20] 2 vessel tests (machined)
University of Waterloo (1992) [36] 12 vessel tests (real corrosion
pits)
University of Waterloo (1994–1996) [37–39] 26 vessel tests (machined pits)
Fu and Batte (1999) [40]3 (British Gas) (LPC) 2 vessel tests (machined patch)
Transgas (2000) [41] 17 vessel tests (some real
corrosion)
Petrobras (2000) [42] 9 vessel tests (machined grooves)
University of Waterloo (1996, 2000) [43–45] 40 vessel tests (real corrosion)
British Gas (1992) [33] 4 ring tests (machined slits)
British Gas, Shell (1992, 1994) [46,47] 9 ring tests (machined grooves)
University of Waterloo (1998) [48] (Roberts and Pick) 10 vessel tests (machined)
DNV (2000) [49] 12 vessel tests (machined)

3
The Linepipe Corrosion Group Sponsored Project (conducted by British Gas) conducted approximately 81 full scale vessel burst tests
and 52 ring expansion tests of artificial corrosion defects (both single defects and interacting defects). However, most of these tests have
not been published.
1250 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

The AGA/PRCI Database of Corroded Pipe Tests [34,35] is the most comprehensive source of publicly
available burst tests of real and simulated corrosion in line pipe material. Although it does not included all
of the tests identified above (based on the 1995 edition of the database, there are an additional 108 vessel tests
and 21 ring tests in the published literature).
The tests include artificial (simulated) corrosion defects (machined pits, slots and patches) and real corro-
sion defects, single defects and interacting defects, burst tests (internal pressure only) and combined loading
(pressure, bend and axial compression) tests. Given the large number of tests from different sources contained
in the Database of Corroded Pipe Tests and in the wider published literature, the data must be used with some
care, otherwise tests that are not directly comparable may be considered together. Some of the test data is not
reliable, due to the test having been subject to a number of pressure cycles (as is the case when a vessel con-
taining multiple defects is tested repeatedly), or the test being terminated prior to failure of the defect. Other
tests involve interaction, or the effects of combined loading, and should not be considered with tests of single
defects. For some tests the available information is incomplete.
The ‘reliable’ test data has been identified as above and guided by the tests omitted from the further val-
idation of RSTRENG [35]. The range of the experimental parameters of all of the ‘reliable’ burst tests of pre-
dominantly longitudinally orientated (real or artificial) ‘corrosion’ defects (a total of 159 tests) is
Pipe diameter, mm 273.0 to 914.4
Wall thickness, mm 4.57 to 22.1
2 R=t ratio 31.5 to 130.3
Grade (API 5L) A25 to X65
Yield strength, Nmm2 196.0 to 515.0
Tensile strength, Nmm2 277.0 to 658.0
Yield to tensile ratio 0.60 to 0.85
2/3 Charpy impact energy, J 18.0 to 90.0
Maximum defect depth (d), mm 1.60 to 17.1
d/t 0.25 to 0.97
Defect length ð2cÞ, mm 19.35 to 3048.0
2c=Rt0:5 0.527 to 49.7
Defect width, mm 0.15 to 304.8
Burst pressure, Nmm2 4.88 to 25.2
Burst stress, Nmm2 145.9 to 589.5
Burst stress (% SMYS) 45.4 to 186.3
Real corrosion defects have an irregular profile, whereas artificial (machined) corrosion defects are typically
flat bottomed. The profile of a corroded area must be considered if an accurate prediction of the burst pressure
is desired. River-bottom profiles are available for all of the defects in the Database of Corroded Pipe Tests. This
information is not available for some of the additional published tests of real corrosion.

(1) The longitudinal extent of a corroded area is the most important length parameter for the burst strength
under internal pressure loading. The circumferential extent has a small influence on the burst strength,
but the effect is sufficiently small to not need considering. However, the circumferential extent must be
considered if external axial and/or bending loads are present.
(2) External loads reduce the burst pressure compared to the case of an end-capped pressure vessel (axial
stress equal to half the hoop stress). The effect of tensile external loads is generally small, whilst compres-
sive loads can cause a significant reduction in the burst pressure.
(3) No difference between the behaviour of internal and external corrosion has been noted in full scale tests
or finite element analyses (but noting that pipelines are thin-walled geometries).
(4) Short defects (typically less than 3t in length) of any depth record high burst pressures, typically above
the pressure required to yield the uncorroded pipe.
(5) In modern, tough, line pipe steel the flow stress for smooth corrosion defects is the ultimate tensile
strength of the material.
(6) The effect of toughness on a sharp defect is more significant than that on a blunt defect.
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1251

5. The role of geometry and flow stress

5.1. Ductile failure

Two possible scenarios for the ductile failure of a blunt part-wall defect in a tough line pipe steel (i.e.
excluding the possibility of cleavage fracture) have been identified, as described by Leis and Stephens
(1997) [50,51] and Fearnehough et al. [52,53].

(1) As the load (pressure) increases, local wall thinning will occur in the remaining net section. This local
wall thinning could continue, leading to necking of the wall and failure due to void nucleation, growth
and coalescence in a manner comparable to that of a tensile test specimen.
(2) Alternatively, a crack could initiate at the base of the defect due to the presence of micro-stress raisers (e.g.
local surface irregularities caused by a corrosion mechanism) through a process of void nucleation and
growth. The behaviour after the initiation of a crack would depend on the toughness of the material.
In a high toughness material, initiation would be delayed to a higher load and further stable ductile tear-
ing would be slower, or a growing crack could blunt; wall thinning would continue and the failure load
would tend to that of plastic collapse. However, in a lower toughness material, once initiated, the crack
would extend by stable ductile tearing, reducing the remaining wall thickness and hence reducing the
degree of wall thinning that occurs before failure. The load at failure would be less than that predicted
by the plastic collapse limit state because of the stable ductile tearing.

5.2. The role of geometry

The failure of a part-wall defect in a pipeline subject to internal pressure has two limits:

(1) a defect with a length and depth tending towards zero (i.e. defect-free pipe), and
(2) an infinitely long defect of finite depth (see Fig. 2).

It is assumed that the line pipe material is tough and that failure occurs due to plastic collapse (i.e. unstable
plastic flow). In the first case, the failure stress tends towards the failure stress of defect-free pipe, based on the

failure pressure of
burst pressure of defect-free pipe (plastic

1.0
burst pressure of defect normalised by

pipe of reduced
blunt defect
cross-section (1-d/t)
collapse limit load)

sharp defect
decreasing increasing
acuity toughness

increasing
defect
depth (d)

0.0
0 ∞

normalised defect length

Fig. 2. The effect of material toughness, defect depth, length and acuity on burst strength.
1252 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

full wall thickness (t), and in the second case it tends towards the failure stress of defect-free pipe, but based on
the reduced wall thickness ðt  dÞ.
The failure stress of a part-wall flaw of finite length lies between the above two extremes; it is a function of:
(1) the geometry of the pipe and the geometry of the defect, and (2) the material.

5.3. The role of flow stress

The failure stress of defect free pipe tends towards the ultimate tensile strength of the material, as measured
in a uniaxial tensile test. However, account must be taken of large scale geometry effects (a cylinder under
internal pressure exhibits geometric softening: as the pressure increases, the diameter increases; the hoop stress
increases because of both the increase in pressure and the increase in diameter). Theoretically, the failure stress
depends upon the strain hardening characteristics of the material and the assumed yield criterion (Tresca or
von Mises) [46,50,51,54,55]. Experimental results indicate that the failure stress lies between the Tresca and
von Mises bounds, and is reasonably approximated by the ultimate tensile strength.
The failure stress of defect free pipe can be interpreted as a flow stress, although the term reference stress4
has also been proposed (by researchers at Battelle [50,51]) to differentiate it from the term flow stress as used in
fracture mechanics.5 The flow or reference stress describes the role of the material.

5.4. The failure of a blunt, part-wall defect

Therefore, the failure stress of a blunt, part-wall defect subject to internal pressure can be predicted by a
failure criterion that comprises a flow stress term and a geometry term. The geometry term includes the effects
of bulging, the global stiffness, the stiffness of the defect, defect acuity, etc. The flow, or reference, stress rep-
resents the material behaviour. Note that the complete separation of material and geometry terms is an
approximation, introducing some scatter into predictions of test data or numerical data.

5.5. The role of geometry and flow stress in the published methods

Failure criteria such as the flow stress dependent forms of the NG-18 equations [7] (and ASME B31G
[27,56], modified B31G [14], etc.) have been described as plastic collapse failure criteria. However in many
of the tests, on which these older semi-empirical failure criteria are based, failure was preceded by significant
amounts of ductile tearing and some of the steels had a low toughness. Furthermore, the geometry term was
empirical and the flow stress was adjusted to fit the test results. This lead to empirical definitions of the flow
stress (reference stress) that were conservative, since they were biased towards the behaviour of older steels.
The NG-18 equations were developed from tests of V-shaped notches, not blunt, part-wall defects. Therefore,
the methods for assessing corrosion based on the NG-18 equations (ASME B31G, modified B31G, etc.) have a
conservative bias when applied to tests of blunt, part-wall defects.
Developments in the accuracy of failure criteria follow from their being better able to describe the effects of
reference stress and geometry. The more recent failure criteria for corrosion (DNV-RP-F101 [13], PCORRC
[21]) have used finite element analyses of blunt, part-wall defects to determine the form of the geometry term,
and have considered the form of the reference stress in more detail. These failure criteria were validated
against burst tests of modern line pipe steels containing blunt, part-wall defects or real corrosion defects. Mod-
ern line pipe steels have a higher toughness than older steels, such that the failure of blunt part-wall defects is
controlled by plastic collapse (where plastic collapse is defined in terms of the defect-free failure stress (i.e. the
ultimate tensile strength)), and the effect of toughness is negligible. However, difficulties can then arise in
applying the more recent methods to older, lower toughness, line pipe. The more recent methods may be none
conservative.

4
The reference stress is the failure stress of defect-free pipe; it represents the plastic collapse limit state. The reference stress is
independent of the defect geometry.
5
The flow stress is an empirical concept. It was introduced to incorporate plasticity into a linear-elastic fracture mechanics analysis. The
flow stress is not necessarily the stress at plastic collapse (where plastic collapse is failure due to plastic flow).
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1253

1.4

CONSERVATIVE
actual failure stress/predicted failure stress
1.2

1.0

0.8
Battelle (1973)
CANMET (1988)
TWI (1982)
0.6 Batelle (1986)
CSM SNAM EUROPIPE (2000)
Keller et al. (1987)
UNCONSERVATIVE Herrera et al. (1992)

0.4
0 10 20 30 40 50 60 70 80 90 100 110
2/3 Charpy V-notch impact energy, J

Fig. 3. The effect of toughness on flow stress dependent NG-18 predictions of burst tests of machined V-shaped notches and slots.

6. The effect of toughness

The effect of toughness on the failure stress of blunt, part-wall defects can be observed through comparisons
with the published burst tests of real and artificial corrosion. The influence of toughness is clear in tests of
part-wall V-shaped notch tests, as conducted by Battelle during the development of the NG-18 equations
[7] (see Fig. 3): as the toughness decreases, a flow stress dependent failure criterion becomes inappropriate
(the predictions become increasingly non-conservatively). The influence of toughness on the failure of corro-
sion defects is less clear (see Figs. 4 and 5) because corrosion defects are blunt, and the irregular profile of a
real corrosion defect introduces experimental scatter. Increasing conservatism with increasing toughness is
apparent for modified B31G (Fig. 4), but not for DNV RP-F101 (Fig. 5). The toughness of the line pipe steel
is not known for a large number of the tests in the AGA/PRCI Database of Corroded Pipe Tests.6
Figs. 4 and 5 include those AGA/PRCI tests of defects which have not been subject to multiple pressure
cycles are shown (since removing all of the ‘unreliable’ tests would remove some tests of low toughness line
pipe and give an incomplete picture). The tensile strength is not available for all of the tests, so some tests
in Fig. 4 are not to be found in Fig. 5. A number of test results are noteworthy7: test index 107 is a corrosion
pit that contained a prior through-wall crack and can therefore be excluded, test indices 215, 9, 6 have tran-
sition temperatures at or above the test temperature, and test index 1 is of a line pipe steel with an unreported
transition temperature. Test indices 1, 6 and 9 are all the first test of a series of multiple tests of a single vessel.
No metallurgical analysis of the fracture surfaces is reported, so the actual failure mechanism (ductile or cleav-
age) is the subject of conjecture.
The tests of real corrosion defects that are non-conservatively predicted by modified B31G and DNV
RP-F101 involve line pipe steels tested at a temperature below the transition temperature (or the transition
temperature is unknown). None of the assessment methods are applicable to line pipe steel that is in the
transitional region or on the lower shelf.

6
Tests with an unknown toughness are plotted as having a zero toughness, to illustrate the range of the predictions.
7
The test index number refers to the number of the test in the AGA/PRCI Database of Corroded Pipe Tests.
1254 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

2.0

Battelle (1973)

actual failure stress/predicted failure stress

1.8 AGA Database (1994, 1995)
British Gas (1999) CONSERVATIVE
Det Norske Veritas (2000)
1.6 Cronin and Pick (2000)
Battelle (1995)

1.4

1.2

1.0 INDEX 1
INDEX 215 INDEX 6

0.8
INDEX 9

0.6
UNCONSERVATIVE
INDEX 107
0.4
0 10 20 30 40 50 60 70 80 90 100
2/3 Charpy V-notch impact energy, J

Fig. 4. The effect of toughness on modified B31G predictions of burst tests of real and artificial corrosion defects.

2.0

Battelle (1973)
AGA Database (1994, 1995)
actual failure stress/predicted failure stress

1.8
British Gas (1999)
CONSERVATIVE
Det Norske Veritas (2000)
Cronin and Pick (2000)
1.6 Battelle (1995)

1.4

1.2

1.0

INDEX 1
0.8
INDEX 6

0.6 INDEX 9
UNCONSERVATIVE

0.4
0 10 20 30 40 50 60 70 80 90 100

2/3 Charpy V-notch impact energy, J

Fig. 5. The effect of toughness on DNV-RP-F101 predictions of real and artificial burst tests of corrosion defects.

Considering all of the published full scale tests, the lowest toughness is 18 J (13 ftlbf) and the maximum wall
thickness is 22.5 mm (1.0 in.).8 Consequently, considering the basis of the various criteria and a comparison
with full scale test data, ASME B31G, modified B31G and RSTRENG are applicable to low, moderate and

8
Fu (1999) has tested line pipe up to 25.4 mm, but the test results have not been published [22].
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1255

high toughness steels (assuming upper shelf behaviour), whilst DNV RP-F101 and PCORRC are only proven
for moderate to high toughness steels (see Sections 8 and 9).

7. Methods for assessing corrosion defects

Numerous methods have been developed for predicting the burst pressure of blunt part-wall defects,
which characterise the behaviour of typical corrosion defects. A number of these methods are listed
below (see also Table 2). All of these methods are primarily concerned with the longitudinal extent of
the corroded area and internal pressure loading. The methods are empirical or semi-empirical; the older
methods are based on the original Battelle part-wall failure criterion (the NG-18 equations), whilst the
more recent methods have partly developed from extensive numerical studies validated against test data
(see below).

(i) ASME B31G [56].

(ii) modified B31G (RSTRENG 0.85) (Kiefner and Vieth (1989)) [14].
(iii) RSTRENG (Kiefner and Vieth (1989)) [14].
(iv) Klever (1992) [54], Stewart et al. (1994) [46].
(v) SHELL92 (Ritchie and Last (1995)) [55].
(vi) DNV-RP-F101 (LPC) [13,40]*.
(vii) PCORRC (Stephens and Leis (2000)) [21,51]*.
(viii) CPS (Cronin and Pick (2000)) [44]*.
(ix) SAFE (SwRI) (Wang et al. (1998)) [57]*.
(an asterisk denotes the ‘new’ methods)

The DNV RP-F101 and SAFE methods can be applied to corrosion subject to axial and bending loads.
A detailed description of all of these methods can be found in the published literature.

(i) Real corrosion.

Corrosion defects are orientated and spaced in a random manner. In the analysis of such a defect an
attempt is made to characterise the corroded area by its projected length and area. The difficulty in
describing a three-dimensional corroded area by a few parameters introduces large scatter in compari-
sons of predicted to actual failure stress. The scatter is significantly reduced by the use of assessment
methods based on a river-bottom profile, but there is still more scatter than for flat-bottomed defects.
River-bottom methods (such as RSTRENG and those given in DNV-RP-F101 and CPS) are based
on iterative algorithms and are not suited to hand calculations. The methods based on a simple geomet-
ric approximation are closed-form methods.
Interaction between defects has been considered empirically, or through finite element analysis of a
narrow range of pipe and defect geometries. Limited guidance is available in the published literature.
(ii) Approximate methods for assessing real corrosion.
The original ASME B31G criterion [56], modified B31G criterion [14], DNV-RP-F101 (LPC), and
PCORRC define simple approximations to the exact corroded area, based on the maximum length
and the maximum depth of the defect. Corrosion typically has an irregular profile. The most conserva-
tive idealisation is a rectangular profile (as in DNV-RP-F101 and PCORRC). ASME B31G assumes a
parabolic profile (the 2/3 factor in the equation, see Table 2) and modified B31G assumes an arbitrary
profile (the 0.85 factor in the equation). The methods for assessing a river-bottom profile are also
approximations, because a river-bottom profile is an idealisation of the actual three-dimensional shape
of a corroded area.

All of the methods considered here assume that failure is due to a flow stress dependent mechanism and
can, therefore, be described by the tensile properties (yield strength, ultimate tensile strength) of the line pipe
steel. It is further assumed that the steel is on the upper shelf; the transition temperature is conventionally
defined as the temperature at which a DWTT specimen exhibits an 85% shear area. A minimum toughness
1256 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

Table 2
Methods for assessing the burst strength of a corroded area (based on longitudinal extent) subject to internal pressure loading
Method Basic equation ‘flow stress’e Defect shape ‘bulging’ factord
a rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  4ffi
NG-18 NG-18 rY þ 10 ksi rectangular (d/t) or defect area ðA=Ao Þ
1 þ 0:6275 p2cffiffiffi Dt
ffi  0:003375 p2cffiffiffiffi
Dt

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2ffi
ASME NG-18 1:1rY parabolic 2/3 (d/t) 1 þ 0:8 p2cffiffiffiffi Dt
3B31G
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  4ffi
modified NG-18 rY þ 10 ksi arbitrary 0.85 (d/t) 1 þ 0:6275 p2cffiffiffiffi  0:003375 p2cffiffiffiffi
Dt Dt
B31G
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  4ffi
RSTRENG NG-18 rY þ 10 ksi effective area and effective length 1 þ 0:6275 p ffiffiffi
2c ffi  0:003375 p2cffiffiffiffi
Dt Dt
(river bottom profile)
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2ffi
SHELL92 NG-18 rU rectangular (d/t) 1 þ 0:8 p2cffiffiffiffi Dt

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
LPC NG-18 rU rectangular (d/t) 1 þ 0:31 p2cffiffiffiffi Dt

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
DNV-RP- NG-18 rU rectangular (d/t) (and river bottom profile) 1 þ 0:31 p2cffiffiffiffi Dt
F101
b c
PCORRC New rU rectangular (d/t)
a
2c is equivalent to L.
b
The basic equation of the part-wall NG-18 failure criterion is (where M is the bulging factor and r
 is the flow stress)
2  3 " #
1  AAo 1  dt
4
rh ¼ r   5¼r  d  1 :
1  AAo M1 1 t M

c
The basic equation of the PCORRC failure criterion is
"   "   0:5 #!#
d 2c d
rh ¼ r
 1 1  exp 0:16 pffiffiffiffiffi 1  :
t Rt t
d
The bulging factor in NG-18, ASME B31G, modified B31G, RSTRENG and SHELL92 is one of the various forms of the Folias
factor. The bulging factor in LPC and DNV-RP-F101 was derived by curve fitting results to a non-linear geometry, elastic–plastic finite
element parametric study. The bulging factor in PCORRC is incorporated into the basic equation (see above).
e
SHELL92, LPC and DNV-RP-F101 state that using 0:9rU gives a conservative bias to the predictions (the 0.9 factor is not included in
Fig. 6).

may need to be satisfied. This is specifically the case for the recent, alternative, assessment methods
(DNV-RP-F101 (LPC), PCORRC, CPS [44], SAFE [57]) which assume that failure is controlled by plastic
collapse (plastic flow) (i.e. the flow stress is the ultimate tensile strength).
The methods are all similar in their general form, being based on the NG-18 equation for the failure of a
part-wall flaw, but differ in respect of assumptions and simplifications made in their derivation. These differ-
ences can be classified in terms of:

(i) The flow stress.

(ii) The geometry correction factor (also referred to as the Folias factor, or the length correction factor, or
the bulging correction factor).
(iii) The defect profile.

Stephens and Francini (2000) have concluded that two categories of assessment method for corrosion
defects can be described [24]:
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1257

1.0

0.9

modified B31G (X42)

0.8
d/t (normalised defect depth)

modified B31G (X65) DNV-RP-F101 (X42)

0.7 DNV-RP-F101 (X65)

0.6 PCORRC (X42)

PCORRC (X65)
0.5
ASME B31G
0.4

0.3

0.2
SHELL92 (X42)
0.1 SHELL92 (X65)

0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
2c/(Rt)^0.5 (normalised defect length)

Fig. 6. Methods for assessing the burst strength of a corroded area.

(1) empirically calibrated criteria that have been adjusted to be conservative for almost all corrosion defects,
irrespective of the toughness of the line pipe (these criteria are variously based on the yield strength, the
flow stress, or the ultimate tensile strength) (the ‘old’ methods), and
(2) plastic collapse criteria that are only appropriate for blunt defects in moderate to high toughness line
pipe (these criteria are based on the ultimate tensile strength) (the ‘new’ methods).

DNV-RP-F101 (LPC), PCORRC, SAFE and CPS should be regarded as belonging to the second category
of assessment method.
The ASME B31G, modified B31G, RSTRENG, SHELL92 [55], LPC, DNV-RP-F101 and PCORRC [21]
methods are summarised in Table 2 and Fig. 6.9 The LPC and DNV RP-F101 methods are essentially the same.
LPC, DNV-RP-F101 and PCORRC were developed from curve fitting to the results of parametric finite ele-
ment analyses of blunt, part-wall defects. These are theoretically calibrated methods (i.e. calibrated to average
data in the form of an experimentally validated finite element model and associated numerical failure criterion).
ASME B31G and related methods are by contrast based on curve fits to empirical data (originally tests on V-
shaped notches, then real corrosion defects). PCORRC and DNV RP-F101 give similar results (see Fig. 6).

8. Comparison of methods for assessing corrosion defects

8.1. Problems with scatter in the data

Large scatter is apparent in the predictions of the burst strength of real corrosion when using a method
based on a simple geometric idealisation (rectangular, parabolic, etc.), because maximum depth and maximum
length are insufficient to describe the irregular shape of a real corrosion defect (see Fig. 7).

9
All of the curves in this figure represent the failure locus of critical defect depth and defect length for a hoop stress equal to 100%
SMYS. For all of the methods except ASME B31G, the failure loci are dependent on the line pipe steel grade. The curves are presented for
two grades, X42 and X65.
1258 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

160

AGA Database (1994, 1995)

140 British Gas (1999)

Predicted Failure Stress/Yield Strength, percent

Det Norske Veritas (2000)
PETROBRAS (2000)
Cronin and Pick (2000)
120
Battelle (1995)

UNCONSERVATIVE
100

80

60

40

20
CONSERVATIVE

0
0 20 40 60 80 100 120 140 160
Failure Stress/Yield Strength, percent

Fig. 7. DNV-RP-F101 predictions of (‘reliable’) burst tests of real and artificial corrosion subject to internal pressure.

8.2. Problems with comparing the methods

There is insufficient data in the published literature to do a thorough comparison of the methods for assess-
ing corrosion. If there were enough detailed data, then the first step in a comparison would be burst tests of
artificial, flat-bottomed corrosion defects, to avoid scatter associated with approximations to an irregular pro-
file. The approach would be to consider those tests which are known to have failed by plastic collapse (i.e. the
flow stress or reference stress (defect-free failure stress) is equal to the ultimate tensile strength). Define an
appropriate failure criterion (as has been done for DNV-RP-F101 and PCORRC). Then identify those tests
which do not follow the predictions of the criterion. Finally, determine what is different about these outliers
and thence define the limitations of the failure criterion. Only then would the methods be compared against
burst tests of real corrosion defects.

8.3. Comparisons of methods in the published literature

Several reviews or comparisons of methods for assessing corrosion defects are described in the published
literature. The Linepipe Corrosion Group Sponsored Project and the DNV Joint Industry Project both con-
ducted a review of existing assessment methods as part of the development of an improved method [22,23].10
Battelle have also reviewed methods for assessing corrosion [24]. Other authors have conducted limited com-
parisons of methods with test data during the course of the development and validation of new or modified
assessment methods. The conclusions of the various reviews are:

10
A PRCI sponsored project is being conducted to further compare existing assessment methods for corrosion subject to internal
pressure and to clarify issues surrounding the behaviour of blunt defects in low and moderate to high toughness line pipe steels. The results
of this study are not currently in the public domain.
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1259

1. Recently developed methods such as DNV-RP-F101 and PCORRC are based on equations fitted to the
results of a large number of finite element analyses of blunt, part-wall defects, these analyses incorporated
a failure criterion validated against actual burst tests. The DNV-RP-F101 and PCORRC methods were
developed to be mean fits to the experimental and numerical data, and so should be the most accurate meth-
ods. This is the consensus view of the reviews in the literature [22–24] (see Fig. 7 and note the accurate pre-
dictions of the artificial corrosion defects11).
2. The modified B31G method is more accurate than the original ASME B31G method [14,24].
3. RSTRENG gives a further improvement in accuracy [14,22,24].
4. The more recent methods, such as DNV-RP-F101 and PCORRC, are only applicable to blunt defects in
tough materials [23,24].

The ASME B31G method (or modified B31G and RSTRENG) for predicting the burst pressure of a cor-
roded pipeline (the ‘old’ methods) were, predominantly, developed and validated through full scale tests on
older line pipe steels. The ‘new’ methods (DNV RP-F101 and the pipeline specific appendix of BS 7910,
and PCORR) were developed and validated through tests on modern, high toughness, line pipe steels. The
‘new’ methods are biased towards the behaviour of modern, high toughness, line pipe steels and the ‘old’ meth-
ods are biased towards older, relatively lower toughness, steels. The difference between the behaviour of older
line pipe steels and modern steels can largely be attributed to the general increase in the toughness of line pipe,
due to improvement in steel production and technological advances. The ‘old’ methods demonstrate greater
scatter than the ‘new’ methods when compared to the (relevant) published full-scale test data; the ‘new’ meth-
ods are more accurate.

9. Recommendations in PDAM

The recommendations in PDAM for assessing the burst strength of a corrosion defect (considering depth
and longitudinal length) are:

1. DNV-RP-F101 for moderate to high toughness line pipe, and

2. Modified B31G and RSTRENG in older, lower grade line pipe, and when there is no confidence that the
requirements for the application of the more recent methods are satisfied.

Moderate to high toughness line pipe is defined as:

(i) modern (clean) line pipe with a 2/3 thickness specimen size upper shelf Charpy V-notch impact energy
equal to at least 18 J (13 ftlbf) (the full size equivalent is 27 J (20 ftlbf)),
(ii) meeting the minimum elongation requirements in API 5L [58], and
(iii) excluding line pipe steels suspected of containing a significant number of inclusions, second phase par-
ticles or other contaminants; typically, this means lower grade line pipe (such as grades A and B) and
other older line pipe.

Note that none of the methods have been proven in line pipe with a wall thickness greater than 25.4 mm.

10. Assessing a corroded area

10.1. The assessment method

The best methods for assessing a corrosion defect (considering depth and longitudinal length) in a pipeline
subject to internal pressure have been identified, and their limitations highlighted in Section 9.

11
The AGA/PRCI tests include tests of older, lower grade line pipe steels, hence some of the non-conservative predictions.
1260 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

CORRODED PIPELINE

Indications of low toughness include: old

line pipe, line pipe not manufactured to DOES THE LINE YES
API 5L (or equivalent), or an operating PIPE HAVE A LOW SEEK SPECIALIST ADVICE
temperature less than the DWTT TOUGHNESS?
transition temperature.
NO

MAXIMUM DEPTH
(d) GREATER YES
THAN 0.85xWALL SEEK SPECIALIST ADVICE
THICKNESS (t)

NO

IS THE YES
CORROSION SEE CHAPTER 26
IN A DENT?

NO

IS THE CORROSION
ASSOCIATED WITH YES
SEE CHAPTER 34
OTHER DAMAGE OR
DEFECTS?

NO

NO
IS THE
CORROSION ON
A WELD OR HAZ?

YES

IS THE WELD YES

Indications of low toughness include: old SEEK SPECIALIST ADVICE
UNDER-MATCHED?
line pipe, line pipe not manufactured to
API 5L, welds not fabricated to a
recognised pipeline welding standard, NO
poor quality welds, low frequency ERW
IS THE WELD
or IW seam welds, oxyacetylene girth YES
SUSPECTED OF
welds, and an operating temperature less SEEK SPECIALIST ADVICE
HAVING A LOW
than the DWTT transition temperature.
TOUGHNESS
If the upper shelf Charpy V-notch
impact energy is less than 30 J NO
(minimum), 40 J (average) then the weld
should be regarded as having a low
toughness.
IS THE PIPELINE YES
PRESSURE SEE SECTION 20.18
CYCLED?

NO

IS THE WALL
THICKNESS YES
SEEK SPECIALIST ADVICE
GREATER THAN
25.4 mm?

NO

Fig. 8. The Assessment of a Corrosion Defect in a Pipeline (note that section numbers refer to the PDAM document section numbering).
A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1261

NO

YES IS THE LINE PIPE

GRADE A OR
GRADE B?

NO

IS THE 2/3-SIZE
NO CHARPY V-NOTCH
IMPACT ENERGY OF
THE LINE PIPE
GREATER THAN 18J?

YES

ARE THE MINIMUM

NO
ELONGATION
NO
REQUIREMENTS IN
API 5L SATISFIED?

YES

IS THE CORROSION YES

ARBITRARILY SEE SECTION 20.17
ORIENTATED?

NO

IS THE CORROSION YES

CIRCUMFERENTIALLY SEE SECTION 20.14 TO 20.16
ORIENTATED?

NO

IS THE CORROSION
LONGITUDINALLY
ORIENTATED?

YES

IS THE CORROSION
SUBJECT TO LOADS OTHER YES
THAN INTERNAL OR
SEE SECTION 20.13
EXTERNAL PRESSURE?

NO

SEE SECTION 20.12.2.2

MAXIMUM DEPTH
YES
(d) GREATER
SEEK SPECIALIST ADVICE
THAN 0.80xWALL
THICKNESS (t)

NO

IS THE CORROSION YES

ARBITRARILY
NO SEE SECTION 20.17
ORIENTATED?

NO

Fig. 8 (continued)
1262 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

NO

IS THE CORROSION YES

CIRCUMFERENTIALLY SEE SECTION 20.14 TO 20.16
ORIENTATED?

NO

IS THE CORROSION
LONGITUDINALLY
ORIENTATED?

YES

IS THE CORROSION
SUBJECT TO LOADS OTHER YES
THAN INTERNAL OR SEEK SPECIALIST ADVICE
EXTERNAL PRESSURE?

NO

SEE SECTION 20.12.2.1

Fig. 8 (continued)

10.2. The assessment procedure

The flowchart in Fig. 8 provides a general overview of the issues that need to be considered when assessing
an area of corrosion in a pipeline, and identifies the appropriate method to be used. The flowchart does not
give practical guidance of how to conduct the assessment.

1.0

corrosion in a single features

0.9
girth weld
interacting features
0.8
features requiring further
coincident internal and
d/t (normalised defect depth)

investigation
0.7 external corrosion

0.6

0.5
UNACCEPTABLE

0.4

0.3

ACCEPTABLE
0.2
corrosion in a section of pipeline subject to large external loads
0.1
metal loss associated with a dent
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

2c/(Rt)^0.5 (normalised longitudinal defect length)

Fig. 9. Illustration of the assessment of corrosion in a pipeline, considering single and interacting features, and then identifying other
issues that may influence the acceptability of the defects and require further investigation.
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1263

What follows is some practical guidance that can be applied to assessing an area of corrosion: it can be
applied to direct measurements obtained from excavating and inspecting a pipeline by hand, or to the results
of an intelligent pig run. The approach adopted is to use the most conservative geometric idealisation to deter-
mine if the defect(s) are acceptable. These assumptions are then revisited and revised systematically to move
from a conservative assessment to a more accurate (but still conservative) assessment. The approach can be
applied to any suitable assessment method. DNV RP-F101 and modified B31G specify an acceptance crite-
rion, providing a necessary safety margin between failure and acceptance.
Specified minimum material properties (as given in the line pipe steel specification) and the specified min-
imum wall thickness should be used. The longitudinal and circumferential dimensions of the defect are defined
by a projection in the respective transverse direction. Inspection tolerances should be added to all defect
dimensions.
The corrosion is assumed to be in undented pipe and away from any welds. The pipeline is assumed to be
subject to only internal pressure loading. The longitudinal extent of the corrosion is likely to be more impor-
tant that the circumferential extent. In general, however, the various steps of the assessment should be applied
to both the longitudinal and circumferential extent of the corrosion.
These assumptions simplify the assessment procedure, because it is only necessary to consider single defects
and interacting defects. However, in practice corrosion coincident with dents, welds or other defects, and
external loads, must be considered to complete the assessment, since they can lead to a very different picture
of the significance of the defects (see Fig. 9).
The corrosion is assumed to be inactive. The guidance is also applicable to active corrosion, except that it is
also necessary to consider the implications of defect growth. It is important to establish the cause of any cor-
rosion in a pipeline.

Screening
(1) Identify the critical defects (i.e. depth greater than 80% of the wall thickness, failure pressure less than
the maximum operating pressure). This assessment assumes that all defects are single defects, it does not
take account of defect interaction. This is non-conservative; therefore the assessment cannot stop at this
stage, it is simply a ranking process to focus attention on the main threats.
Interaction
(2) Determine whether the defect(s) can be considered as a single feature or as part of a group of inter-
acting features.A number of different interaction rules have been described in the literature. One com-
monly used rule is that adjacent defects are considered to interact if the spacing (in the longitudinal or
circumferential direction) between the defects is less than the respective dimension (i.e. length or
width) of the smaller defect.12 The depth of the composite defect is defined by the maximum depth,13
and the length and width by the dimensions of an enveloping rectangleIt is always conservative to
assume that all of a cluster of adjacent defects interact. The dimensions of the composite defect are
defined as above.
Assessment
(3) Assess the single defect(s).
(4) Assess the interacting defect(s), using the dimensions of the composite defect(s).
Review
(5) Consider more accurate assessment methods (less conservative) interaction rules, a river-bottom profile,
etc. for those defect(s) which are not acceptable. Alternatively, repair the defect or downrate the pipeline.

References

[1] Bolt R, Owen RW. Recent trends in gas pipeline incidents (1970–1997): a report by the European Gas Pipeline Incidents Data
Group (EGIG), Paper C571/032/99, Ageing pipelines, optimising the management and operation: low pressure–high pressure,

12
This interaction rule is based on linear elastic fracture mechanics. It is to be found in documents such as BS 7910 [4].
13
Assuming that all of the defects are on a coincident surface.
1264 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265

IMechE Conference Transactions 1999–8 (C571), Institution of Mechanical Engineers, Newcastle upon Tyne, UK, 11–13
October; 1999.
[2] Eiber RJ, Miele CR, Wilson PR. An analysis of DOT reportable incidents on gas transmission and gathering lines for June 1984
Through 1992, In: Topical report to line pipe research supervisory committee of the pipeline research committee of the American gas
association, NG-18 Report No. 213, American Gas Association, July 1995.
[3] http://ops.dot.gov/stats.htm.
[4] Anon. Guide on methods for assessing the acceptability of flaws in fusion welded structures. London (UK): British Standards
Institution; 2005 [BS 7910:2005].
[5] Anon. Managing system integrity for hazardous liquid pipelines, API Standard 1160 (ANSI/API STD 1160-2001), 1st ed.; November
2001.
[6] Lewis K. Integrity management of pipelines. Congreso Internacional de Ductos (International Pipeline Congress). Mexico: Mérida,
Yucatán; 2001 (14–16 November).
[7] Kiefner JF, Maxey WA, Eiber RJ, Duffy AR. The failure stress levels of flaws in pressurised cylinders. ASTM STP, vol. 536.
Philadelphia: American Society for Testing and Materials; 1973. p. 461–81.
[8] Leis BN, Brust FW, Scott PM. Development and validation of a ductile flaw growth analysis for gas transmission line pipe, Final
Report to A.G.A. NG-18, Catalog No. L51543; June 1991.
[9] Leis BN, Ghadiali ND. Pipe axial flaw failure criteria – PAFFC, Version 1.0 users manual and software, Topical Report to A.G.A.
NG-18, Catalog No. L51720; May 1994.
[10] Knauf G, Hopkins P. The EPRG Guidelines on the assessment of defects in transmission pipeline girth welds, 3R International, 35,
Jahrgang, Heft, 10–11/1996, p. 620–4.
[11] Roovers P, Bood R, Galli M, Marewski U, Steiner M, Zaréa M. EPRG Methods for assessing the tolerance and resistance of
pipelines to external damage, pipeline technology, In: Proceedings of the third international pipeline technology conference, Brugge,
Belgium, 21–24 May 2000, R. Denys, Ed., Elsevier Science, 2000, pp. 405–425.
[12] Hopkins P. The application of fitness for purpose methods to defects detected in offshore transmission pipelines, In: Conference on
welding and weld performance in the process industry, London; 27–28 April 1992.
[13] Anon; DNV-RP-F101, Corroded Pipelines, Det Norske Veritas, 1999.
[14] Kiefner JF, Vieth PH. A modified criterion for evaluating the strength of corroded pipe, Final Report for Project PR 3-805 to the
Pipeline Supervisory Committee of the American Gas Association, Battelle (OH); 1989.
[15] Kastner W, Rohrich E, Schmitt W, Steinbuch R. Critical crack sizes in ductile piping. Int J Pressure Vessels Piping 1981;9:
197–219.
[16] Schulze HD, Togler G, Bodman E. Fracture mechanics analysis on the initiation and propagation of circumferential and longitudinal
defects in straight pipes and pipe bends. Nucl Eng Des 1980;58:19–31.
[17] Anon. Fitness-for-service, API Recommended Practice 579, 1st ed., American Petroleum Institute; January 2000.
[18] Cosham A, Hopkins P. The pipeline defect assessment manual, IPC02-27067, In: Proceedings of IPC 2002, international pipeline
conference, American Society of Mechanical Engineers, Calgary, Alta., Canada; 2002.
[19] Anon, Specifications and requirements for intelligent pig inspection of pipelines, version 2.1, Pipeline Operators Forum; 6 November
1998.
[20] Stephens DR, Bubenik TA, Francini RB. Residual strength of pipeline corrosion defects under combined pressure and axial loads,
Final Report to Line Pipe Research Supervisory Committee of the Pipeline Research Committee of the American Gas Association,
NG-18 Report No. 216, A.G.A. Catalog No. L51722, Battelle Memorial Institute; February 1995.
[21] Stephens DR. Leis BN. Development of an alternative criterion for residual strength of corrosion defects in moderate- to high-
toughness pipe, In: Proceedings of the third international pipeline conference (IPC 2000), vol. 2, Calgary, Alta., Canada, American
Society of Mechanical Engineers, 1–5 October 2000, pp. 781–92.
[22] Fu B. Advanced engineering methods for assessing the remaining strength of corroded pipelines, ageing pipelines, optimising the
management and operation: low pressure – high pressure, In; IMechE conference transactions 1999–8 (C571), Institution of
Mechanical Engineers, Newcastle upon Tyne, UK, 11–13 October 1999.
[23] Bjørnøy OH. Marley M. Assessment of corroded pipelines/past, present and future, In: Eleventh international conference on offshore
and polar engineering (ISOPE 2001), Stavanger (USA); June 2001.
[24] Stephens DR, Francini RB. A review and evaluation of remaining strength criteria for corrosion defects in transmission pipelines, In:
Proceedings of ETCE/OMAE 2000 joint conference on energy for the new millennium, New Orleans (LA, USA); 14–17 February 2000.
[25] Barkow AG. Don’t bet on a pit, Materials protection and performance, vol. 11 (10), October 1972, p. 11–7.
[26] Marvin CW. Determining the strength of corroded pipe. Mater Perform 1972;11(11):34–40.
[27] Kiefner JF, Duffy AR. Criteria for determining the strength of corroded areas of gas transmission lines, In: Paper T, American gas
association operating section on transmission conference, AGA; 1973. p. T86–91.
[28] Shannon RWE. The failure behaviour of line pipe defects. Int J Pressure Vessel Piping 1974;2:243–55.
[29] Mok DRB, Pick RJ, Glover AG. Behaviour of line pipe with long external corrosion. Mater Perform 1990;29(5):75–9.
[30] Mok DRB, Pick RJ, Glover AG, Hoff R. Bursting of line pipe with long external corrosion. Int J Pressure Vessels Piping
1991;46:195–216.
[31] Coulson KEW, Worthingham RG. Pipe corrosion-1: standard damage assessment approach is overly conservative. Oil Gas J 1990.
April 9th.
[32] Coulson KEW, Worthingham RG. Pipe corrosion-conclusion: new guidelines promise more accurate damage assessment. Oil Gas J
1990. April 16th.
 A. Cosham et al. / Engineering Failure Analysis 14 (2007) 1245–1265 1265

[33] Hopkins P, Jones DG. A study of the behaviour of long and complex-shaped corrosion in transmission pipelines. In: Proceedings of
eleventh international conference on offshore mechanics and Arctic engineering (OMAE 1992). Calgary (Canada): American Society
of Mechanical Engineers; 1992.
[34] Vieth PH. Kiefner JF. Database of Corroded Pipe Tests, Final Report on Contract No. PR 218-9206 to Line Pipe Research
Supervisory Committee of the American Gas Association, Kiefner and Associates, Inc.; April 1994.
[35] Kiefner JF, Vieth PH, Roytman I. Continued Validation of RSTRENG, Updated Draft Final Report on Contract No. PR 218-9304
to Line Pipe Research Supervisory Committee, Pipeline Research Committee of the American Gas Association, Kiefner and
Associates, Inc.; 31 March 1995.
[36] Chouchaoui BA, Pick RJ. Residual burst strength of pipe with internal corrosion pits. In: Proceeding of the international conference
on pipeline reliability, Calgary, Canada; June 1992.
[37] Chouchaoui BA, Pick RJ. Behaviour of isolated pits within general corrosion. Pipes Pipeline Int 1994(January–February):12–20.
[38] Chouchaoui BA, Pick RJ. Behaviour of circumferentially aligned corrosion pits. Int J Pressure Vessels Piping 1994;57:187–200.
[39] Chouchaoui BA, Pick RJ. Behaviour of longitudinally aligned corrosion pits. Int J Pressure Vessels Piping 1996;67:17–35.
[40] Fu B, Batte AD. Advanced methods for the assessment of corrosion in linepipe. Health and Safety Executive Summary Report, OTO
1999-051, HSE Books; 1999.
[41] Nehoda J, Horalek V. Long term experiences with the maintenance of high pressure pipelines at transgas s.p.. In: Third international
conference on pipeline rehabilitation & maintenance, Prague, Czech Republic; 4–7 September 2000.
[42] Benjamin AC, Vieira RD, Freire JLF, De castro JTP. Burst tests on pipeline with long external corrosion. Proceedings of the third
international pipeline conference (IPC 2000), vol. 2. Calgary (Alta., Canada): American Society of Mechanical Engineers; 2000. p.
793–9.
[43] Cronin DS, Pick RJ. Experimental database for corroded pipe: evaluation of RSTRENG and B31G. Proceedings of the third
international pipeline conference (IPC 2000), vol. 2. Calgary (Alta., Canada): American Society of Mechanical Engineers; 2000. p.
757–67.
[44] Cronin DS, Pick RJ. A new multi-level assessment procedure for corroded line pipe. Proceedings of the third international pipeline
conference (IPC 2000), vol. 2. Calgary (Alta., Canada): American Society of Mechanical Engineers; 2000. p. 801–8.
[45] Cronin DS, Roberts KA, Pick RJ. Assessment of long corrosion grooves in line pipe. Proceedings of the first international pipeline
conference (IPC 1996), vol. 1. Calgary (Alta., Canada): American Society of Mechanical Engineers; 1996.
[46] Stewart G, Klever FJ, Ritchie D. An analytical model to predict the burst capacity of pipelines. Pipeline technology, international
conference on offshore mechanics and arctic engineering (OMAE 1994), vol. V. American Society of Mechanical Engineers; 1994. p.
177–88.
[47] Jones DG, Turner T, Ritchie D. Failure behaviour of internally corroded linepipe. In: Proceedings of eleventh international
conference on offshore mechanics and arctic engineering (OMAE 1992). American Society of Mechanical Engineers; 1992.
[48] Roberts KA, Pick RJ. Correction for longitudinal stress in the assessment of corroded line pipe. In: Proceedings of second
international pipeline conference (IPC 1998). Calgary (Alta., Canada): American Society of Mechanical Engineers; 1998. p. 553–61.
[49] Bjørnøy OH, Sigurdsson G, Cramer EH. Residual strength of corroded pipelines, DNV test results. In: Tenth international
conference on offshore and polar engineering (ISOPE 2000), Seattle, USA; 28 May–2 June 2000.
[50] Leis BN, Stephens DR. An alternative approach to assess the integrity of corroded line pipe – part I: current status. In: Proceedings of
the seventh (1997) international offshore and polar engineering conference, Honolulu, USA; 25–30 May 1997. p. 624–34.
[51] Leis BN, Stephens DR. An alternative approach to assess the integrity of corroded line pipe – part II: alternative criterion. In:
Proceedings of the seventh (1997) international offshore and polar engineering conference, Honolulu, USA; 25–30 May 1997. p. 635–
41.
[52] Fearnehough GD, Lees GM, Lowes JM, Weiner RT. The role of stable ductile crack growth in the failure of structures. In: Paper 33,
conference on practical application of fracture mechanics to pressure vessel technology. Institution of Mechanical Engineers; 1971. p.
119–28.
[53] Fearnehough GD, Jones DG. An approach to defect tolerance in pipelines. In: Paper C97/78, conference on tolerance of flaws in
pressurised components. London, UK: Institution of Mechanical Engineers; 16–18 May 1978. p. 179–92.
[54] Klever FJ. Burst strength of corroded pipe: flow stress revisited. In: Paper OTC 7029, 24th annual offshore technology conference.
Houston, Texas, USA; 4–7 May 1992. p. 417–31.
[55] Ritchie D, Last S. Burst criteria of corroded pipelines – defect acceptance criteria. In: Paper 32, proceedings of the EPRG/PRC 10th
biennial joint technical meeting on line pipe research. Cambridge, UK; 18–21 April 1995. p. 32-1–32-11.
[56] Anon. Manual for determining the remaining strength of corroded pipelines. In: A Supplement to ASME B31 code for pressure
piping, ASME B31G-1991 (Revision of ANSI/ASME B31G-1984), New York, USA: The American Society of Mechanical Engineers;
1991.
[57] Wang W, Smith MQ, Popelar CH, Maple JA. A new rupture prediction model for corroded pipelines under combined loading. In:
Proceedings of second international pipeline conference (IPC 1998). Calgary (Alta., Canada): American Society of Mechanical
Engineers; 1998.
[58] Anon. Specification for Line Pipe, Exploration and Production Department, API Specification 5L, American Petroleum Institute,
Forty Second Edition, 1 April 2000.