This document was downloaded from the Penspen Integrity Virtual Library

For further information, contact Penspen Integrity:

Penspen Integrity
Units 7-8
St. Peter's Wharf
Newcastle upon Tyne
NE6 1TZ
United Kingdom

Telephone: +44 (0)191 238 2200

Fax: +44 (0)191 275 9786
Email: [email protected]
Website: www.penspenintegrity.com

Proceedings of IPC 2002:

International Pipeline Conference
29 September - 3 October, 2002; Calgary, Alberta, Canada

IPC02-27067
THE PIPELINE DEFECT ASSESSMENT MANUAL
Andrew Cosham,
Phil Hopkins
[email protected]
[email protected]
Penspen Andrew Palmer, Newcastle Business Park, Newcastle upon Tyne, NE4 7YL, UK
Tel: +44 (0)191 273 2430, Fax: +44 (0)191 273 2405

ABSTRACT
Oil and gas transmission pi pelines h ave a g ood s afety
record. This is due to a com bination of good design, materials
and operatin g practices . How ever, lik e an y en gineering
structure, pipelines do occas ionally f ail. T he m ajor cau ses of
pipeline failures around the world are external interference and
corrosion; th erefore, as sessment m ethods are needed to
determine the severity of such defects when they are detected in
pipelines.
Defects occu rring du ring th e f abrication of a pipelin e are
usually assessed ag ainst reco gnised an d p roven q uality co ntrol
(workmanship) limits. These workmanship limits are somewhat
arbitrary, but they h ave been prov en ov er tim e. How ever, a
pipeline w ill in variably co ntain lar ger defects at some stage
during its life, and th ese w ill req uire a fitness-for-purpose
assessment to d etermine w hether o r n ot to repair the pipeline.
Consequently, the past 40 y ears has seen a large number of full
scale tests of def ects in pipelin es, an d th e dev elopment of a
number of m ethods f or as sessing the significance of defects.
Some of th ese m ethods h ave been in corporated in to in dustry
guidance, o thers are to b e f ound in th e published literature.
However, there is no definitive guidance that draws together all
of th e as sessment tech niques, or assesses each method against
the p ublished test d ata, o r reco mmends b est p ractice in th eir
application.
To ad dress this industry need, a J oint Industry P roject has
been sp onsored b y fifteen international oil and gas companies1
to d evelop a P ipeline Def ect A ssessment Manual (PDAM).
PDAM docu ments th e bes t av ailable tech niques currently
available f or th e as sessment of pipelin e def ects (such as
corrosion, den ts, g ouges, w eld def ects, etc.) in a simple and
easy-to-use manual, and gives guidance in their use. PDAM is
based on an ex tensive critical rev iew of pipeline fitness-for1

Advantica Technologies, BP, CSM, DNV, EMC, G az de France, Health and

Safety Executive, MOL, Petrobras, PI I, SNA M R ete G as, Sh ell G lobal
Solutions, Statoil, Toho Gas and TotalFinaElf.

purpose methods an d pu blished t est dat a. It i s i ntended t o be

another to ol to h elp p ipeline en gineers maintain th e h igh lev el
of pipeline safety.
In ad dition to id entifying th e b est m ethods, PDAM has
served to identify a n umber o f lim itations in th e cu rrent
understanding of the behaviour of defects in pipelines, and the
empirical lim its in th e ap plication o f ex isting m ethods. T his
paper discusses the PD AM proj ect, i n t he con text of bot h t he
current bes t practice av ailable f or def ect as sessment and the
limitations of current knowledge.
1. INTRODUCTION
The most common causes of dam age an d f ailures in
onshore an d of fshore, oi l an d g as t ransmission pi pelines i n
Western Eu rope an d North A merica are external interference
(mechanical damage) an d corros ion. A ccordingly, th e
behaviour of def ects i n pi pelines h as been the subject of
considerable study over the past 40 y ears, with a large number
of f ull s cale tes ts, an alyses an d oth er work having been
undertaken. M any d ifferent fitness-for-purpose methods have
been developed.
Fitness-for-Purpose. F itness-for-purpose, as discussed
here, means that a particu lar s tructure is con sidered to be
adequate for its p urpose, p rovided th e co nditions to reach
failure are n ot reach ed[1]. No te th at f itness-for-purpose m ay
also have a legal and contractual meaning in different countries.
Fitness-for-purpose is based on a det ailed technical assessment
of the significance of the defect. Local and national legislation
and reg ulations m ay n ot p ermit certain ty pes o f d efects to b e
assessed by f itness-for-purpose m ethods or may mandate
specific limits. Su ch issues should always be considered prior
to an assessment.
Safety m ust alw ays b e th e p rime consideration in any
fitness-for-purpose as sessment. It
is always necessary to
appreciate the co nsequences o f a f ailure. T hese will influence
the necessary safety margin to be applied to the calculations.

Pipeline Integrity Management. P ipeline f ailures are

usually related to a b reakdown in a system, e.g. the corrosion
protection system h as b ecome f aulty, an d a co mbination o f
ageing coating, a ggressive e nvironment, a nd r apid c orrosion
growth may lead to a co rrosion failure. This type of failure is
not simply a corrosion failure, but a corrosion control system
failure. Sim ilar o bservations can b e d rawn f or f ailures due to
external interference, stress corrosion cracking, etc..
These considerations lead to the conclusion that a holistic
approach to p ipeline d efect assessm ent an d in tegrity is
necessary; understanding the equation that quantifies the failure
load is only one aspect.
Pipeline integrity management is th e general term given to
all efforts (design, co nstruction, o peration, m aintenance, etc.)
directed to wards en suring co ntinuing p ipeline integrity. The
American P etroleum In stitute (API) h as d eveloped an in dustry
consensus standard that gives guidance on developing integrity
management programmes (API 1160)[2]. The American Society
of Mechanical Engineers (A SME) is als o dev eloping an
integrity management appendix for ASME B31.8[3].
The Pipeline Defect Assessment Manual. T he P ipeline
Defect Assessment Manual (PDAM) presents a considered view
of th e best cu rrently av ailable m ethods for assessing the
fitness-for-purpose of defects in pi pelines. It i s bas ed on a
critical review of the published fitness-for-purpose methods and
test data. P DAM intended to be a d ocument that will assist in
maintaining pipeline in tegrity. T he P DAM p roject is d ue f or
completion i n August 2002. PDAM will be m ade available to
the pipeline industry.
This paper s ummarises t he m ethodology an d g ives an
outline of th e co ntents o f P DAM. T he b est m ethods f or
assessing a v ariety of different types of defect are summarised
(see Table 3). E mpirical t oughness l imits d erived fr om
published test data are g iven an d th e as sessment of ex ternal
interference (dents and gouges) is described in more detail. The
PDAM recom mendations f or th e as sessment of oth er ty pes of
defect will be described in future papers.
NOMENCLATURE
2c
d
t
A
CV
D
E
H
Ho
Hr
K1
K2

length of part-wall metal loss defect (mm)

depth of part-wall metal loss defect (mm)
pipe wall thickness (mm)
fracture area of a 2/3 Charpy specimen (53.55 mm2 for
a 2/3 Charpy specimen) (mm2)
2/3 t hickness sp ecimen up per she lf Charpy V-notch
impact energy (J)
outside diameter of pipe (mm)
Youngs modulus (207,000 Nmm-2)
dent depth (mm)
dent depth measured at zero pressure (mm)
dent depth measured at pressure (mm)
non-linear regression parameter
non-linear regression parameter

Y
U

outside radius of pipe (mm)

flow stress (Nmm-2)
hoop stress at failure (Nmm-2)
yield strength (Nmm-2)
ultimate tensile strength (Nmm-2)

2. FITNESS
FOR
PURPOSE,
ENGINEERING
CRITICAL ASSESSMENTS (ECAs) AND PIPELINES
The f itness-for-purpose of a def ect i n a pi peline may be
determined by a v ariety of m ethods ran ging f rom prev ious
relevant experience (including workmanship acceptance levels),
to model testing, to engineering critical assessm ents (ECAs),
where a defect is appraised analytically.
2.1 GENERIC
Various technical procedures are available for assessing the
significance of defects in a ran ge of structures. T hese methods
use a com bination of fracture mechanics and limit state (plastic
collapse) methods. Bot h BS 7910 : 1999 [1] and API RP 579 [4]
contain detailed engineering critical as sessment methods which
can be applied to defects i n p ipelines ( although t he l atter
document is biased towards defects in process plant).
2.2 PIPELINE-SPECIFIC
Documents such as th e abov e are g eneric; th ey can be
conservative w hen applied to s pecific structures such as
pipelines. T herefore, t he pi peline i ndustry h as dev eloped i ts
own f itness-for-purpose m ethods ov er t he pas t 40 years (and,
indeed, docu ments s uch as BS 7910 recom mend that such
methods be u sed). T hese pipeline specific methods are u sually
based o n ex periments, so metimes w ith limited theoretical
validation; they are semi-empirical methods. Consequently, the
methods may become in valid if th ey are ap plied o utside th eir
empirical limits. Accordingly, PDAM has considered the limits
of th e ex perimental v alidation of com monly u sed pipelin e
specific methods.
Methods and guidelines developed by the pipeline industry
range fr om t he N G-18 e quations[5] (which f ormed th e bas is of
methods s uch as A SME B31G [6] an d R STRENG[7]) an d th e
Ductile Flaw Growth Model (DFGM) (im plemented as PAFFC
(Pipe Axial Flaw Failure Criteria))[8,9] developed by the Battelle
Memorial In stitute in th e USA o n b ehalf o f th e P ipeline
Research Council International (PRCI), to the guidelines for the
assessment o f gir th weld d efects[10], mechanical damage[11] and
ductile f racture p ropagation[12] produ ced by t he European
Pipeline Research Group (EPRG).
The conservatism of generic methods compared to pipeline
specific methods can largely be attributed to issues of constraint
and ductile tearing. Co nstraint is th e restriction of plastic flow
in th e v icinity o f th e crack tip d ue to stress triaxiality. Stress
triaxiality is in duced b y lo ad an d g eometry. T he stan dard test
methods use d t o m easure fr acture toughness are designed to
give co nditions o f h igh co nstraint at th e crack tip to ensure
conservative res ults. P ipelines h ave low con straint becau se
they are th in walled (g eometry) an d are predom inantly s ubject
to membrane tensile loading ( loading m ode). Co nventional

(single param eter) f racture m echanics does not consider the

elevation in fracture toughness due to a reduction in the level of
constraint, and hence an in herent m argin of s afety is in cluded
when applied to low constraint structures. T he semi-empirical
pipeline specific methods consider constraint implicitly because
they h ave b een d eveloped f rom f ull scale tests in w hich th ese
effects manifest themselves directly. Sim ilarly, th e in crease in
toughness with d uctile c rack gr owth ( a r ising r esistance c urve)
is also co nsidered im plicitly. The difference between pipeline
specific an d g eneric m ethods dim inishes when sophisticated
fracture m echanics (tw o-parameter f racture m echanics, tearin g
analysis, etc.) and limit state methods are applied.
2.3 A BRIEF HISTORY OF PIPELINE DEFECT
ASSESSMENT
i. The Early Days.
Fracture mechanics is the science of why things fail. T he
effect of defects on structures was studied qualitatively as lo ng
ago as the 15 th century by Leonardo da Vinci; he measured the
strength of lengths of iro n wire, illu strating th e effect o f f laws
on strength and observing t hat sho rt w ires w ere st ronger t han
long w ires (d ue to th e lo wer p robability o f th e sh orter wire
containing a defect). Notch ed bar im pact tes ting of iron an d
steel was w idely u sed by th e en d of th e 19 th cen tury to
determine ductile to brittle transition temperatures[13].
In 1920, Grif fith pu blished a qu antitative relation ship
between the fracture stress and th e s ize of a f law, deriv ed in
terms of a s imple en ergy balan ce f rom a stress analysis of an
elliptical hole by Inglis and the First Law of Thermodynamics.
However, the work of Griffith was only applicable to perfectly
elastic m aterials (b rittle m aterials) an d efforts to apply the
theory to metals were initially not successful.
ii. The Start.
Prior to circa 1950, failure reports of engineering structures
did not usually con sider th e pres ence of crack s. C racks w ere
considered unacceptable in term s o f q uality, b ut w ere n ot
considered q uantitatively. T here w ere ex ceptions: th e Liberty
Ship failures (during the Secon d W orld W ar) are com monly
cited as one of the prime instigators for the further development
of the science of fracture mechanics.
In t he 1950s t here w as m ajor i nterest i n f racture i n t he
aircraft industry in th e USA, p articularly in alu minium, an d in
the 1960s there was an increased interest in fracture in nuclear
power plan ts. T his lead to th e dev elopment of f racture
mechanics using various approaches (stress intensity factor (K),
J-integral and crack tip open ing displacement ( )). T he 1950s
and 1960s was al so a peri od w here t he s afety of t ransmission
pipelines was of interest, primarily in the USA due to its lar ge
and aging pipeline system.
iii. The Pipeline Pioneers.
Workers at th e B attelle Mem orial In stitute in Co lumbus,
Ohio extensively studied the failure of defects in line pipe steel
through both theoretical work a nd ful l sc ale t esting, und er t he
auspices of th e th en P ipeline R esearch C ommittee of th e
American Gas A ssociation. T he prin cipal obj ective of th is

early w ork w as to p rovide a so und and quantitative technical

understanding of the relationship b etween th e h ydrostatic test
level and the number and size of defects removed. The concept
of the flow stress was introduced and a correction for plasticity
at the crack tip, requ ired w hen apply ing lin ear-elastic f racture
mechanics theory to elastic- plastic m aterials, w as
proposed[14,15].
The researchers noted that defects in line pipe tended to fail
in a d uctile m anner, b ut th at tw o basic distinctions could be
made:
1. Toughness d ependent fa ilures t o p redict t he fa ilure
stress of these tests a measure of the fracture toughness was
required (th e critical stress intensity factor, Kc, or an
empirical correlation with the upper shelf Charpy V-notch
impact energy).
2. Flow stress d ependent ( plastic co llapse) f ailures to
predict the failure stress of these tests only a measure of the
strength of the material was required.
The work at Battelle led to th e d evelopment o f th e f low
stress dependent and the toughness dependent, through-wall and
part-wall NG- 18 eq uations[5]. A su mmary o f th e test d ata an d
the transition from toughness to flow stress dependent failure is
given in Fig . 1. T he u nderlying ex pressions an d con cepts are
still widely used today.
The orig inal w ork an d m odels accou nted f or th e very
complex failure process of a d efect in a p ipeline, in volving
bulging o f th e p ipe w all, p lastic flow, crack initiation and
ductile te aring, a lthough m uch o f this is im plicit a nd follows
from the semi-empiricism. T hese pioneering models were safe
due to inherently conservative assumptions and verification via
full scale testing, b ut th ey are lim ited b y th e ran ge o f th e
experiments (generally, thin walled, lo wer g rade, lo w y ield to
tensile ratio line pipe). T he DFGM, d eveloped b y B attelle in
the early 1990s, is a revision and update of the original NG-18
equations a nd b etter d escribes t he si gnificance o f t oughness,
ductile tearing and plastic collapse[8,9].
iv. The Future.
Recent w ork h as sh own th ese o ld m ethods to still be
applicable to many newer p ipeline ap plications, b ut th ere h as
been a h eavy relian ce on ex periments and, more recently,
numerical an alysis. W ith s ome n otable ex ceptions, th ere h as
been little f undamental w ork rep orted, an d th is is a m ajor,
serious an d s omewhat pu zzling om ission. There has been a
focus on dev eloping patches t o existing methods, and of
proving that th ese old m ethods are eith er (1) h ighly
conservative, or (2) applicable to n
ewer m aterials or
applications via simple testing or numerical analysis.
These are u ltimately sh ort-sighted ap proaches to solving
problems; rath er ef fort s hould be directed tow
ards th e
fundamental reasons why the older methods do not work (or are
conservative) and to d eveloping ne w m ethods. I t i s
unreasonable to expect that 30 y ear old methods developed for
thin wall, moderate toughness line pipe steels will be applicable
to n ewer s teels of h igher s trength (g rade X100 or above) and
toughness, larger d iameter, t hicker w all ( deep w ater p ipelines
3

are approaching 50 mm in thickness), higher strains (deep water

and arctic co nditions (frost heave) will give rise to greater than
1 percen t pl astic s trains). T he ori ginal f low s tress dependent
methods w ere n ot con servative (s ee Fig . 1), an d th ey, an d the
methods th at w ere bas ed on th em, are not necessarily
theoretically applicable to newer, thicker materials.
The pi oneering w ork i n t he 1960s and 70s made use of
leading edge k nowledge of f racture m echanics, an d th is
fundamental research w as act ively s upported by t he pi peline
industry. Whether this can be said of the industry at the start of
the 21st century is an other matter. Su ch a f ailing will impede
the d evelopment o f n ew d esign and integrity solutions (high
grade, high pressure, high stress, high strain, etc.).
1.4

CONSERVATIVE

normalised failure stress

1.2

1.0

K c2
=
8c 2

0.8

12
E
M
A
= ln sec

8c 2
2

Cv

UNCONSERVATIVE

0.6
through-wall defects
part-wall defects
0.4
0

10

12

normalised flaw size

Fig. 1 The NG-18 equations and test data, illustrating flow

2
stress and toughness dependent behaviour

3. THE PIPELINE DEFECT ASSESSMENT MANUAL

PDAM is bas ed u pon a com prehensive, critical an d
authoritative review of av ailable pipelin e def ect as sessment
methods. This critical review in cludes a co mpilation o f all o f
the p ublished f ull-scale test d ata u sed in th e d evelopment an d
validation of existing defect assessment methods. The full-scale
test data is u sed to as sess th e in herent accu racy of the defect
assessment m ethods, an d to id entify th e best methods
(considering relev ance, accu racy an d eas e of use) and their
range o f ap plicability. P DAM d escribes th e best method f or
assessing a particular type of defect, defines the necessary input
data, g ives th e lim itations o f th e m ethod, and defines an
appropriate factor to account f or th e m odel u ncertainty. T he
model uncertainty for each assessment method has been derived
from a statistical comparison of th e p redictions o f th e m ethod
with the published test d ata, based on the prediction interval of
the classical linear regression model.
PDAM p rovides th e written tex t, th e m ethods, recip es for
application, acceptan ce ch arts an d s imple ex amples, an d is
supported by literature reviews. Sim ple electron ic workbooks
2

T he e quation is the to ughness dependent through-wall failure criterion,

expressed in imperial units[5].

have b een d eveloped to p ermit easy im plementation o f the

best m ethods. T he ro le o f P DAM in the fitness-for-purpose
assessment of a defect in a pipeline is summarised in Fig. 9.
PDAM ha s b een c losely scrutinised throughout its
development b y th e sp onsors, an d all literatu re rev iews an d
chapters of the m anual h ave been in dependently rev iewed by
international experts in the field of pipeline defect assessment.
PDAM does not present new defect assessment methods; it
presents the current s tate of th e art in f itness-for-purpose
assessment o f d efective p ipelines. L imitations o f th e m ethods
recommended in P DAM rep resent lim itations o f th e available
methods, and of the current state of knowledge.
4. TYPES OF DEFECT CONSIDERED IN THE
PIPELINE DEFECT ASSESSMENT MANUAL
PDAM con tains g uidance f or th e as sessment of the
following types of defect:
defect-free pipe
corrosion
gouges
plain dents
kinked dents
smooth dents on welds
smooth dents containing gouges
smooth dents containing other types of defects
manufacturing defects in the pipe body
girth weld defects
seam weld defects
cracking
environmental cracking
In ad dition, g uidance is g iven o n th e treatm ent o f the
interaction between def ects, an d th e as sessment of def ects in
pipe fittings (pipe work, fittings, elbows, etc.). Guidance is also
given on predicting the b ehaviour o f d efects up on p enetrating
the pipe wall (i.e. leak or rupture, and fracture propagation).
The following types of loading have been considered in the
development of th e g uidance: in ternal pressure, external
pressure, axial force and bending moment.
Methods a re gi ven i n P DAM fo r assessing the burst
strength of a defect subject to s tatic loadin g an d f or as sessing
the fatigue strength of a defect subject to cyclic loading. There
are s ome com binations of def ect type, orientation and loading
for which there are n o clearly defined assessment methods. In
summary, th e as sessment of def ects s ubject to s tatic or cyclic
internal pres sure l oading i s w ell understood, but, in general,
other loads and combined loading are not.
5. THE LAYOUT OF THE PIPELINE DEFECT
ASSESSMENT MANUAL
The P ipeline Def ect A ssessment Manual follows the
following format for each defect type and assessment method:
1. A brief definition of the type of defect.
2. A f igure illu strating th e d imensions an d o rientation o f th e
defect relative to the axis of the pipe, and a nomenclature.

Brief n otes th at h ighlight particu lar problem s as sociated

with the defect.
4. A flow chart summarising the assessment of the defect.
5. The minimum required information to assess the defect.
6. The assessment method.
7. The ran ge o f ap plicability o f th e method, its background,
and any specific limitations.
8. An appropriate m odel u ncertainty f actor t o be appl ied t o
the assessment method.
9. An example of the application of the assessment method.
10. Reference to alternative sources of information available in
national or international guidance, codes or standards.
The flow charts included for each defect type consist of a
number of yes-no type questions designed to identify whether or
not the methods contained in that chapter are appropriate to the
given case, and to in dicate the appropriate m ethod to u se. An
example of the flow chart for the assessment of a smooth dent
containing a gouge is given in Fig. 10.
6. ASSESSMENT METHODS IN THE PIPELINE
DEFECT ASSESSMENT MANUAL
A summary of all o f th e m ethods reco mmended in th e
Pipeline Defect A ssessment Man ual f or p redicting th e b urst
strength of a defect subject to internal pressure is given in Table
3. Longitudinally and circu mferentially orien tated def ects are
considered. T he primary methods (in dicated in n ormal font)
are plastic collapse (flow stress d ependent or limit state) failure
criteria, and a re o nly a ppropriate i f a m inimum t oughness i s
attained (see below). T he s econdary m ethods (in dicated in
italic font) are th e altern ative m ethods recommended when a
minimum toughness is no t a ttained. U pper she lf b ehaviour i s
assumed t hroughout. T he ge neral procedures for assessing
flaws in s tructures, bas ed on f racture m echanics, g iven in BS
7910 (and API 579) can be appl ied i n g eneral (i rrespective of
upper or lower sh elf b ehaviour), b ut w ill g enerally b e
conservative compared to the pipeline specific methods3.
Having given an overview o f t he c ontents o f P DAM, t he
remainder of this paper (1) describes the role of toughness and
gives empirical toughness limits fo r t he a pplication o f fl ow
stress dependent a ssessment m ethods, a nd ( 2) give s sp ecific
guidance on the assessment of gouges and dents and gouges.
7. TOUGHNESS LIMITS
Line pipe steels is generally tough and ductile, and operates
on t he u pper s helf4. In itiation an d p ropagation o f a p art-wall
3

P AFFC inco rporates co rrelations be tween the f racture to ughness 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 ) is measured).
4
Brittle (cleavage) fracture can occur in older line pipe steels or under unusual
(typically upset) conditions which can cause low temperatures. I f the DWTT
(Drop Weight Tear Test) transition temperature is less than the minimum design
temperature, then initiation will be ductile. A high upper shelf Charpy V-notch
impact e nergy is al so de sirable to e nsure that f ailure is co ntrolled by pl astic

flaw thr ough the w all o ccurs und er a d uctile fracture

mechanism, involving some combination o f p lastic f low an d
crack initiation and ductile tearing, involving a p rocess of void
nucleation, growth and coalescence. The relative importance of
plastic f low an d crack in itiation an d tearing depends on the
toughness of the material and the geometry of the defect. Fig. 2
is an illustration of the role of toughness in the failure of a partwall defect.
bu rst pressure / bu rst pressure of defect-free p ipe

3.

1.0

failure pressure
of pipe of
reduced crosssection (1-d/t)

blunt defect

sharp defect
decreasing increasing
acuity
toughness

increasing
defect
depth (d)

0.0

0
n o rm a lis e d d e fe c t le n g th

Fig. 2 The effect of material toughness, defect depth,

length and acuity on burst strength

As t he t oughness d ecreases t he b urst st rength o f a defect

will decrease. As the toughness increases the burst strength of a
defect w ill in crease, b ut ten ding towards an upper limit
corresponding to th e p lastic co llapse lim it state, where failure
occurs due to plastic flow (and can be predicted using limit state
methods). T herefore, i f t he t oughness i s gr eater than some
minimum value then the failure of a defect will be controlled by
plastic collapse and only knowledge of the tensile properties of
the material is req uired to p redict th e b urst stren gth (as
demonstrated in the transition between the toughness dependent
and flow stress forms of the NG-18 equations).
The u pper b ound to th e stren gth of a material is the
ultimate tensile strength. If f ailure is d ue to p lastic co llapse
then th e f low stress sh ould b e th e u ltimate tensile strength;
failure will o ccur w hen th e stress in th e rem aining lig ament
exceeds U. T he minimum to ughness necessar y to ensur e that
failure is controlled by plastic co llapse m ay b e h igh; L eis
suggests a full size equivalent upper shelf Charpy impact energy
of between 60 an d 75 f tlbf (81 J and 102 J)5 for a f ully ductile
response[16]. C onsidering Fig. 1 an d Fig. 3, it is clear that flow
stress dependent behaviour, as defined in the context of the NG18 equations, manifests itself at a lower toughness.
This in troduces an im portant d istinction. A minimum
toughness may b e d efined e mpirically a bove w hich a gi ven
collapse[16-18]. Th e DWTT transition temperature is defined as the temperature
at which a DWTT specimen exhibits 85 percent shear area. The steel is on the
upper s helf if the D WTT tr ansition te mperature is less than the current
temperature of the steel.
5
The 2/3 thickness specimen size equivalent is between 54 J and 68 J.

1.4
CONSERVATIVE
actual failure stress/predicted failure stress

flow s tress depen dent (or ps eudo plastic col lapse) failure
criterion will g ive reaso nably co nservative p redictions (tak ing
into account ex perimental s catter). T his is n ot equ ivalent to
stating that failure is due to plas tic collaps e. T he em pirical
minimum toughness may be l ower t han t he t rue m inimum
toughness f or p lastic co llapse b ecause o f the
inherent
conservatism in th e f low stress d ependent f ailure criterion
(consider that f low stress d ependent f ailure criteria ty pically
define the flow stress as some function of Y, or th e average of
Y an d U, an d im plicitly co nsider so me d egree o f d uctile
tearing (tearing was observed in the original full scale tests used
to develop the NG-18 equations[14]).
Wall th ickness is also im portant b ecause of the transition
from plane stress to plane s train beh aviour an d th e in creasing
constraint with increasing w all th ickness. P ipelines are
typically th in w alled stru ctures (th e w all thickness is seldom
greater t han 1 i n. ( 25.4 m m)). A m inimum t oughness limit
should b e d efined w ith resp ect to a m aximum w all th ickness.
Defect acuity is also a con sideration, blu nt def ects are les s
sensitive to toughness tha n sha rp d efects ( blunt d efects r ecord
higher burst strengths in low to moderate toughness steels).
Toughness Limits for the NG-18 Equations Em pirical
minimum toughness limits for the a pplicability o f the flo w
stress dependent though-wall a nd p art-wall N G-18 e quations
can be def ined by reference to th e results of relevant full scale
burst tests (see section 8.1).
The ef fect o f to ughness o n the accur acy o f predictions of
the bu rst s trength of an ax ially orien tated, machined, part- wall
defect made w ith t he f low s tress depen dent part -wall N G-18
equations is illustrated in Fig. 3. A flow stress of the average of
Y and U and a two term Folias factor has been used (Eqs. (1)
to (3), below ). T he prediction s becom e in creasingly nonconservative a t a l ower t oughness. T he sc atter in the range
from 20 J t o 45 J i s al so cl ear, w ith s ome t ests bei ng n onconservatively predicted an d oth ers bein g con servatively
predicted, i n an approx imate ran ge from 0.80 to 1.20 (ratio of
the actual to predicted failure stress). Consequently, taking into
account the observed scatter, it is reasonable to apply the flow
stress d ependent p art-wall NG-18 eq uation if the 2 /3 thickness
specimen s ize u pper s helf C harpy V- notch impact en ergy is at
least 21 J (16 ftlbf). The maximum wall thickness in this set of
test d ata i s 2 1.7 m m. Therefore, this minimum toughness
requirement is o nly v alid f or lin e p ipe o f a thickness less than
21.7 mm. It is sh own later in Fig . 5 th at co nservative
predictions of th e f ull scale tests can b e o btained if th is
toughness limit is applied together with a suitable correction for
the model uncertainty.
It is important to note that whilst this approach to deriving
a toughness limit is simple and practical, it has the disadvantage
of i ntroducing fur ther c onservatism fo r hi gher toughness line
pipe steels. Fu rthermore, it is n ot a limit for failure by plastic
collapse, as def ined by L eis (2001). A more sophisticated
approach, s uch as PAFFC, w ould be m ore robu st f or a w ider
range of material toughness.

1.2

1.0

0.8

0.6
UNCONSERVATIVE

Battelle (1973)
CANMET (1988)
TWI (1982)
Batelle (1986)
CSM SNAM EUROPIPE (2000)
Keller et al. (1987)
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 predictions of part-wall

burst tests made using the flow stress dependent part-wall
6
NG-18 equation

A sim ilar an alysis o f b urst tests o f axially orientated,

machined, t hrough-wall d efects i n l ine pipe indicates that a
minimum 2/3 th ickness s pecimen s ize u pper s helf Charpy Vnotch impact energy of 40 J (29.5 f tlbf) is n ecessary f or th e
flow st ress d ependent t hrough-wall N G-18 fa ilure criterion to
be applied. T he m aximum w all th ickness is 2 1.9 m m. T his
difference b etween p art-wall a nd t hrough-wall d efects follows
the same trend as tes ts th at h ave in dicated th at th e f racture
initiation tran sition tem perature (FIT T) (th e tem perature at
which a f racture ch anges f rom b rittle to ductile) of a part-wall
defect is lower than that of a through-wall defect[17,18].
Range of Toughness from Published Data The minimum
toughness (2/3 specimen thickness upper shelf Charpy V-notch
impact en ergy) an d m aximum w all thickness derived from the
published full scale test d ata f or sev eral ty pes o f d efect are
summarised b elow7. T hese v alues in dicate th e potential limits
of t he v arious as sessment m ethods. The methods may be
applicable outside of these lim its, b ut th ere is lim ited
experimental evidence. T he res ults of s pecific s tudies of th e
range of v alidity of s pecific as sessment m ethods are als o
indicated. In all cases, the basic assumption is that the line pipe
steel is on the upper shelf.
Corrosion The lowest toughness is 18 J (13 ftlbf) and the
maximum wall thickness is 22.5 mm (1.0 in.).
ASME B31G , m odified B31G an d R STRENG are
applicable t o l ow t oughness st eels (on the upper shelf)[19,20].
The recen tly dev eloped m ethods f or assessing corrosion, such
as DNV- RP-F101[21] a nd P CORRC[20] are on ly prov en f or
moderate to high toughness steels; a minimum toughness of 41 J
(30 f tlbf) h as been propos ed[20]. N one of t he m ethods f or

The toughness is not reported in a num ber of tests; these tests are shown in
Fig. 3 as having zero toughness to indicate the range of the test data.
7
Note that the Charpy impact energy is not reported for all of the tests.

assessing corrosion have been validated in line pipe with a wall

thickness greater than 25.4 mm.
Gouges T he l owest t oughness i s 1 4 J ( 10 ft lbf) a nd t he
maximum wall thickness is 21.7 mm (0.854 in.).
Changes to th e local m icrostructure at the base of a gouge,
as a consequence of the gouging process, have been studied by
CANMET. It is indicated that the effect of such changes were
not significant if the upper shelf Charpy V-notch impact energy
(2/3 specimen s ize) ex ceeded 20 J [42]. T he flo w str ess
dependent part-wall NG-18 equation can be used to predict the
burst strength o f a go uge ( see se ction 8 ). T he m inimum
toughness to apply this m ethod i s 2 1 J ( maximum t hickness
21.7 mm), see above.
Dent and Gouge The l owest t oughness i s 1 6 J ( 12 ft lbf)
and the maximum wall thickness is 20.0 mm (0.787 in.).
Dent The lowest toughness is 2 0 J ( 15 ft lbf) a nd t he
maximum wall thickness is 12.7 mm (0.500 in.).
8. THE ASSESSMENT OF THE BURST STRENGTH
OF A GOUGE IN PDAM
A gouge is surface damage to a pipelin e caused by contact
with a foreign object that has scrapped (gouged) material out of
the p ipe, resu lting in a m etal lo ss d efect. T he m aterial at th e
base o f a g ouge w ill h ave b een sev erely cold worked as a
consequence of the gouging process. T his work hardened layer
will h ave a red uced d uctility an d m ay co ntain crack ing. A
gouge may be in fully rerounded pipe (i.e. a dent of zero depth).
A gouge reduces the burst and fatigue strength of the pipe.
A gouge may be of any orientation with respect to the pipe
axis. A longitudinally o rientated g ouge is th e m ost sev ere
condition for internal pressure loading; therefore, the following
discussion concentrates on this orientation.
8.1 FULL SCALE BURST TESTS OF GOUGES
A large number of f ull scale b urst tests o f lo ngitudinally
orientated gouges (part- wall def ects) in lin e pipe s teel h ave
been conducted by a number of different organisations. Tests in
other pres sure v essel s teels h ave als o been carried ou t. The
total n umber of pu blished bu rst t ests i s of t he order of 190,
although only the most relevant 115 tests are referred to here.
The tests can be variously described as follows8:
1. machined V-shaped notch or slot (artificial gouge)
- Battelle (1965 - 1974)[5] (vessels) (48 tests)
- Bat telle (1986)[22] (vessels) (3 tests)
- British Gas (1974)[23] (vessels) (3 tests)
- British Gas (1981, 1982)[24] (vessels) (1 test)
- Iron and Steel Institute o f J apan (Ku bo et al.)
(1993*)[25] (vessels) (19 tests)9
- CSM SNAM EUROPIPE (2000)[26] (vessels) (2 tests)
2. scrape (g ouge) t he pi pe u sing a t ool bi t m ounted on a
pendulum

- CANMET (1985, 1988)[27,28] (vessels) (12 tests)

3. fatigue pre-cracked semi-elliptical machined notch
- TWI (Garwood et al.) (1982)[29] (vessels) (2 tests)
- TV and Mannesmann (K eller et al .) (1987) [30]
(vessels) (15 tests)
- University of T ennessee (H errera et al.) (1992)[31]
(vessels) (10 tests)
It is noteworthy that a larger degree of scatter is noticeable
in the results of tests o f f atigue p re-cracked n otches, w hen
compared to the tests of machined notches.
8.2 METHODS FOR PREDICTING THE BURST
STRENGTH OF A GOUGE
The assessment of the burst strength of part-wall defects in
pipelines derives from work conducted at Battelle in the 1960s
and 70s , cu lminating i n t he development of flow stress
dependent and toughness dependent forms of through-wall and
part-wall failure criteria (the NG-18 equations)[5]. The throughwall and part-wall criteria are semi-empirical. The through-wall
failure criterion was developed and validated against the results
of 92 full scale v essel b urst tests co ntaining artif icial,
longitudinally-orientated, t hrough-wall d efects. T he p art-wall
failure criterion was developed and validated against the results
of 48 full scale v essel b urst tests co ntaining artif icial,
longitudinally-orientated, machined V-shaped notches.
The flow stress depen dent f orm of t he part -wall f ailure
criterion has been widely used as a plastic collapse solution for
axial crack-like flaws s ubject to in ternal pres sure, an d appears
in docu ments s uch as BS 7910 an d API 579. Several
previously pu blished rev iews h ave con cluded that the NG-18
equations are the best equations for assessing part-wall defects
such a s go uges[32,33]. The part- wall NG- 18 equ ations are als o
recommended in th e EP RG g uidelines f or th e assessment of
mechanical damage[11].
The flow stress dependent part -wall N G-18 equ ation i s as
follows

d
1
t
=
d 1
1 t M

(1)

is the flow stress, which is an empirical concept intended to

represent th e s tress at w hich u nconstrained plas tic f low occurs
in a s train h ardening elas tic-plastic material via a single
parameter. One commonly used definition of the flow stress is10
=

Y +U
2

(2)

M is the Folias factor, representing the stress concentration due

to the bulging that occurs under internal pressure loading. T he

The tests marked with an asterisk hav e no t be en incl uded in the statistical
comparison of the two methods.
9
Note that there is a la rge d ifference b etween t he t est t emperature a nd t he
temperature at which the material properties were measured.

10

A SME B31G us es a f low s tress o f 1.1 tim es the yield strength, modified
B31G and RSTRENG (and the NG-18 equations) use a flow stress of the yield
strength plus 10 ksi (68.95 Nmm-2).

analytical so lution f or th e Fo lias f actor is an infinite series.

Three commonly used approximations are given below.
2c
M = 1 + 0.26
Rt

2c
M = 1 + 0.314
Rt
2c
M = 1 + 0.40
Rt

(3)

2c

0.00084

Rt

(4)

(5)

Equation (5) is th e ex pression th at appears in ASME

B31G. It is the most conservative approximation. Equation (4)
appears i n m odified B31G an d R STRENG. Equ ation (3) i s a
close approximation to Eq. (4) that is valid for 2c/(Rt)0.5 greater
than 8.0.
The growth through wall of a sha rp, p art-wall d efect i n
ductile line p ipe o ccurs tho ugh some combination of plastic
flow and ductile tearing. The NG-18 equations do not explicitly
consider the effects of ductile tearing on the failure of throughwall an d p art-wall d efects. A m ore so phisticated m ethod f or
assessing part-wall defects, such as gouges, is PAFFC[9].

Predicted Failure Stress/Yield Strength, percent

160

120

100

UNCONSERVATIVE

80

60

CONSERVATIVE

40

20

0
0

20

40

60

80

mean

standard
deviation

coefficient
of variation

(1)

two term Folias (Eq. 5)

three term Folias (Eq. 4)
approximate Folias (Eq. 3)

1.06
1.02
0.99

0.16
0.14
0.13

0.15
0.14
0.13

(2)

two term Folias

three term Folias
pproximate Folias

1.05
1.01
0.98

0.15
0.13
0.12

0.15
0.13
0.13

two term Folias

three term Folias
pproximate Folias

0.95
0.92
0.89

0.15
0.14
0.13

0.16
0.15
0.14

a
(3)
a

Note : (1) average of yield strength and tensile strength, (2) yield strength plus
10 ksi, and (3) tensile strength.

Table 1 Statistical comparison of NG-18 equation with

several forms of the Folias factor and flow stress

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

140

the p ublished test d ata f or p redicting th e b urst strength of a

gouge. H owever, t his e quation ha s b een published with
different definitions of the flow stress and the Folias factor (M).
Consequently, t he va rious fo rms o f t he N G-18 e quations ha ve
been compared using th e p ublished test d ata. On ly tests o n
machined n otches h ave been con sidered. T ests where there is
insufficient data and where the u pper sh elf 2 /3 th ickness size
Charpy im pact en ergy is les s th an 21 J (s ee s ection 7, above)
have been ex cluded. T he total n umber of f ull scale tests
considered in the comparison is 71. The statistics of the ratio of
the actual failure stress to the predicted failure stress are given
in Table 1.

100

120

140

160

Failure Stress/Yield Strength, percent

Fig. 4 Failure stress of axially orientated part-wall defects

predicted using the part-wall NG-18 equation

8.3 COMPARISON WITH TEST DATA

The flow stress dependent f orm of t he part -wall N G-18
equations is the best method in terms of the quality of fit with

There is little d ifference b etween th e three forms of the

Folias factor, the approximate two term factor (Eq. (3)) an d the
three term factor (Eq. (4)) being almost identical; similarly for a
flow stress of the average of Y and U, and one of Y plus 10
ksi (as quoted in Kiefner et al. (1973)). A flow stress equal to
U gives, on average, non-conservative predictions, and a slight
increase in th e s catter. A com parison between the predictions
made u sing th e NG- 18 eq uation, w ith a flow stress of the
average of Y and U and the two term Folias factor (Eqs. (1) to
(3)), and the published full scale test data is shown in Fig. 4.
8.4 RECOMMENDATION IN PDAM
PDAM recommends th e s emi-empirical NG- 18 part- wall
flow stress dependent failure criterion with the approximate two
term Folias f actor an d a f low s tress of th e average of yield
strength a nd t ensile st rength ( Eqs. ( 1) t o ( 3)). The equations
should not b e ap plied if the 2 /3 thickness sp ecimen size u pper
shelf Charpy V-notch impact energy is less than 21 J (16 ftlbf).
The wall thickness must be less than 21.7 mm.
The part-wall NG-18 equation does not give a lower bound
estimate; accordingly, a model u ncertainty h as been deriv ed.
The effect of applying a con fidence interval corresponding to a
95 percent one-tail confidence level is illustrated in Fig. 5; note
that a ll o f t he t ests w ith a toughness greater than 21 J are
conservatively predicted.
8

lower bound Predicted Failure Stress/Yield Strength, percent

When assessing a gouge it is im portant to co nsider th e

possibility of cracking at the base of the gouge and the presence
of a den t. A n as sessment can be n on-conservative if these
issues are not considered. This may mean that it is necessary to
excavate the pipeline to p erform a d etailed in spection o f th e
damage. It is suggested that the measured depth of a gouge be
increased by 0.5 mm to account for the possibility of cracking at
the base of the gouge, unless an inspection technique is used to
detect and measure cracking.
160
unknown Charpy
Charpy less than 21 J
Charpy greater than or equal to 21 J

140

120

UNCONSERVATIVE
100

80

60

40

CONSERVATIVE

20

0
0

20

40

60

80

100

120

140

160

Failure Stress/Yield Strength, percent

Fig. 5 Failure stress of axially orientated part-wall defects

predicted using a lower bound to the part-wall NG-18
equation

8.5 RANGE OF APPLICABILITY

The recommended method for assessing the burst strength
of a lon gitudinally orientated gouge has been compared against
the results of 92 f ull scale b urst tests o f v essels co ntaining
artificial, machined p art-wall d efects an d g ouges, in cluding
some materials other than line pipe s teel. T he range of the test
data included in the comparison is as follows (in SI units). This
gives an indication of the range of applicability of the part-wall
NG-18 equation.
Pipe Diameter, mm
Wall Thickness, mm
2R/t ratio
Grade (API 5L)
Yield strength, Nmm-2 379.2
Tensile strength, Nmm-2 483.3
yield to tensile ratio
2/3 Charpy Impact Energy, J
Notch Depth (d), mm

114.0
5.6
13.3
X52
0.69
13.6
0.49

to
to
to
to
to
to
to
to
to

1422.4
21.7
104.0
X100
878.0
990.0
0.99
261.0
16.8

d/t 0.088
Notch Length (2c), mm
2c/(Rt)0.5 0.41
Burst Pressure, Nmm-2 1.84
Burst Stress, Nmm-2 61.4
Burst Stress (percent SMYS)

14.0

13.7

to
to
to
to
to
to

0.92
609.6
8.16
142.0
880.7
132.5

9. THE ASSESSMENT OF THE BURST STRENGTH

OF A DENT AND GOUGE IN PDAM
A dent is a depres sion which produces a g ross disturbance
in the curvature of th e pipe w all, cau sed by con tact w ith a
foreign body resulting in plas tic def ormation of th e pipe w all.
External in terference can cau se both m etal loss defects
(gouging) and dents.
A d ent c ontaining a go uge ( or o ther type of metal loss
defect) is a very severe form of damage. The burst strength of a
smooth dent containing a gouge is lower than the burst strength
of an equivalent plain dent, and lower than that of an equivalent
gouge in undented pipe. T he fatigue strength of a smooth dent
containing a gouge is lower than that of an equivalent plain dent
9.1 FULL SCALE BURST TESTS OF DENTS AND
GOUGES
A large number of full scale ring and vessel burst tests of a
smooth dent containing a si ngle go uge ha ve b een c onducted
by a variety of different or ganisations, s ee below . T he total
number of pu blished t ests i s 242. However, most of the tests
have actu ally been of m achined n otches or s lots, rath er th an
gouges. A variety of different test methods have been used, as
indicated b elow. A ll o f th e m achined notches (slots) and
gouges ha ve b een l ongitudinally o rientated. A ll o f t he d ents
have been longitudinally orientated, except for the Gasunie tests
in which transverse dents were introduced into pipe.
The tests can be variously described as follows11:
1. damage introduced at zero pressure; introduce the dent and
then m achine a V-shaped n otch (artif icial g ouge) in the
base of the dent
- British Gas (1982, 1989) [24,34] (108 ri ng t ests an d 23
vessel tests)
- Tokyo Gas (1998*)[35] (vessels) (3 tests)
2. damage introduced at zero pres sure; machine a V-shaped
notch (artificial gouge) and then introduce the dent
- Battelle (1979, 1986)[22,36-38,39] (vessels) (30 tests)
- Nanyang Technical University (1992*)[40] (vessels) (17
tests)
3. damage introduced at zero pres sure; machine a V-shaped
notch (artificial gouge) and then introduce the dent (a sharp
steel trian gle w as in serted in th e notch between the
cylindrical indenter and the pipe)
- DNV (2000)[41] (vessels) (1 test)
4. damage introduced at zero pressure; introduce the dent and
then scrape (gouge) the pipe using a t ool bit mounted on a
pendulum
11
T he te sts m arked with an aste risk hav e no t be en included in the statistical
comparison of the two methods.

- CANMET (1985, 1988)[28,42] (vessels) (11 tests)

5. damage (dent) in troduced at pres sure; m achine a Vshaped notch (artif icial g ouge) at zero pres sure an d th en
introduce the dent at pressure
- S ES (1996)[43,44] (vessels) (14 tests)
6. damage (den t) in troduced at pres sure; gouge at zero
pressure and then introduce the dent at pressure
- EPRG (1991*, 1992*)[45,46] (vessels) (8 tests)
7. damage i ntroduced at a l ow pres sure (150 ps i) or zero
pressure; damage introduced u sing an in denter w ith a
machined sharp edge (w ith a 60 deg ree in cluded an gle)
along its length
- Bat telle (1978)[36] (vessels) (2 tests)
8. damage introduced at pressure; dent and gouge introduced
simultaneously using a specially designed test rig
- British Gas (1983*)[47] (vessel) (1 test)
- Bat telle (1986*)[22,39] (vessels) (17 tests)
9. damage (transverse dent) introduced at pressure and gouge
introduced at zero pres sure; den t at pres sure, depres surise
(holding indenter in place) and then scrape (gouge) the pipe
using the indenter
- Gasunie (1986*, 1990*)[48,49] (vessels) (10 tests)
10. damage introduced at pres sure; machine a blunt (rounded)
notch at zero pressure an d th en in troduce th e den t at
pressure
- University of Cambridge (1992* , 1993* , 1996* )[50-52]
(vessels) (20 tests)
11. damage in troduced at zero pres sure; machine a 1 in. wide
slot (artificial corrosion) and then introduce the dent
- S ES (1997*)[53] (vessels) (3 tests)
Internal pressure s tiffens t he res ponse of t he pi pe t o
indentation, such that dents in troduced at p ressure w ill b e
smaller than those introduced at zero pres sure, and puncture is
more likely (if the indenter is sharp). Introducing dents at zero
pressure allow s deeper den ts to be f ormed than would be
observed in practice[22]. A ring test simulates an infinitely long
gouge in a co ntinuous d ent. A continuous dent will spring
back an d rerou nd m ore th an a s hort den t becau se it is
geometrically less stif f (there is n o co nstraint from the ends of
the dent). Introducing the den t af ter th e g ouge in creases th e
likelihood of cracking occurring at the base of the gouge. T he
most realistic tests are th ose in w hich th e dent and gouge are
introduced into pressurised pipe under dynamic conditions.
9.2 METHODS FOR PREDICTING THE BURST
STRENGTH OF A DENT AND GOUGE
The behaviour of a dent containing a gouge is complex. A
dent and gouge is a g eometrically unstable structure. T he base
of the gouge m ay con tain crack ing an d th e properties of th e
material i n t he d ent a nd go uge m ay ha ve b een adversely
affected. Ou tward m ovement o f th e d ent promotes initiation
and growth of cracking i n t he b ase o f t he go uge, c hanging t he
compliance of th e den t an d g ouge s tructure. The failure of a
dent and go uge d efect i nvolves hi gh p lastic st rains, w all
thinning, movement of the dent, crack initiation, ductile tearing

and plastic flow. An analysis of the failure mechanism of a dent

and gouge defect is described by Leis et al. (2000)[54,55].
Empirical relatio nships for p redicting the b urst strength of
a smooth dent con taining a g ouge h ave been propos ed by
British Gas [24,47], th e EP RG[11] an d B attelle[22,37]. A s emiempirical fracture model f or as sessing th e bu rst s trength of a
dent-gouge d efect h as b een d eveloped b y British Gas[56], an d
has subsequently been included in the EPRG recommendations
for the as sessment of m echanical dam age[11]. More
sophisticated m odels are u nder dev eloped (e.g . L eis et al.
(2000)), w hich attem pt to m ore accu rately m odel the failure
mechanism of a dent and gouge defect.
The t wo m ost w idely qu oted m odels f or predi cting the
failure stress of a dent and gouge defect are:
1. T he empirical Q factor model developed by Battelle under
the auspices of the Pipeline Research Council International
(PRCI)[22,37].
2. The d ent-gouge f racture m odel d eveloped by British Gas
and adopted by the EPRG[11,56].
Both of these m odels are bas ed on t he den t dept h af ter
spring back and measured at zero pressure.
The Empirical Q Factor Model B attelle d eveloped an
empirical m odel f or predi cting t he bu rst strength of a smooth
dent c ontaining a go uge b ased o n t he r esults o f 3 0 ful l scale
burst tests [22,36-38], in which the d amage was introduced at zero
pressure by n otching an d t hen den ting t he pi pe. The failure
stress, normalised by the flow stress, was related to an empirical
parameter, denoted Q. T he Q factor is defined as a function of
the upper shelf Charpy impact energy (for a 2/3 size specimen),
the dent depth (af ter s pring back an d m easured at zero
pressure), the gouge length, and the gouge depth.
The e mpirical r elationship i s gi ven b y t he following
equations (in imperial units)
f

Q=

(Q 300 )0.6
90

(6)

Cv
(7)
H
d

(2c )
2R
t

= Y + 10000 psi (8)

Fig. 6 s hows a com parison betw een the predictions made
using the empirical Q factor model and the published full scale
test data.
The Dent-Gouge Fracture Model The dent-gouge defect
is m odelled as an ax ially o rientated, co ntinuous dent (of
constant width) with a single, infinitely long, axially orientated,
sharp notch located at the base of the dent. T he length of the
dent or the gouge is n ot con sidered. T he elev ated m embrane
and bending stresses at th e bas e of th e den t are con sidered,
through an approximate solution based on thin shell theory and
Castiglianos second th eorem. T he u nderlying f racture model,

10

considering the r eaction b etween fr acture ( toughness) a nd

plasticity, is a col lapse modified strip-yield model. The model
was calibrated using the results of 111 ring and 21 vessel burst
tests of smooth dents containing machined notches (notch then
dent) introduced at zero pressure carried out by British Gas [24].
A relationship b etween t he i mplied fr acture t oughness a nd t he

upper shelf Charpy impact energy (for a 2/3 size specimen) was
determined from a non-linear regression analysis of the dent and
gouge test data (therefore, th e co rrelation b etween Ch arpy
energy and fracture toughness is not generally applicable).
The dent-gouge fracture model is defined as follows (in SI
units)

ln(0.738C v ) K 1
Ho

2
1.5E
R H o

1
= cos exp 113 2
+ Y2 10.2
exp
Y1 1 1.8
(9)
D
t D
K2

Ad

d
= 1.15 Y 1
t

(10)
2

d
d
d
d
Y1 = 1.12 0.23 + 10.6 21.7 + 30.4
t
t
t
t
2

d
d
d
d
Y2 = 1.12 1.39 + 7.32 13.1 + 14.0
t
t
t
t

(11)

(12)

K1 = 1.9

(13)

K 2 = 0.57

(14)

H o = 1.43H r

(15)

The f low s tress as sumed in th e den t-gouge fracture model

is not appropriate f or h igher g rade s teels (g reater th an X65),
due to the increasing yield to tensile ratio with line pipe grade.
The dent-gouge fracture model i s bas ed on t ests i n which
the damage was introduced at zero pressure, and the dent depth
is th at af ter s pring back an d m easured at zero pres sure.
Therefore, a correction m ust be m ade f or den ts in troduced at
pressure and m easured at pres sure. A n em pirical rerou nding
correction factor developed by t he EPR G i s propos ed (Eq.
(13))[11]. This correction factor relates the dent depth (after the
removal of the indenter) measured at pres sure to th at measured
at zero pres sure, f or den ts in troduced at pres sure. It is w orth
noting th at th is em pirical co rrection is based on limited test
data, an d t hat al ternative m ethods h ave been dev eloped which
should be more robu st (e.g . R osenfeld (1998) [57]), a lthough
there is limited test d ata available to validate such methods and
they require more in formation th an is g iven in th e relev ant
published t ests. T here ha ve b een no b urst tests which have
directly compared the effect of denting at pres sure and denting
at zero pressure on t he f ailure beh aviour of a s mooth den t
containing a go uge. Co nsequently, c orrecting for denting at
pressure remains an area of considerable uncertainty.

Fig. 7 s hows a com parison betw een the predictions made

using th e s emi-empirical den t-gouge f racture m odel and the
published full scale test data.
9.3 COMPARISON WITH TEST DATA
The empirical Q factor model and the dent-gouge fracture
model are compared against the published test d ata in order to
determine th e best m ethod in term s o f th e quality of fit with
the test d ata. A n umber o f th e tests can not b e co nsidered
because of the absence of toughness, actual m aterial properties
or dent depth after spring back measured at zero pressure. Tests
involving transverse dents or t ests i n w hich t he go uge ha s
been ground smooth have also been excluded.
The total number of full scale tests co nsidered in th e
comparison i s 162, i ncluding 93 ri ng t ests an d 69 v essel tests.
The formulation o f the Q factor model is su ch that if Q is less
than 300 ft.lbf.in-1, th en th e f ailure stress can not b e d efined.
Therefore, although the gouge length is given for all of the 69
vessel te sts, the Q factor model can on ly be appl ied t o 55 of
these tests.
mean

standard
deviation

coefficient
of variation

(1) f

racture model
Q factor

1.09
1.80

0.48
2.02

0.44
1.12

(2) f

racture model
Q factor

1.23
1.45

0.64
0.88

0.52
0.61

Note : (1) all tests, (2) limited number of tests (refer to text).

Table 2 Statistical analysis of predictions made using the

semi-empirical dent-gouge fracture model (EPRG) and the
empirical Q factor model (PRCI)

The statistics o f the ratio o f the actual failure stress to the

predicted failure stress for the two models are given in Table 2.
Two subsets of the test data are considered: in (1) all of the tests
applicable to each model are con sidered, whilst in (2) the tests
are limited to those to which the Q factor model can be applied,
and tw o ap parent o utliers in th e p redictions o f th e Q f actor
model, o ne B attelle test an d o ne B ritish Gas test (see Fig. 6)
have been removed. T he den t-gouge f racture model is clearly
the better model. Note that there is a larger amount of scatter in

11

280

Predicted Failure Stress/Yield Strength, percent

260

British Gas (1982, 1989)

Battelle (1979)
CANMET (1985, 1988)
SES (1996)
Battelle (1978)

240
220
200

UNCONSERVATIVE

180
160
140
120
100
80
60

CONSERVATIVE

40
20
0
0

10

20

30

40

50

60

70

80

90

100

110

120

130

140

measured depth of th e g ouge be in creased by 0.5 mm, as

discussed above.
lower bound Predicted Failure Stress/Yield Strength, percent

the predictions of dent-gouge tests co mpared to the predictions

of gouges and notches in undented pipe using the part-wall NG18 equation (see above) .

Fig. 6 Failure stress of dent and gouge defects predicted

using the empirical Q factor model
110

UNCONSERVATIVE
British Gas (1982, 1989)
Battelle (1979)
Det Norske Veritas (2000)
CANMET (1985, 1988)
SES (1996)
Battelle (1978)

Predicted Failure Stress/Yield Strength, percent

90
80

CONSERVATIVE

70
60
50

0
40

50

60

70

80

90

100

40
30
20
10
0
-10
-20

110

120

130

10

20

30

40

50

60

Pipe Diameter, mm
Wall Thickness, mm
2R/t ratio
Grade (API 5L)
Yield strength, Nmm-2 279.2
Tensile strength, Nmm-2 475.0
yield to tensile ratio
2/3 Charpy Impact Energy, J
Dent Depth, mm
H/2R 0.42
Notch Depth (d), mm
d/t 0.014
Notch Length (2c), mm
2c/(Rt)0.5 0.84
Burst Pressure, Nmm-2 0.972
Burst Stress, Nmm-2 29.2
Burst Stress (percent SMYS)

10

30

50

70

80

90

100

110

120

130

140

9.5 RANGE OF APPLICABILITY

The dent-gouge fracture model has been compared against
the results of 162 full scale bu rst t ests of ri ngs an d v essels
containing dent-gouge defects or dent-notch defects. The range
of the test data included in the comparison is given below (in SI
units). This gives an indication of the range of applicability of
the dent-gouge fracture model.

20

20

60

Fig. 8 Failure stress of dent and gouge defects predicted

using a lower bound to the semi-empirical dent-gouge
fracture model

30

10

70

CONSERVATIVE

Failure Stress/Yield Strength, percent

40

UNCONSERVATIVE
British Gas (1982, 1989)
Battelle (1979)
Det Norske Veritas (2000)
CANMET (1985, 1988)
SES (1996)
Battelle (1978)

80

Failure Stress/Yield Strength, percent

100

90

140

Failure Stress/Yield Strength, percent

Fig. 7 Failure stress of dent and gouge defects predicted

using the semi-empirical dent-gouge fracture model

9.4 RECOMMENDATION IN PDAM

PDAM recom mends th e den t-gouge fracture model for
assessing t he b urst st rength o f a sm ooth dent containing a
single, axially orientated gouge.
The dent-gouge fracture model d oes no t gi ve a l ower
bound es timate of th e bu rst s trength of a combined dent and
gouge, accordingly a model u ncertainty h as been deriv ed.
The effect of applying a con fidence interval corresponding to a
95 percent one-tail confidence level is illustrated in Fig. 8.
The assessment of a dent and gouge defect is difficult. The
morphology o f th e d amage is su ch that ultrasonic inspection
techniques may no t b e r eliable. I t i s sugge sted t hat t he

12

216.3
4.8
33.6
X42
0.61
16.3
1.5
0.18
50.8

7.05

to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to

1066.8
20.0
107.7
X65
543.3
701.2
0.87
130.7
146.5
18.0
6.1
0.51
810.0
8.98
25.24
626.8
151.5

ACKNOWLEDGMENTS
The au thors ack nowledge th e s ponsors of th e P ipeline
Defect Assessment Manual J oint In dustry P roject f or th eir
permission to publish this paper.

15.
16.

REFERENCES
1.

2.
3.
4.
5.

6.

7.

8.

9.
10.
11.

12.

13.
14.

Anon; Guide on methods for assessing the acceptability of flaws

in f usion w elded s tructures, B S 7910 : 1999, Incorporating
Amendment No. 1, Britis h Sta ndards Ins titution, L ondon, U K,
1999.
Anon; Managing Sy stem I ntegrity f or H azardous L iquid
Pipelines, A PI Sta ndard 1 160 ( ANSI/API ST D 1160-2001),
First Edition, November 2001.
Lewis, K., 2001, Integrity Management of Pipelines, Congreso
Internacional de D uctos ( International P ipeline C ongress),
Mrida, Yucatn, Mexico.
Anon; Fitness-For-Service, A PI R ecommended P ractice 579,
First Edition, American Petroleum Institute, January 2000.
Kiefner, J. F., Maxey, W. A., Eiber, R. J., and Duffy, A. R., 1973,
The F ailure S tress L evels o f F laws in P ressurised Cy linders,
ASTM ST P 536, A merican Soc iety for Testing and Materials,
Philadelphia, pp. 461-481.
Anon; Manual f or D etermining the R emaining Str ength of
Corroded Pipelines, A Supplement to A SME B 31 C ode f or
Pressure P iping, A SME B 31G-1991 (Revision of ANSI/ASME
B31G-1984), The American Soc iety of Me chanical Eng ineers,
New York, USA, 1991.
Kiefner, J . F ., V ieth, P . H ., 1989, A Modif ied Criterion for
Evaluating the Str ength of C orroded P ipe, Fina l R eport for
Project P R 3-805 to the P ipeline Supe rvisory Committee of the
American Gas Association, Battelle, Ohio.
Leis, B . N ., B rust, F. W., a nd Sc ott, P. M., 1991, Development
and V alidation of a Duc tile Fla w G rowth A nalysis f or Gas
Transmission Line Pipe, Final Report to A.G.A. NG-18, Catalog
No. L51543.
Leis, B. N., G hadiali, N . D ., 1994, Pipe A xial Fla w Fa ilure
Criteria - P AFFC, V ersion 1.0 U sers Ma nual a nd Sof tware,
Topical Report to A.G.A. NG-18, Catalog No. L51720.
Knauf, G ., H opkins, P ., 1996, The EP RG Guidelines on the
Assessment of Defects in Transmission Pipeline Girth Welds, 3R
International, 35, Jahrgang, Heft, pp. 620-624.
Roovers, P., Bood, R., G alli, M., Ma rewski, U., Steiner, M., a nd
Zara, M., 2000, EPRG Me thods f or A ssessing the Tolerance
and Resistance of Pipelines t o E xternal Dam age, P ipeline
Technology, V olume I I, Proceedings o f t he T hird In ternational
Pipeline T echnology C onference, B rugge, Belgium, R. Denys,
Ed., Elsevier Science, pp. 405-425.
Re. G ., P istone, V., Vogt, G ., D emofonti, G ., and Jones, D. G.,
1993, EPRG Recommendation for Crack Arrest Toughness for
High Strength Line P ipe Steels, P aper 2, P roceedings of the 8 th
Symposium o n L ine P ipe Research , A merican G as A ssociation,
Houston, Texas, pp. 2-1-2-13.
Rossmanith, H. P., 1999, The Strug gle f or Re cognition of
Engineering Fracture M echanics, F racture Research i n
Retrospect, H. P. Rossmanith, Ed., A.A. Balkema Publishers.
Duffy, A. R., McClure, G. M., Maxey, W. A., and Atterbury, T. J.,
1968, Study of the Fe asibility of Ba sing N atural G as P ipeline
Operating Pressure on Hydrostatic Test Pressure, Final Report to

17.

18.

19.

20.

21.
22.

23.
24.
25.

26.

27.

28.

29.

13

the Ame rican Ga s As sociation, AGA C atalogue N o. L 30050,

Battelle Memorial Institute.
Hahn, G . T., Sa rrate, M., a nd Rosenfield, A. R., 1969, Criteria
for Cra ck Ex tension in Cy
lindrical P ressure V essels,
International Journal of Fracture Mechanics, 5, pp. 187-210.
Leis, B. N., Thomas, T. C ., 2001, Line-Pipe P roperty I ssues in
Pipeline Design a nd in R e-Establishing MA OP, C ongreso
Internacional de D uctos ( International P ipeline C ongress),
Mrida, Yucatn, Mexico.
Eiber, R . J ., B ubenik, T . A ., 1993, Fracture Control Plan
Methodology, Paper 8, Eighth Sy mposium on L ine P ipe
Research, P ipeline Research Co mmittee o f t he A merican G as
Association, Catalogue No. L51680, Houston, Texas, USA.
Eiber, R. J., Leis, B. N., 2002, Fracture Control Technology for
Pipelines Circa 2000, Final Report on P roject PR-3-00108 to
the L ine P ipe Re search Supe rvisory Com mittee of the Pipeline
Research C ouncil I nternational, P RCI Report PR-3-00108,
Battelle.
Kiefner, J. F., V ieth, P. H ., a nd R oytman, I ., 1995, Continued
Validation of R STRENG, U pdated D raft Final Report on
Contract No. PR 218-9304 to L ine P ipe R esearch Supe rvisory
Committee, P ipeline Research Co mmittee o f t he A merican Gas
Association, Kiefner and Associates, Inc..
Stephens, D. R., L eis, B . N ., 2000, Development of a n
Alternative Criterion for Residual Strength of Corrosion De fects
in Moderate- to High-Toughness Pipe, Volume 2, Proceedings of
the Third International Pipeline Conference (IPC 2000), Calgary,
Alberta, Canada, American Society of Mechanical Engineers, pp.
781-792.
DNV-RP-F101, Corroded Pipelines, Det Norske Veritas, 1999.
Maxey, W. A ., 1986, Outside For ce Defect Behaviour, Report
to L ine P ipe R esearch Supe rvisory C ommittee of the P ipeline
Research Committee of t he A merican G as A ssociation, NG -18
Report No. 162, AGA Catalogue No. L51518, Battelle.
Shannon, R . W . E., 1974, Failure B ehaviour of L ine P ipe
Defects, International Journal of Pressure Vessels and Piping, 2,
pp. 243-255.
Jones, D. G., 1982, The Significance of Mechanical Damage in
Pipelines, 3R International, 21, Jahrgang, Heft.
Kubo, T., Shiwaku, T., Kondo, J., Miyazaki, H., and Kawaguchi,
Y., 1993, Proposal of Modif ied Spe cimen f or Chevron Notch
Drop W eight T ear T est, P aper 4 , P roceedings o f t he 8th
Symposium on Line Pipe Research, Pipeline Research Committee
of the American Gas Association, Houston, Texas, USA.
Demofonti, G., Mannucci, G ., Ba rsanti, L ., Spine lli, C. M., a nd
Hillenbrand, H . G ., 2000, Fracture Be haviour a nd D efect
Evaluation of Large Diameter, HSLA Steels, Very High Pressure
Pipelines, Volume 1, P roceedings o f t he T hird In ternational
Pipeline Conference (IPC 2000) , C algary, A lberta, C anada,
American Society of Mechanical Engineers, pp. 537-545.
Wang, K . C ., Sm ith, E. D ., 1988, The Ef fect of Mechanical
Damage on Fra cture Initia tion in L inepipe: P art II - Gouges,
Canadian C entre f or Mine ral a nd Ene rgy T echnology
(CANMET), Canada, Report ERP/PMRL 88-16 (TR).
Tyson, W. R ., Wang, K . C ., 1988, Effects of Ex ternal D amage
(Gouges a nd D ents) on P erformance of L inepipe. A R eview of
Work at MTL, CANMET, C anadian C entre f or Mine ral a nd
Energy T echnology ( CANMET), C anada, Report MTL 88-34
(OP).
Garwood, S. J., Willoughby, A. A., and Rietjens, P., 1981, The
Application of CTOD Methods for Safety Assessment in Ductile

30.

31.

32.

33.
34.

35.

36.

37.

38.

39.
40.
41.

42.

43.

44.

Pipeline Ste els, P aper 22, I nternational C onference on Fitness

for P urpose V alidation of W elded C onstructions, T he Welding
Institute, London, UK.
Keller, H . P ., J unker, G ., a nd Me rker, W ., 1987, Fracture
Analysis of Surface Cracks i n Cy lindrical P ressure V essels
Applying the T wo P arameter Fr acture Criterion (TPFC),
International Journal of Pressure Vessels and Piping, 29, pp 113153.
Herrara, R ., C arcagno, G ., L andes, J ., a nd Zhou, Z., 1992,
Predicting F ailure for In ternally P ressurised Pipes with Surface
Flaws, International Conference on Pipeline Reliability, Calgary,
Canada.
Hopkins, P., Corbin, P., 1988, A Study of Ex ternal D amage of
Pipelines, P aper 5, N G-18/EPRG Se venth J oint Biennial
Technical M eeting o n L ine P ipe Research , Cal gary, A lberta,
Canada.
Miller, A . G ., 1988, Review of L imit L oads of Structures
Containing Defects, I nternational J ournal of P ressure V essels
and Piping, 32.
Hopkins, P ., J ones, D . G ., a nd Clyne, A. C., 1989, The
Significance of Den ts an d Def ects i n T ransmission P ipelines,
Paper C 376/049, P roceedings In ternational Co nference on
Pipework, Engineering and Operation, IMechE, London.
Hagiwara, N., Oguchi, N., 1998, Fatigue B ehaviour of L ine
Pipes S ubjected t o S evere M echanical Dam age, Volume 1,
Proceedings o f S econd I nternational P ipeline C onference, IPC
1998, C algary, C anada, A merican Soc iety of Mechanical
Engineers, pp. 291-298.
Mayfield, M. E., W ilkowski, G . M., a nd Eibe r, R . J ., 1978,
Influence o f T oughness o n Resi stance t o M echanical Dam age
and Ability of Line P ipe to W ithstand Damage, Paper 2, AGAEPRG Line Pipe Research Seminar III, Houston, Texas.
Mayfield, M. E., Maxey, W. A ., a nd W ilkowski, G . M., 1979,
Fracture Initia tion T olerance of L ine Pipe, Paper F, 6th
Symposium o n L ine P ipe Research , A merican G as A ssociation,
Houston, Texas.
Eiber, R. J ., Ma xey, W. A ., Be rt, C. W., and McClure, G. M.,
1981, The Ef fects of D ents on the Fa ilure C haracteristics of
Linepipe, B attelle C olumbus L aboratories, N G-18, R eport N o.
125, AGA Catalogue No. L51403.
Maxey, W . A ., 1986, Outside For ce Defect Behaviour, 7th
Symposium on Line Pipe Research, Houston, Texas.
Ong, L. S., Soh, A . K., and Ong, J. H., 1992, Experimental and
Finite Element Investigation o f a L ocal Den t o n a P ressurised
Pipe, Journal of Strain Analysis, 27, pp. 177-185.
Bjrny, O. H., Rengrd, O., Fredheim, S., a nd Bruce, P., 2000,
Residual Strength of D ented P ipelines, D NV T est R esults,
Tenth I nternational C onference on O
ffshore a nd Polar
Engineering (ISOPE 2000), Seattle, USA.
Wang, K . C ., Sm ith, E. D ., 1985, The Ef fect of Mechanical
Damage on Fracture In itiation in Linepipe P art III - G ouge in a
Dent, C anadian C entre f or Mine ral a nd Energy Technology
(CANMET), Canada, Report ERP/PMRL 85-69 (TR).
Kiefner, J. F., Alexander, C. R., and Fowler, J. R., 1996, Repair
of Dents Containing Minor Sc ratches, P aper 9, 9 th Symposium
on L ine P ipe Research , P ipeline Research Committee of the
American Gas Association, Houston, Texas.
Alexander, C. R., Fowler, J. R., and Kiefner, J. F., 1997, Repair
of Dents Com bined w ith G ouges Conside ring Cy clic P ressure
Loading, Pipeline Engineering, American Society of Mechanical
Engineers, Houston, Texas, USA.

45. Hopkins, P., 1991, The Sig nificance of Mechanical Damage in

Gas Transmission P ipelines, P aper 25, V olume II, Proceedings
of EP RG/NG-18 Eig hth B iennial J oint T echnical Me eting on
Line Pipe Research, Paris, France.
46. Hopkins, P., Corder, I. and Corbin, P., 1992, The Resistance of
Gas Transmission Pipelines to Mechanical Damage, Paper VIII3, Inte rnational Conf erence on P ipeline Reliability, Calgary,
Canada.
47. Hopkins, P., Jones, D. G ., a nd C lyne, A . C ., 1983, Recent
Studies of the Significance of Mechanical Damage in Pipelines,
The American Gas Association an d E uropean P ipeline Research
Group Research Seminar V, San Francisco, USA.
48. Spiekhout, J ., G resnigt, A . M., K oning, C., and Wildschut, H.,
1986, Calculation Mode ls f or the Ev aluation of the R esistance
Against Mechanical Damage of P ipelines, 3R International, 25.
Jahrgang, Heft, pp 198-203.
49. Muntinga, T . G ., K oning, C ., 1990, Verification of External
Damage Mode ls by B urst T ests on P ipe Se ctions, P aper 13,
Proceedings o f In ternational P ipeline T echnology C onference,
Oostende, Belgium, pp. 13.25-13.32.
50. Lancaster, E. R., P almer, S. C ., 1992, Model T esting of
Mechanically D amaged P ipes C ontaining D ents a nd Gouges,
PVP-Vol. 235, D esign a nd A nalysis of P ressure Vessels, Piping
and Components, American Society of Mechanical Engineers, pp
143-148.
51. Lancaster, E. R ., P almer, S. C ., 1993, Assessment of
Mechanically D amaged P ipes C ontaining D ents a nd Gouges,
PVP-Vol. 261, Service Ex perience a nd L ife Ma nagement:
Nuclear, F ossil, an d P etrochemical P lants, A merican Society of
Mechanical Engineers, pp 61-68.
52. Lancaster, E. R ., P almer, S. C ., 1996, Burst Pressures of Pipes
Containing Dents and G ouges, P art E: J ournal of P rocess
Mechanical En gineering, P roceedings o f the Institution of
Mechanical Engineers, 210, pp. 19-27.
53. Alexander, C . R ., K iefner, J . F., 1997, Effects of Sm ooth a nd
Rock Dents on Liquid Petroleum Pipelines, Final Report to The
American P etroleum Institute , Stre ss Eng ineering Se rvices, Inc .,
and Kiefner and Associates, API Publication 1156, First Edition.
54. Leis, B. N., Bubenik, T. A., Francini, R. B., Nestleroth, J. B., and
Davis, R . J ., 2000, Recent Developments i n A voiding,
Detecting, an d A ssessing S everity o f M echanical Dam age,
Pipeline T echnology, V olume I I, P roceedings of the Third
International Pipeline Technology Conference, Brugge, Belgium,
R. Denys, Ed., Elsevier Science, pp. 405-425.
55. Leis, B . N ., Fr ancini, R . B ., 1999, Linepipe Resistance to
Outside Force V olume T wo: A ssessing Serv iceability o f
Mechanical Damage, Final Report on P roject PR 3-9305 to the
Line P ipe Re search Supe rvisory Com mittee of the Pipeline
Research Council International, Battelle.
56. Hopkins, P ., 1992, The A pplication of Fitne ss f or P urpose
Methods to D efects D etected in O ffshore Transmission
Pipelines, Conference on Welding and Weld Performance in the
Process Industry, London.
57. Rosenfeld, M. J ., 1998, Investigations of D ent R erounding
Behaviour, V olume 1, P roceedings of Se cond I nternational
Pipeline Conference, IPC-98, Calgary, Canada, American Society
of Mechanical Engineers, pp. 299-307.
58. Kastner, W., Rohric h, E., Sc hmitt, W. a nd Ste inbuch, R., 1981,
Critical Crack Sizes In Duc tile Piping, International Journal of
Pressure Vessels and Piping, 9, pp 197-219.

14

59. Schulze, H. D., Togler, G ., a nd B odman, E., 1980, Fracture

Mechanics Analysis on the Initia tion a nd P ropagation of

Circumferential and Longitudinal Defects in Str aight P ipes a nd

Pipe Bends, Nuclear Engineering and Design, 58, pp 19-31.

internal pressure (static)

longitudinally orientated

internal pressure (static)

circumferentially orientated

DNV-RP-F101[21]
corrosion

modified B31G[6,7]

Kastner local collapse solution[58]

RSTRENG[7]
NG-18 equations[5]
gouges

Kastner local collapse solution

PAFFC[8,9]

BS 7910 (or API 579)

BS 7910[1] (or API 579[4])

plain dents

empirical limits

kinked dents

no method1

smooth dents on welds

smooth dents and gouges
smooth dents and other types of defect
manufacturing defects in the pipe body2
girth weld defects
seam weld defects

no method
[11,56]

dent-gouge fracture model

dent-gouge fracture model

no method

NG-18 equations

Kastner local collapse solution

BS 7910 (or API 579)

BS 7910 (or API 579)

workmanship, EPRG[10]

BS 7910 (or API 579)

workmanship

BS 7910 (or API 579)

BS 7910 (or API 579)

cracking

PAFFC
BS 7910 (or API 579)

environmental cracking3
leak and rupture

no method

PAFFC
NG-18 equations
PAFFC

Schulze global collapse solution[59]

Note:
1.
No method represents both limitations in existing knowledge and circumstances where the available methods are too complex for inclusion in
a document such as PDAM.
2.
The term manufacturing defect covers a wide range of pipe body defect (laminations, inclusions, seams, cold shuts, gouges, plug scores, 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, it is then generally necessary to rely on workmanship limits and industry experience.
3.
Environmental cracking (stress corrosion cracking, hydrogen blisters, hydrogen stress cracking, etc.) can be very difficult to assess and cannot
necessarily be simply characterised as a crack-like defect.
Table 3 Recommended methods the Pipeline Defect Assessment Manual for assessing the burst strength of defects subject
to static internal pressure loading

15

REGULATIONS
IS A FITNESS-FORPURPOSE ASSESSMENT
APPROPRIATE?

TYPE OF
DEFECT/DAMAGE

DESIGN CODES
AND STANDARDS

YES

LOADS

IDENTIFY APPROPRIATE
SECTION OF THE
PIPELINE DEFECT
ASSESSMENT MANUAL

PIPE GEOMETRY

DEFECT
DIMENSIONS

CONSULT DEFECT
SPECIFIC FLOW CHART

CONSIDER
CONSEQUENCES
OF A FAILURE

ACCEPTANCE CRITERION
(SAFETY FACTOR)

CONSULT BACKGROUND
INFORMATION AS
NECESSARY

MINIMUM INFORMATION
REQUIRED TO
UNDERTAKE THE
ASSESSMENT

IDENTIFY DEFECT
ASSESSMENT METHOD

CONSULT DESCRIPTION
OF METHOD AS
NECESSARY

APPLICABILITY OF
METHOD

CONDUCT FITNESS-FORPURPOSE ASSESSMENT

1. STATIC LOADS
2. CYCLIC LOADS

MODEL UNCERTAINTY

NO

YES

IS THE DEFECT
ACCEPTABLE?

REFINE FITNESS-FOR-PURPOSE
ASSESSMENT, SEEK SPECIALIST
ASSISTANCE, OR TAKE
APPROPRIATE REMEDIAL ACTION

NO FURTHER
ASSESSMENT REQUIRED

DOCUMENT FITNESS-FORPURPOSE ASSESSMENT

Fig. 9 The role of the Pipeline Defect Assessment Manual in the fitness-for-purpose assessment of a pipeline defect

16

DENTED PIPELINE
Indications of low toughness include: old
linepipe, lin epipe not m anufactured to
API 5L, or an operating temperature less
than the DWTT transition temperature.

DOES THE LINE

PIPE HAVE A LOW
TOUGHNESS?

YES

SEEK SPECIALIST ADVICE

NO

IS THE DENT
KINKED?

YES

SEE CHAPTER 23

NO

DOES THE
DENT CONTAIN
ANY DEFECTS?

NO

SEE CHAPTER 22

YES

IS THE DEFECT
A GOUGE?

NO

SEE CHAPTER 26

YES

IS THE DENT
ON A WELD?

YES

SEEK SPECIALIST ADVICE

NO
IS THE DENT SUBJECT
TO LOADS OTHER THAN
INTERNAL OR
EXTERNAL PRESSURE?

YES

SEEK SPECIALIST ADVICE

NO

IS THERE
MORE THAN
ONE DEFECT?

YES

SEEK SPECIALIST ADVICE

YES

ARE THE DENT OR

THE DEFECT
CIRCUMFERENTIALLY
ORIENTATED?

YES

SEEK SPECIALIST ADVICE

NO

IS THE PIPELINE
PRESSURE
CYCLED?

YES

SEE SECTION 25.11

NO

SEE SECTION 25.10

Fig. 10 The assessment of a smooth dent containing a gouge

17