Journal of Pipeline Science and Engineering 1 (2021) 36–50

Contents lists available at ScienceDirect

Journal of Pipeline Science and Engineering

journal homepage:
http://www.keaipublishing.com/en/journals/journal-of-pipeline-science-and-engineering/

A comparative study of burst failure models for assessing remaining

strength of corroded pipelines
Xian-Kui Zhu
Materials Science and Engineering, Savannah River National Laboratory, Aiken, SC 29808, USA

a r t i c l e i n f o a b s t r a c t

Keywords: This paper performs a comparative study of remaining strength assessment models for corroded pipelines. A
Pipeline brief review of burst prediction models is first given for defect-free pipes, including strength solutions and flow
Corrosion defect solutions of burst pressure and experimental validations. Followed is a review of corrosion assessment models
Remaining strength
that are categorized into three generations in terms of the reference stress used in each model. Those corrosion
Burst pressure
models are then evaluated in comparison to full-scale burst data, with a focus on validating the newly proposed
Assessment model
third-generation models. Next, recent progresses are discussed, including the PRCI corrosion assessment projects,
constraint effect, bulging factor and defect width effect. Major technical challenges facing the corrosion model
improvement are finally discussed with regards to full-scale tests, numerical modeling, material failure criteria,
and real corrosion defects.

1. Introduction
Transmission pipeline system is an important infrastructure used
worldwide to transport crude oil or natural gas over long distances to
meet the increasing demands of energy. It has been showed that the
pipeline is an effective and safe vehicle for the oil and gas transporta-
tion. The transmission pipelines are made of carbon steels and most are
buried underground per the regulation requirements. Because of under-
ground water or sour soil, steel pipes are susceptible to external cor-
rosion that threatens the pipeline integrity. As such, pipeline corrosion
has been a major challenge facing the oil and gas industry. In addi-
tion to external corrosion attack, the buried pipelines under high pres-
sure may fail occasionally due to geohazards, natural forces, mechani-
cal damage, steel and weld quality, stress corrosion cracking, and other
anomalies.
Pipeline and Hazardous Materials Safety Administration (PHMSA) of
the U.S. Department of Transportation (DOT) reported the U.S. pipeline
incidents and the statistical trends over a period of 20 years at its
website (US DOT PHMSA 2018) and updates the database regularly.
Based on the PHMSA database of pipeline failure, Dai et al. (2017) sta-
tistically analyzed recent six-year significant incidents from 2010 to
2015 that contain 432 crude oil pipeline failures and 238 natural gas
pipeline failures. Fig. 1(a) and 1(b) show the failure causes that were
categorized for the oil pipelines and gas pipelines, respectively. As ev-
ident in those figures, the three major failure causes are corrosion,
pipe/weld material failure, and equipment failure for the oil pipelines,
while the three major failure causes are pipe/weld material failure, ex-
cavation damage, and corrosion for the gas pipelines. As a result, cor-
rosion is one of the major failures causes for both oil and gas pipelines,
and so corrosion assessment is critical to manage the buried pipeline
integrity.
Unprotected buried steel pipelines are highly susceptible to exter-
nal corrosion due to corrosive environments and different corrosion
mechanism, including sweet corrosion, sour corrosion and microbiolog-
ical induced corrosion (Ossai et al., 2015). Those corrosions weaken
the pipeline strength and degrade its loading-carrying capacity, and
eventually make steel pipelines deteriorated without a proper corro-
sion protection. To prevent pipelines from external corrosion, pipeline
coating and cathodic protection are required by regulations. Usually,
buried pipeline management needs to cope with other extreme geo-
logical conditions in a range of geohazards, such as ground shaking
in seismically active regions, land sliding, ground subsidence and set-
tlement, geological fault and freeze-thaw displacement (Nyman et al.,
2008; Porter et al., 2016). Because soil overburden pressure acting
on buried pipelines is insignificant compared to internal pressure
(Peng, 1978; American Lifeline Alliance, 2001), this work only consid-
ers internal pressure applied to thin wall pipelines containing corrosion
defects.
For determining the load-carrying capacity of pressure vessels,
many empirical, analytical or numerical methods have been devel-
oped for defect-free pressure vessels, see the review by Zhu (2016).
Christopher et al. (2002) evaluated a set of available burst pressure
prediction models for both thin-wall and thick-wall pressure vessels.
Law and Bowie (2007) assessed the burst pressure prediction models

E-mail address: Xiankui.Zhu@srs.gov

https://doi.org/10.1016/j.jpse.2021.01.008

2667-1433/© 2021 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
X.-K. Zhu
Fig. 1. Failure causes (Dai et al., 2017) of (a) oil and (b) gas pipelines.
for high-strength line pipes. More recently, Zhu and Leis (2012a) eval-
uated a series of burst pressure prediction models for a variety of
pipeline steels in terms of the strength theories and plastic flow the-
ories. The most accurate model was identified for defect-free line
pipes.
The natural gas industry began work to investigate the threat posed
by corrosion in the late 1960s (Keifner and Atterbury, 1971), with a
view to quantify when reinforcement or replacement of corroded pipes
was required. Pipe segments in need of replacement were removed dur-
ing rehabilitation work and then end capped and pressurized to fail-
ure. The experimental data were trended to quantify burst pressure as a
function of the length and depth of the metal-loss defects (Keifner and
Atterbury, 1971; Kiefner and Duffy, 1971), and the trending function
originated in the NG-18 equations (Maxey et al., 1972; Kiefner et al.,
1973) that were developed at Battelle in the early 1970s for deter-
mining fracture failure of pipelines containing cracks. Therefore, the
early burst pressure models are empirical in nature. Based on the NG-
18 equation for collapse-controlled failure, in the early 1980s, the oil
and gas industry developed the corrosion assessment code - ASME B31G
(ASME, 1991), and then modified it as Mod B31G (ASME, 2009) in 1989.
For a general pressure vessel containing blunt defects, Miller (1988) de-
livered a good review of early investigations on the limit load. For
pipelines containing corrosion defects, Cosham et al. (2007) collected
and analyzed the available corrosion assessment methods for corroded
pipelines. After a decade of numerous efforts, in the 1990s, the cor-
rosion assessment model was adopted in BS 7910 (British Standards
Institution, 2015; Cosham and Andrews, 2019), and a general fitness
for service code of API 579 (API/ASME, 2016) was developed for engi-
neering critical analysis of pressure vessels. Nevertheless, ASME B31G
(ASME, 1991) and ModB31G (ASME, 2009) remain in use today in the
daily pipeline integrity management, even though those methods are
known to be (overly) conservative.
Recently, Zhu (2018) summarized the primary existing assessment
methods for corrosion defects in pipelines and discussed several prac-
tical challenges facing the oil and gas industry. Zhu (2015a) discussed
the numerical approaches used to predict burst pressure of line pipes
37
Journal of Pipeline Science and Engineering 1 (2021) 36–50
with metal loss defects. Leis et al. (2017) presented their numerical re-
sults to minimize the B31G model uncertainty in the corrosion assess-
ment. Zhou and Huang (2012) assessed the model errors of corrosion
models. Amaya-Gomez et al. (2019) assessed the reliability of corrosion
methods, and Bhardwaj et al. (2020) quantified the uncertainty of burst
pressure models of corrode pipelines. However, those publications have
limit discussions on strain hardening effect, and a categorized analysis is
needed for better understanding the development of corrosion models.
So motivated, this paper performs a comparative study of remain-
ing strength assessment models for corroded pipelines. A brief review
of burst prediction models is first given to defect-free pipes and then to
corroded pipes. All burst models are critically evaluated in comparison
to full-scale burst data, with a focus on validating the newly proposed
third-generation models. Recent progresses are then presented, and ma-
jor technical challenges are finally discussed for improving the available
corrosion models for engineering applications.
2. Brief review of burst models for line pipes
The methods and models of the burst pressure prediction for defect-
free pipes are the basis for developing corrosion models for assessing
pipes containing metal-loss defects, where the reference stress of cor-
rosion model is determined by the burst pressure of defect-free pipe.
A recent review (Zhu, 2016) on the burst models, challenges and new
technologies for defect-free line pipes was published in the Journal of
Pressure Vessel Technology in its Special Issue for celebration of the
50th Golden Anniversary of ASME Pressure Vessel and Piping (PVP) Di-
vision. Those results are briefly reviewed here.
2.1. Strength models and experimental validation
For a large diameter, thin wall defect-free pipe, four strength models
(Zhu, 2016) of burst pressure were obtained in terms of the ultimate ten-
sile stress (UTS) and using the Tresca criterion (Tresca, 1864), von Mises
criterion (von Mises, 1913) and Zhu-Leis criterion (Zhu and Leis, 2006a)
as well as the flow stress criterion (Kiefner et al., 1973).
X.-K. Zhu
(1) Tresca strength solution:
2𝑡
𝑃0 = 𝜎 (1)
𝐷 uts
(2) von Mises strength solution:
4𝑡
𝑃M0 = √ 𝜎uts
3𝐷
(3) Zhu-Leis strength solution:
( )
1 2 2𝑡
𝑃ZL0 = 1+ √ 𝜎uts
2 3 𝐷
(4) Flow stress-based failure pressure:
2𝑡
𝑃f0 = 𝜎 (4)
𝐷 f low
where D is the outside diameter, 𝐷̄ is the mean diameter, and t is the wall
thickness. The flow stress 𝜎 flow = (𝜎 ys +𝜎 uts )/2, in which 𝜎 ys is the yield
stress (YS) and 𝜎 uts is the UTS. Zhu-Leis strength solution in Eq. (3) was
determined using a new multi-axial strength theory – average shear
stress criterion developed by Zhu and Leis, (2006a, 2004). Eqs. (1) to
(3) show that the Zhu-Leis strength solution is equal to the average of
Tresca and Mises strength solutions.
To evaluate the accuracy of four strength models in Eqs. (1) to (4), ex-
perimental data of burst pressure (Pb ) for more than 100 full-scale tests
of thin-wall cylindrical pressure vessels made of carbon steels were used
to compare with the burst pressure predictions from the four strength
criteria, as shown in Fig. 2. All materials used in the experiments are
ductile steels with a wide range of strain hardening exponent n from
0.02 to 0.18 that covers low to high-strength pipeline steels. Experi-
mental details and References cited in Fig. 2 were described by Zhu and
Leis, 2006a. All points in the figure are experimental data, and all lines
are burst pressure predictions.
As shown in Fig. 2, three circles contain three groups of test data.
The purple, blue and red circles represent test data of burst pressure for
low, intermediate, and high strength carbon steels, respectively. Four
dash lines in red, green, blue, and yellow denote the burst pressure
predictions, respectively by the von Mises, Zhu-Leis, Tresca, and flow
stress criteria. This figure shows that (1) von Mises strength criterion
overestimates burst pressure for all steels, (2) Zhu-Leis strength crite-
rion is good for high-strength steels but overestimated for low-strength
steels, (3) Tresca strength criterion adequately predicts burst pressure
for intermediate-strength steels, and (4) the flow stress criterion predicts
the most conservative results for all steels.
n Eq. (6) and von Mises flow solution in Eq. (7). This figure shows that
Fig. 2. Comparison of experimental data of burst pressure and predictions by
von Mises, Zhu-Leis, Tresca, and flow stress criteria for various pipeline steels
(Zhu, 2016).
Journal of Pipeline Science and Engineering 1 (2021) 36–50
(2)
(3)
Fig. 3. Comparison of three flow solutions and experimental data for various
carbon steels (Zhu and Leis, 2006a).
As evident in Fig. 2, experimental data of burst pressure are func-
tions of strain hardening exponent, and the normalized burst pressure
decreases with the increase of strain hardening rate. Except for the flow
stress criterion, the other three strength criteria do not consider the ef-
fect of strain hardening rate on burst pressure. To improve those strength
solutions, Zhu (2016) recommended the use of flow theories to develop
more accurate burst prediction models in reference to two material pa-
rameters.
2.2. Flow models and experimental validation
The Tresca theory (Tresca, 1864) and von Mises theory
(von Mises, 1913) are two classic theories of plasticity developed
to describe the nonlinear plastic deformation in a metal during loading
and determine two bound solutions of burst pressure for the same pipe.
To predict more accurate burst failure, the present author (Zhu and
Leis, 2006a; Zhu and Leis, 2004) proposed a new mechanics concept
of “average shear stress”, and developed an associated multi-axial flow
theory of plasticity, i.e. average shear stress flow theory that was often
referred to as Zhu-Leis flow theory in public literature. For a power-law
hardening material, using the new flow theory, Zhu and Leis, (2006a,
2004) obtained a new flow solution of burst pressure. Accordingly,
three flow solutions of burst pressure were determined for defect-free
pipes in terms of the Tresca, Zhu-Leis and von Mises flow theories as:
( )𝑛+1
1 4𝑡
𝑃T = 𝜎uts (5)
2 𝐷
( √ )𝑛+1
2+ 3 4𝑡
𝑃A = √ 𝜎uts (6)
4 3 𝐷
( )𝑛+1
1 4𝑡
𝑃M = √ 𝜎uts (7)
3 𝐷
where n is the strain hardening exponent and measured from a tensile
test or estimated from the Y/T ratio (Zhu and Leis, 2006a; Zhu and
Leis, 2007).
The experimental data of burst pressure shown Fig. 2 is reused to
assess and validate the flow theories. Fig. 3 compares the experimen-
tal data with Tresca flow solution in Eq. (5), Zhu-Leis flow solution in
(1) all test data and predictions of burst pressure are functions of strain
hardening exponent in the similar trend, (2) von Mises flow solution
gives an upper bound prediction, (3) Tresca flow solution gives a lower
bound prediction, and (4) Zhu-Leis flow solution predicts a better result
38
X.-K. Zhu
that matches well with the burst data on average. Thus, those experi-
ments validate the proposed Zhu-Leis flow theory and the associate flow
solution of burst pressure for defect-free pipes.
Using other full-scale test data and the statistical analysis,
Seghier et al. (2018) confirmed that the average shear stress yield cri-
terion is the best plastic flow criterion to predict burst pressure of in-
tact pipes for different steel grades ranging from X46 to X100. Like-
wise, Zimmermann et al. (2007a, 2007b), Zhou and Huang (2012) and
Bony et al.( 2010), Alamilla et al. (2013) among others assessed the
available burst prediction models and concluded that the Zhu-Leis flow
solution is the most accurate for predicting burst pressure of line pipes.
Thus, Zhu-Leis flow theory (Zhu and Leis, 2006a) not only fills in the
technical gap between Tresca and von Mises flow theories, but also de-
termines a more accurate burst pressure.
Recently, the Zhu-Leis criterion has obtained extensive applica-
tions in the corrosion assessment and pipeline integrity manage-
ment (Amaya-Gomez et al., 2019; Bhardwaj et al., 2020). In addi-
tion, Chen et al. (2015) adopted the Zhu-Leis flow theory and de-
termined the burst pressure of pipes with geometric eccentricity.
Chen et al. (2020) extended the Zhu-Leis flow theory to determine the
burst pressure of cylindrical explosion containment vessels.
2.3. Numerical validation of new flow solution
Because all finite element analysis (FEA) software codes, including
ABAQUS (ABAQUS Standard 2018) determine the von Mises-based burst
pressure as in Eq. (7) at plastic instability that for a defect-free pipes. To
obtain a more accurate result, Zhu and Leis (2007) developed a Mises
stress-based failure criterion in reference to the Zhu-Leis flow solution
and von Mises flow theory. Fig. 4 shows the FEA results of von Mises ef-
fective stress versus the normalized internal pressure that were obtained
in Ref Zhu and Leis, 2007 for an X65 defect-free pipe using a modified
RIKS model (Riks, 1979) that was built in ABAQUS (ABAQUS Standard
2018). Also included in this figure are two theoretical solutions deter-
mined using the von Mises flow theory and the average shear stress yield
(ASSY) criterion or Zhu-Leis flow theory, respectively. This figure shows
that (1) von Mises effective stress at the global geometric instability is
higher at about 18% than the Zhu-Leis effective stress, (2) the von Mises
burst pressure is higher at about 8% than the Zhu-Leis burst pressure,
and (3) the proposed Mises stress-based failure criterion determines a
better numerical prediction that agrees well with the Zhu-Leis flow so-
lution. As a result, the proposed material failure criterion is adequate to
be used in ABAQUS or any other FEA software to accurately determine
the burst pressure defined by the Zhu-Leis flow theory.
Recently, four FEA calculations were performed on a X65 pipe of 30
inch-diameter and 0.5 inch-wall thickness, where the yield stress is fixed
at 72 ksi but the UTS varies, leading to the stain hardening exponent n
Fig. 4. Variation of Mises stress with pressure for X65 (Zhu and Leis, 2007).
39
Journal of Pipeline Science and Engineering 1 (2021) 36–50
Fig. 5. Comparison of FEA results and theoretical solutions for X65 over a large
range of strain hardening exponent n.
values equal to 0.05, 0.1, 0.15 to 0.2. Using the material failure crite-
rion (Zhu and Leis, 2007) and the ABAQUS RIKS model (Riks, 1979), the
FEA results of Zhu-Leis and von Mises failure pressures (Pf ) were deter-
mined and compared with the theoretical solutions, as shown in Fig. 5.
It shows that the FEA results and theoretical solutions agree very well
over the large range of strain hardening rates for the two flow theories.
This numerically validates the Zhu-Leis flow solution of burst pressure
and is complementary to the experimental validation in Fig. 4.
3. Corrosion assessment models
Real corrosion defects formed in an ageing pipeline can be very com-
plicated in size, shape, location and orientation. Only axially oriented
isolated metal-loss defects subject to internal pressure are considered in
this work. Other anomalies and loading conditions will not be consid-
ered.
A general expression of failure stress or pressure of a corrosion defect
at burst failure can be written as:
𝑆f = 𝑆R × 𝑓 (defect geometry) (8)
𝑃b =𝑆R 2𝑡∕𝐷 × 𝑓 (defect geometry) (9)
where Sf denotes the hoop stress at burst failure, SR is a reference stress,
and f(defect geometry) is a function of defect geometry (defect depth d,
length L and width W). In many cases, defect width has a small effect and
√
f(defect geometry) can be simplified as f(d/t, L/ (Dt)). Absent the effect
of a defect, Eq. (8) must predict the failure stress of the corresponding
defect-free pipe, and thus the SR must be equal to the failure stress of the
defect-free pipe. Note that Eq. (9) is consistent with the notation used
in API 579 (Cosham and Andrews, 2019) for the Level-1 defect-severity
criterion.
All corrosion models are categorized here into three generations. The
first generation was empirically developed in the 1960s to 1970s with
SR =the flow stress, the second one was numerically developed in the
1990s to 2000s with SR =the UTS, and the third one remains in develop-
ing with SR =f(UTS, n), a function of the UTS and the strain hardening
exponent n (or the YS). Those corrosion models are reviewed below.
3.1. The first-generation models (1960–1970s)
3.1.1. ASME B31G
Based on full-scale test data obtained in the early 1970s for pipeline
steels up to X65, Maxey et al. (1972) and Kiefner et al. (1973) developed
a set of semi-empirical equations (i.e., NG-18 equations) to estimate the
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

burst pressure of line pipes with cracks. Based on NG-18 equation for defect length L. This method allows to determine more accurate cor-
flow-controlled failure, a corrosion assessment model B31G was devel- roded area using the discrete approach, and thus to determine more ac-
oped in 1981 by American Society for Mechanical Engineers (ASME), curate burst pressure of a real corrosion defect. A personal-computer
and updated periodically, such as the versions in 1991 (ASME, 1991) code called RSTRENG (Kiefner and Vieth, 1989) was developed by
and 2009 (ASME, 2009). For a short defect, corrosion area was assumed Pipeline Research Council International (PRCI) for predicting more ac-
to be parabolic in shape with a curved bottom, and the burst failure curate burst pressure of a corroded pipe. This effective area model was
pressure of the corroded pipeline is estimated by: adopted in the Level-2 procedure of B31G (ASME, 2009) and API 579
[ ] √ (API/ASME, 2016).
2 𝑡 𝜎f 1 − (2∕3)(𝑑∕𝑡)
𝑝b = , for 𝐿 ≤ 20𝐷𝑡 (10) For a machined rectangular defect with a uniform depth d, Eq. (14)
𝐷 1 − (2∕3)(𝑑∕𝑡)∕𝑀
becomes a simple one:
where the crack depth is limited within 80% of t, or d/t ≤ 0.8. [ ]
2 𝑡 𝜎f 1 − (𝑑∕𝑡)
For a long corrosion defect, the corrosion area was assumed to have 𝑃b = (16)
𝐷 1 − (𝑑∕𝑡)∕𝑀
a rectangular shape with a uniformed bottom, and the failure pressure
in Eq. (10) is simplified as: If a corrosion defect is long enough, 1/M approaches zero and
2 𝑡 𝜎f [ 𝑑
] √ Eq. (16) reduces to Eq. (11). In 1992, Ritchie and Last (1995) adapted
𝑝b = 1− , for 𝐿 > 20𝐷𝑡 (11) Eq. (16) as a corrosion defect acceptance criterion for Shell Oil Com-
𝐷 𝑡
pany that was named as Shell-92, where the flow stress 𝜎𝑓 = 0.9𝜎uts and
In both equations above, the flow stress 𝜎f = 1.1 SMYS, where SMYS is the maximum defect depth was allowed d/t ≤ 0.85. The Shell-92 model
the specified minimum yield stress defined in API 5L (API, 2018), d is first used the UTS.
the defect depth, L is the total defect length, and M is a Folia’s bulging
factor (Folias, 1964; Folias, 1965) and expressed as:
3.1.4. CSA Z662
√
√ ( )2 Based on Eq. (16) and the Shell-92 assessment criterion, Canadian
√
√ 𝐿
√
𝑀 = 1 + 0.8 √ (12) Standards Association (CSA) developed their corrosion assessment code
𝐷𝑡 – CSA Z662 (CSA, 2019). In this code, the burst pressure of corrosion
defects was predicted by:
ASME B31G model was calibrated with the vintage pipeline grades
[ ( ) ]
up to X65, and thus may be inadequate to use for modern pipeline steels 2𝑡𝜎f 1 − 𝑑ave ∕𝑡
like X80 or X100. 𝑃b = ( ) (17)
𝐷 1 − 𝑑ave ∕𝑡 ∕𝑀

3.1.2. Mod B31G (0.85 dL) where dave is the average defect depth, the bulging factor M is the same
Practical applications showed that ASME B31G model can be overly as Eq. (14), and the flow stress is defined as:
conservative (Coulson and Worthingham, 1990), and thus a mod B31G { ( )
was proposed in 1989 by Kiefner and Vieth (Kiefner and Vieth, 1989; 1.15𝜎ys , SMYS ≤ 241 MPa or 35 ksi
𝜎𝑓 = ( ) (18)
Kiefner et al., 1996) in the associated software RSTRENG as follows: 0.9𝜎uts , SMYS > 241 MPa or 35 ksi
[ ]
2 𝑡 𝜎f 1 − 0.85(𝑑∕𝑡)
𝑝b = (13) 3.1.5. Comparison of the first-generation models
𝐷 1 − 0.85(𝑑∕𝑡)∕𝑀
Figs. 6 and 7 compare the first-generation (1st-G) models discussed
where the defect depth is limited within 80% of t or d/t ≤ 0.8, the flow above for a fixed uniform defect depth of d/t=0.5 and for two lower
stress was redefined as𝜎f = SMYS + 69 (MPa), and the bulging factor M strength steels X52 and GrB, where the x-axis denotes the normalized
was replaced by: defect length, and the y-axis denotes the normalized burst pressure with
√ Pb the model predicted burst pressure and P0 the Tresca strength solu-
√ ( )2 ( )4
√ √
√ 𝐿 𝐿 tion in Eq. (1). In those figures, the 1st-G models include ASME B31G,
√
𝑀 = 1 + 0.6275 √ − 0.003375 √ , if 𝐿 ≤ 50𝐷𝑡 (14a)
Mod B31G, RSTRENG, Shell-92 and CSA Z662 criteria. The material
𝐷𝑡 𝐷𝑡
properties were assumed to be the SMYS and the specified minimum
( )2 tensile stress (SMTS), see API 5L (API, 2018).
𝐿 √
𝑀 = 3.3 + 0.032 √ , if 𝐿 > 50𝐷𝑡 (14b) For X52, Fig. 6 shows that the Mod B31G model predicts the highest
𝐷𝑡 results of burst pressure for all given defect lengths; the CSA Z662 and
Note that 1) the actual yield stress was used in the original NG-18
equations, 2) the use of SMYS makes B31G and Mod B31G more con-
venient and conservative for practical applications, 3) replacement of
2/3 factor in B31G using the 0.85 factor in Mod B31G reduces model
conservatism. The Mod B31G model was accepted in the newer version
of ASME B31G-2009 (ASME, 2009) and in the Level-1 procedure of the
general FFS code API 579-2016 (API/ASME, 2016). Ma et al., 2011 com-
pared different versions of B31G.

3.1.3. PRCI RSTRENG (effective area model)

For better approximating a real corrosion area with a river bottom
profile, an effective area method was proposed in 1989 by Kiefner and
Vieth (1989) to estimate the remaining strength:
[ ]
2 𝑡 𝜎f 1 − 𝐴 d ∕ 𝐴 0
𝑃b = (15)
𝐷 1 − 𝐴d ∕𝐴0 𝑀
where the flow stress and the bulging factor are the same as defined
in Mod B31G, Ad denotes the effective area of a complex corrosion de-
fect profile, and A0 =tL is the axial cross-section area with the maximum Fig. 6. Comparison of the 1st-G corrosion models for X52.

40
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

Therefore, both LPC and PCORRC criteria are identical for long corro-
sion defects. If a defect is absent, Eq. (22) reduce to Eq. (1), the Tresca
strength solution for defect-free pipes. This implies that both LPC and
PCORRC models have the basis of the Tresca criterion with one single
material parameter.

3.2.3. Choi limit load

Choi et al. (2003) and Kim et al. (2002) carried out a series of full-
scale experimental tests on machined metal-loss defects for X65 pipeline
steel and performed the detailed 3D elastic-plastic FEA calculations of
line pipes with blunt defects, where all axial defects were assumed to be
elliptically curved in the circumferential cross-section. By trending the
FEA outcomes, they obtained the following pressure limit load:
⎧ [ ( ) ( )2 ] ( )
⎪0.9𝜎 2𝑡 𝐶 + 𝐶 √𝐿 + 𝐶2 √ 𝐿
, for √𝐿 <6
⎪ uts 𝐷
𝑖
0 1
Rt Rt Rt
𝑃𝑏 = ⎨ [ ( )] ( ) (23)
Fig. 7. Comparison of the 1st-G corrosion models for GrB. ⎪ 𝜎uts 𝐷2𝑡 𝐶3 + 𝐶4 √𝐿 , for √𝐿 ≥6
⎪ 𝑖 Rt Rt
⎩
where Di is the internal diameter of the pipe, R is the mean radius of the
Shell-92 are identical with predictions slightly lower than the RSTRENG
pipe, and C0 , C1 , C2 , C3 , C4 are given by:
results; and ASME B31G predicts a comparable result to Mod B31G for a
short corrosion defect, but the lowest result for a long corrosion defect. 𝐶0 = 0.06(𝑑∕𝑡)2 − 0.1035(𝑑∕𝑡) + 1.0
For GrB, Fig. 7 shows that the Shell-92 model predicts the highest 𝐶1 = −0.6913(𝑑∕𝑡)2 + 0.4548(𝑑∕𝑡) − 0.1477
results of burst pressure over the given defect length, and followed in 𝐶2 = 0.1163(𝑑∕𝑡)2 − 0.1053(𝑑∕𝑡) + 0.0292 (24)
order are Mod B31G, RSTRENG, and CSA Z662 models. ASME B31G 𝐶3 = −0.9847(𝑑∕𝑡) + 1.1101
model may predict a comparable result to RSTRENG for a short corro- 𝐶4 = 0.0071(𝑑∕𝑡) − 0.0126
sion defect, but the lowest result for a long corrosion defect. Absent a metal-loss defect, Eq. (23) reduces to SR =0.9𝜎 uts or
Pb =0.9𝜎 uts × 2t/Di for defect-free pipes. This burst pressure is less
3.2. The second-generation models (1990s–2000s) than the Tresca strength solution in Eq. (1), and thus the limit load in
Eq. (23) can be conservative for some steel grades.
3.2.1. LPC model
Based on a burst test database and a large amount of elastic-plastic 3.2.4. Recalibrated PCORRC model
FEA numerical results for X65 corroded pipes, Fu and Kirkwood (1995), Recently, based on full-scale experimental test data and a set of
Batte et al. (1997), Fu and Batte (1999) at British Gas in 1995 developed 3D FEA calculations for X70 corroded pipes and for X70 corroded
a line pipe corrosion (LPC) criterion to determine burst pressure: girth/seam welds, Yeom et al. (2015, 2016) in 2015 recalibrated the
( ) PCORRC model:
2𝑡𝜎uts 1 − 𝑑∕𝑡 [ ( ( ))]
𝑃b = (19) 2𝑡 𝑑 𝐿
𝐷−𝑡 1 − 𝑑∕Qt 𝑝b = 0.9𝜎uts × 1 − 1 − exp −0.224 √ (25)
𝐷 𝑡 𝐷(𝑡 − 𝑑 )∕2
where the parameter Q is a curve-fit bulging factor from the FEA results
Absent a corrosion defect, again, Eq. (25) reduces to Pb =0.9𝜎 uts × (2t/D)
as:
√ for defect-free pipes. This outcome is less than the Zhu-Leis strength at
√ ( )2
√ about 16% and the Tresca strength at 10% for X70 defect-free pipe.
√ 𝐿
√
𝑄 = 1 + 0.31 √ (20)
𝐷𝑡 3.2.5. Comparison of the second-generation models
Fig. 8 compares the second-generation (2nd-G) models discussed
The LPC criterion was adopted into DNV RP-F-101 (Det Norske above for a fixed uniform defect depth of d/t=0.5 and for a moderate
Veritas 2015) in 1999 with SR =0.9UTS and statistical factors, and strength steel X65, where the material properties were assumed to be the
then adopted in BS7910-1999 (British Standards Institution, 2015; SMYS and SMTS. In this figure, the 2nd-G models include LPC, BS7910,
Cosham and Andrews, 2019) with SR =(𝜎 ys +𝜎 uts )/2. Note that BS7910- PCORRC, Limit load by Choie (2003), and Reformulated PCORRC by
2015 is the newest version. Yeom (2015). For comparison, RSTRENG and CSA Z662, are also in-
cluded in Fig. 8. This figure shows that 1) LPC and PCORRC predict
3.2.2. PCORRC model comparable results, 2) BS7910 and RSTRENG predict similar results that
In parallel to the development of the LPC model, Leis and are lower than the LPC predictions, 3) the CSA Z662 and Yeom (2015)
Stephens (1997), Stephens et al. (1999), Stephens and Leis (2000) at predict comparable results that are lower than RSTRENG predictions,
Battelle made a similar effort to improve the flow stress-based corrosion and 4) Choie (2003) and Yeom (2015) are comparable for short defects,
models. Based on the FEA results and experimental data of corrosion and become almost identical to be the lower bound for long defects.
defects for X52 to X70, those authors in 1997 developed the PCORRC
model: 3.3. The third-generation models (2000s to present)
[ ( ( ))]
2𝑡𝜎uts 𝑑 𝐿 In the first two generation models as discussed above, the reference
𝑃𝑏 = 1− 1 − exp −0.157 √ (21)
𝐷 𝑡 𝐷(𝑡 − 𝑑 )∕2 stress SR is one single material strength parameter, either a flow stress
or the UTS that corresponds to the strength criteria in Section 2.1 for
For a very long metal-loss defect, Eqs (19) and (21) reduce to the defect-free pipes. As shown in Section 2.2, a more reasonable reference
simple equation: stress SR should contain two material parameters, i.e., the UTS and strain
2𝑡𝜎uts ( 𝑑
) hardening exponent n (or the YS): SR = f(UTS, n or YS). This section
𝑝b = 1− (22) reviews the corrosion models containing the two material parameters.
𝐷 𝑡

41
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

3.3.3. Zhu-Leis flow model coupling with PCORRC

In order to improve the Mod PCORRC in Eq. (26) for a corroded
pipeline, Zhu (2019) in 2018 obtained another M-PCORRC by utilizing
the Zhu-Leis flow solution and the geometric term of PCORRC. This M-
PCORRC is expressed as:
( √ )𝑛+1 [ ( ( ))]
2+ 3 4𝑡𝜎uts 𝑑 0.157𝐿
𝑝b = √ × 1− 1 − exp − √ (28)
4 3 𝐷 𝑡 𝐷(𝑡 − 𝑑 )∕2

3.3.4. Zhu-Leis flow model coupling with LPC

Similarly, using the Zhu-Leis flow solution and the geometric term
of LPC, the following Mod LPC is obtained for predicting burst pressure
of a corroded pipe as:
( √ )𝑛+1 ( )
2+ 3 4𝑡𝜎uts 1 − 𝑑∕𝑡
𝑝b = √ (29)
4 3 𝐷 1 − 𝑑∕(Qt)

Fig. 8. Comparison of the 2nd-G corrosion models for X65. 3.3.5. Tresca flow model coupling with ASME B31.G
More recently in 2018, Abdelghani et al. (2018) adopted the Tresca
flow strength as the reference stress SR and coupled this SR with the
3.3.1. Mises flow model coupling with PCORRC geometric term of ASME B31.G in Eq. (15). Using their FEA results, the
In the 2000s, more experimental analyses and numerical investi- bulging factor was recalibrated. And then, they obtained the following
gations of corroded pipes have been conducted (Yeom et al., 2015; reformulated B31G model for predicting the burst pressure of a corroded
Yeom et al., 2016; Benjamin and Andrade, 2003; Souza, 2004; pipeline:
Souza et al., 2007; Zhu and Leis, 2005; Shuai, 2010; Besel and Zim- ( )𝑛 2𝑡𝜎 1 − 𝑑∕𝑡
1 uts
merman, 2010; Chiodo and Ruggieri, 2009; Zhu and Leis, 2012b; 𝑃b = (30)
2 𝐷 1 − 𝑑∕(tM)
Zhu, 2015b; Gao et al., 2019; Chen et al., 2015; Karstensen and
where the bulging factor was recalibrated as:
Smith, 2001; Noronha et al., 2002; Chauhan and Brister, 2009), with √
√ ( )2 ( )4
a focus on the high-strength pipeline grades X70, X80 and X100. For √
√ 𝐿 𝐿
those high-strength steels, the plastic deformation local to a corrosion √
𝑀 = 1 + 0.45 √ + 0.050625 √ (31)
defect is significant at collapse due to strain hardening and large plas- 𝐷𝑡 𝐷𝑡
tic strain. In order to quantify the effect of strain hardening, Zhu and
3.3.6. Analytic model for long corrosion defects
Leis (2005) started to use the von Mises flow theory to characterize the
For long corrosion defects, both LPC and PCORRC methods reduce
plastic collapse of line pipes, and the Mise flow solution of burst pres-
to Eq. (22) that is the Tresca strength solution coupling with a geomet-
sure was obtained as Eq. (7). After combining the reference stress SR
ric factor of (1-d/t) due to a blunt defect. In a similar way, Zhu and
from Eq. (7) and the defect geometric term of the PCORRC in Eq. (20),
Leis (2012b) and Zhu (2015b) simply extended the plastic flow solu-
they in 2005 obtained a modified PCORRC model:
tions in Eqs. (5) to (7) for the defect-free pipes to a long corrosion defect
4 𝑡 (L>D) as:
𝑝d = ( ) 𝜎uts ×
(√ )1+0.239 1 −1 0.596 𝐷 (1) Tresca flow solution for a long corrosion defect:
𝑌 ∕𝑇
3 ( )𝑛+1 4𝑡𝜎 ( )
1 𝑑
[ ( ( ))] 𝑃T = uts
1− (32)
𝑑 𝐿 2 𝐷 𝑡
1− 1 − exp −0.157 √ (26) (2) Zhu-Leis flow solution for a long corrosion defect:
𝑡 𝐷(𝑡 − 𝑑 )∕2
( √ )𝑛+1
2+ 3 4𝑡𝜎uts ( 𝑑
)
This is the first 3rd -G model that helped us understand the effect of 𝑃ZL = √ 1− (33)
4 3 𝐷 𝑡
the UTS and YS on the collapse of a defect in carbon steels. Because
of the nature of non-conservatism of von Mises solution, the Zhu-2005 (3) Mises flow solution for a long corrosion defect:
( )𝑛+1
model in Eq. (26) may overestimate burst pressure for corroded pipes.
1 4𝑡𝜎uts ( 𝑑
)
𝑃M = √ 1− (34)
3 𝐷 𝑡
3.3.2. Mises flow model coupling with reformulated PCORRC Using full-scale experimental burst data (Souza, 2004; Souza et al.,
In 2013, Ma et al. (2013) also adopted the von Mises flow strength 2007), Zhu and Leis (2012b) and Zhu (2015b) confirmed that the Zhu-
as the reference stress SR and coupled it with the geometric term of Leis solution in Eq. (33) is a better prediction of burst pressure for long
PCORRC. After recalibration for various machined defects in X70 and corrosion defects. Thus, Eq. (33) can be regarded as a 3rd-G corrosion
X80 pipes, these authors obtained a reformulated PCORRC model in model for a very long defect. Fig. 9 compares six corrosion criteria and
the following equation for determining the burst pressure for a high- the variations of their predictions with pipeline grades for a long corro-
strength pipeline: sion defect, where all predictions are normalized by the PCORRC result
(Ppcorrc ). This figure shows that 1) the Mises criterion in Eq. (34) over-
4 𝑡
𝑝 d = (√ ) 𝜎 × predicts the burst pressures for all grades, 2) the Tresca criterion in
1+𝑛 𝐷 uts
3 Eq. (32) underestimates the burst pressure for all grades, 3) the PCORRC
[ ( ( ) )] model slightly underpredicts the burst pressure in comparison to the
( )
𝑑 −0.4174𝐿 𝑑 −0.1151 Zhu-Leis solution in Eq. (33), and (4) ASME B31G and RSTRENG sig-
1− 1 − 0.7501 exp √ 1− (27)
𝑡 Dt 𝑡 nificantly underestimate the burst pressure for low and intermediate
strength steels with grade under X70, but their predictions are compa-
Comparison of Eqs. (27) and (26) shows the geometric term in Eq. (27) is rable to the PCORRC results for high strength steels with grade X80 and
quite different from that in the PCORRC model. above.

42
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

d/t

Pb / P0
p
p

L / (Dt)^0.5

Fig. 11. Comparison of LPC, PCORRC and their modified models for X70.
Fig. 9. Variation of different models with pipeline grades for long corrosion
defect.
above for a fixed uniform defect depth of d/t=0.5 and for a high
3.3.7. A new corrosion model strength steel X70, where the material properties were assumed to be
In reference to the Zhu-Leis flow solution in Eq. (31) for a long defect, the SMYS and SMTS. In this figure, the 3rd-G models include The first
Zhu (2015a) assumed a general format of burst pressure for a finite-sized Mod PCORRC model by Zhu-2005, a reformulated PCORRC model by
corrosion defect as follows: Ma-2013, a new corrosion model by Zhu-2015, another Mod PCORRC
( by Zhu in 2018, a Mod LPC model, a Ref-B31G model by Abdelghani
√ )𝑛+1 [ ( )]
2+ 3 4𝑡𝜎uts 𝑑 𝐿 in 2018. In order to comparison with the 2nd-G models, the LPC and
𝑃b = √ 1− 𝑓 √ (35) PCORRC are included in Fig. 10. This figure shows that 1) the Zhu-2005
4 3 𝐷 𝑡 Dt
is an upper bound solution, 2) the Ref-B31G is a lower bound solution,
√
where f(L/ (Dt)) is a function of defect length and was determined from and 3) all other models determine comparable results. The two bound
treading FEA outcomes. For an X65 steel, the FEA results of burst pres- models determine two extreme solutions, either a non-conservative re-
sure were obtained for the simulated metal-loss defects, and the data sult or an over-conservative result.
regression determined the f function in Eq. (35). Accordingly, a new Fig. 11 compares the Mod-LPC with LPC and the M-PCORRC with
corrosion model for predicting more accurate burst pressure was devel- PCORRC. The new model Zhu-2015 is also included in this figure. It is
oped as: observed from this figure that 1) the Mod-LPC is shifted up from the LPC
in a factor of 1.03, 2) the Mod-PCORRC is shifted up from the PCORRC
( √ )𝑛+1
2+ 3 4𝑡𝜎uts ⎡⎢ 𝑑
⎛ ⎛
𝐿 in the same factor of 1.03, and 3) the two modified models are close
𝑃b = √ 1 − ⎜1 − 1∕⎜1 + 0.1385 √ to the Zhu-2015 model. This comparison implies that the strain hard-
4 3 𝐷 ⎢ 𝑡 ⎜ ⎜ Dt
⎣ ⎝ ⎝ ening effect on the burst pressure is insignificant for the X70 using the
( )2
⎞⎞⎤ minimum specified material properties.
𝐿 ⎟⎟⎥
+ 0.1357 √ (36)
Dt ⎟⎠⎟⎠⎥⎦
3.4. Validation of corrosion assessment models
Note that the defect geometric function in the equation above is com-
pletely different from those for the LPC and PCORRC, and very easy for 3.4.1. Evaluation of the first two generation models
engineers to understand and to use. The first two generation models are frequently used in the integrity
assessment of aging pipelines. To better understand the difference of
3.3.8. Comparison of the third-generation models available models, the PHMSA office in the U.S. DOT sponsored a large
Fig. 10 compares the third-generation (3rd-G) models discussed project (Chauhan and Brister, 2009) in the mid-2000s to thoroughly
evaluate those corrosion models. A large integrated test database that
contains 313 full-scale vessel and ring expansion test data for real and
machined metal loss defects was collected for line pipes ranging from
d/t
Grade A25 to X100. The experimental tests were carried out by differ-
ent organizations over more than fifty years in the oil and gas industry.
The metal-loss defects were axially isolated, only internal pressure was
applied with known pipe geometries, defect sizes, and material proper-
ties. Actual material properties were used as needed in the calculations
Pb / P0

of failure pressure.
Fig. 12(a) and 12(b) compare the actual and predicted results of
failure pressure using the B31G model and Mod B31G model. Fig. 13
compares experimental data and predicted results using the RSTRENG.
Fig. 14(a) and 14(b) compares experimental data and predicted results
using the LPC and PCORRC. All those figures were taken from Report
(Chauhan and Brister, 2009), where the y-axis denotes actual measured
failure pressure (Pt ) divided by the model predicted failure pressure
L / (Dt)^0.5 (Pm ): Pt /Pm . The scatter of Pt /Pm ratio is a measure of the model ac-
curacy.
Fig. 10. Comparison of the 3rd-G corrosion models for X70.
43
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

(a) (a)
5.0 5.0
Actual failure pressure/Predicted failure pressure

4.5

Actual failure pressure/Predicted failure pressure

4.5

A25 A25
4.0 4.0
B B
3.5 BX42 3.5 BX42
X42 X42
3.0 X46 3.0 X46
X52 X52
2.5 X55 2.5 X55
X56 X56
2.0 X60 2.0 X60
X65 X65
1.5 X80 1.5
X80
X100 X100
1.0 Line10 1.0
Line10
0.5 0.5

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Normalized defect depth d/t
Normalized defect depth d/t

(b)
5.0
(b)
5.0

Actual failure pressure/Predicted failure pressure

Actual failure pressure/Predicted failure pressure

4.5
4.5 A25
4.0 B
4.0 A25 BX42
B 3.5
X42
3.5 BX42 X46
3.0
X42 X52
3.0 X46 X55
2.5
X52 X56
2.5 X55 X60
2.0
X56
X65
2.0 X60
1.5 X80
X65
1.5 X80 X100
1.0 Line10
X100
1.0 Line10
0.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0 Normalized defect depth d/t
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Normalized defect depth d/t
Fig. 14. Comparison of actual and predicted failure pressure for integrated test
Fig. 12. Comparison of actual and predicted failure pressure for integrated test database using (a) LPC, and (b) PCORRC (Chauhan and Brister, 2009).
database using (a) ASME B31G, and (b) Mod B31G (Chauhan and Brister, 2009).

accepted the 2nd-G models of the LPC or PCORRC models for predict-
Figs. 12–14 show that (1) the RSTRENG model is the most accurate ing the remaining strength of actual corrosion defects, and continues to
one with the least scatter, and most data points fall into a relatively use the ASME B31G, or Mod B31G, or RSTRENG in the daily pipeline
small scatter; (2) the B31G model is the most conservative one with a integrity assessment. On this basis, a decision is then made for corroded
large scatter range; (3) the Mod B31G is less conservative than the B31G pipelines if they need a repair, replacement or rerun without an action.
with a reduced scatter range; and (4) the LPC and PCORRC models pre-
dict comparable results with a scatter range similar to the Mod B31G. 3.4.2. Experimental validation of all three generation models
Note that the two 2nd-G models were developed for defects with uni- Kim et al. (2002) conducted a set of full-scale experimental tests
form bottom, but not for actual river-bottom corrosion defects, and thus for machined metal-loss defects in a Korean X65 pipeline steel, and re-
predict conservative results. ported the burst data for machined defects with a fixed uniform depth
If a database was only for the machined uniform bottom defects, the of d/t=0.5 and six lengths of L=50, 100, 200, 300, 600, and 900 mm.
two 2nd-G models should predict the best results of burst pressure. Be- The pipe diameter is 762 mm (30 in.) and the wall thickness is 17.5 mm
cause of the large scatters shown in Fig. 14, the pipeline industry has not (0.69 in.). The actual YS and UTS of the X65 pipe are 495 MPa and 565
MPa. The burst pressures are predicted using all 3-generaton corrosion
models, and compared with test data, as shown in Fig. 15(a) and 15(b).
5.0 Fig. 15(a) compares the full-scale test (FST) data with predictions
Actual failure pressure/Predicted failure pressure

4.5 from four 1st-G models (B31G, Mod B31G, RSTRENG, and CSA Z662),
4.0 A25 two 2nd-G models (LPC and PCORRC), and six 3rd-G models (Mod-LPC,
B Mod-PCORRC, Zhu-2005, Ma-2013, Zhu-2015, and Ref-B31G). For the
3.5 BX42
X42 1st-G models, 1) the RSTRENG predictions are conservative and reason-
3.0 X46
X52
ably good for all defects, 2) the Mod B31G predictions are very accu-
2.5 X55 rate for short defects but non-conservative for long defects with length
X56 √
2.0 X60 L> (20Dt), 3) the B31G predictions are conservative, particularly for
X65 long defects, and 4) the CSA Z662 predictions are the lower bounds
1.5 X80
X100 and too conservative. For the 2nd-G models, the LPC and PCORRC pre-
1.0 Line10
dictions are comparable and match with the test data in a small con-
0.5 servative error. For the 3rd-G models, 1) the Zhu-2005 model is non-
0 conservative, 2) the Ref-B31G model is too conservative, and 3) all other
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Normalized defect depth d/t four models are comparable for all defects.
Fig. 15(b) compares the four comparable 3rd-G models and the two
Fig. 13. Comparison of actual and predicted failure pressure for integrated test 2nd-G models with the experimental data. Again, all those six mod-
database using RSTRENG (Chauhan and Brister, 2009). els predict comparable results and closely match the test data. How-

44
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

d/t
Pb / P0

L / (Dt)^0.5

Fig. 16. Comparison of Ma-2013 predictions with full-scale burst data of burst
pressure of defects in various pipeline grades (Ma et al., 2013).
d/t

sion using full-scale experimental datasets for machined flat-bottomed

defects and complex actual defects for a range of pipeline grades from
vintage to modern pipeline productions. Fig. 17 compares the predicted
Pb / P0

L / (Dt)^0.5
Fig. 15. Comparison of corrosion model predictions with experimental data,
(a) all models, and (b) selected 2nd-G models.
ever, further observation shows that 1) the PCORRC is slightly conser-
vative for all defects, 2) the LPC is conservative for short defects, 3) the
Ma-2013 is conservative for short defects and slightly non-conservative
for long defects, 4) the Mod-LPC is accurate for short defects and
slightly non-conservative for long defects, and 5) the Zhu-2015 and
Mod-PCORRC are nearly identical and the most accurate in compari-
son to the full-scale test data.
failure pressure using the Mod-PCORRC model in Eq. (28) with the ac-
tual failure pressure for the machined defects. As evident in this figure,
the predictions agree well with the actual data in a very high best-fitting
correlation factor of 0.991 between the model predictions and actual
data of burst pressure.
Similarly, Abdelghani et al. (2018) also evaluated the Mod-PCORRC
model using the calibration database of ASME B31G and Mod B31G for a
set of complex actual corrosion defects that have irregular river-bottom
shapes. The comparison demonstrated a good agreement between the
predicted failure pressure and the actual failure pressure with a best-
fitting correlation factor of 0.922, while the corresponding best-fitting
correlation factor is 0.617 and 0.698, respectively for B31G and Mod
B31G. Therefore, those full-scale burst data validate the accuracy of the

3.4.3. Experimental validation of Ma-2013 model

Ma et al. (2013) collected 79 full-scale experimental data for cor-
rosion defects in pipelines from public literature. The pipeline grades
ranging from X42 to X100 were categorized into three groups, i.e., low
strength steels (X42 to X56), mid-strength steels (X60 and X65) and
high-strength steels (X80 to X100). Those authors compared their model
(Ma-2013) predictions with the full-scale burst data, as shown in Fig. 16,
and concluded that the Ma-2013 model is very accurate for the high-
strength and mid-strength steels, and conservative but less accurate for
the low-strength steels.
Recall that the 3rd-G models of Mod-PCORRC, Mod-LPC and Zhu-
2015 are comparable to the Ma-2013 model, and thus it is reasonably
anticipated that those 3rd-G models are accurate to predict burst pres-
sures at least for corrosion defects in the high-strength and mid-strength
pipeline steels.

3.4.4. Experimental validation of Mod-PCORRC model Fig. 17. Comparison of the M-PCORRC predictions with the experimental data
More recently, Leis (2020) assessed the reference stress and the geo- for flat-bottom defects in the modern pipeline grades ranging X52 to X100
metric term of the PCORRC model and its modified or reformulated ver- (Leis, 2020).

45
X.-K. Zhu Journal of Pipeline Science and Engineering 1 (2021) 36–50

proposed Mod-PCORRC model for more accurately predicting the burst

pressure of metal-loss defects in pipelines. d/t

4. Recent progresses

4.1. PRCI research on corrosion assessment

Pb / P0
Corrosion assessment has been one of the major research topics for
managing pipeline integrity at PRCI for over a half century. Since the
early 2000s, the present author has participated in a series of PRCI re-
search projects on the corrosion assessment. This section briefs a review
on the recent PRCI projects from EC-2-5 to EC-2-10 and their major re-
sults.
Project EC-2-5 (Leis and Zhu, 2013) completed in 2011 assessed
the corrosion severity for high-strength pipeline steels. Project EC-2-
6 (Leis et al., 2017) completed in 2017 developed a more accurate L / (Dt)^0.5

reference stress for metal-loss assessment in reference to the Zhu-Leis

flow solution (Zhu and Leis, 2006a) for defect-free pipes. Project EC-2-7
Phase I (Leis et al., 2016) completed in 2016 numerically assessed the
PRCI corrosion model errors for metal-loss defects in pipelines. Project
EC-2-7 Phase II (Orth et al., December 2017) completed in 2017 con-
ducted a set of full-scale experimental tests on machined metal-loss de-
fects to assess corrosion model error for X70 pipeline steels. Recently,
Leis et al. (2017) and Leis and Zhu (2019) presented the experimental
and numerical results achieved in the Project EC-2-7.
Project EC-2-8 (Leis et al., 2019) completed in 2019 assessed the ap-
plicability of existing metal-loss criteria for low-hardening steels, where
a large amount of tensile stress-strain curves for a wide range of pipeline
steel grades were analyzed and an alternative equation was obtained to
estimate the strain hardening exponent n from the known YS and UTS.
Project EC-2-9 (Kiefner, 2019) completed in 2019 provided a technical
review of the plausible profiles (Psqr) corrosion assessment model that
Fig. 18. Comparison of burst pressure predicted from four corrosion models
with burst test data.
wall thickness of 17.5 mm. The defects have a fixed depth of d/t=0.5
√
and ten lengths of L/ (Dt) = 0.5, 1, 2, 3, 4, 5, 6, 8, 10 and 15. Fig. 19(a)
shows the variation of pressure normalized by the Zhu-Lies flow solution
with radial displacement at the defect center. Fig. 19(b) shows variation
of von Mises stress normalized by the UTS with pressure. Also included
are the results for defect-free pipes with a full and a half wall thickness.
The failure locus defined by the Zhu-Leis flow solution for defect-free
(a)
1.2

was recently developed at TC Energy (Zhang et al., 2018). Based on the Potential failure locus for defects

in-line inspection (ILI) data, RSTRENG uses one single profile to deter- 1.0
mine failure pressure. Psqr proposed multiple plausible profiles to use
RSTRENG to determine a minimum failure pressure. This approach re- 0.8
duces the conservatism and improves the accuracy of RSTRENG. Most
defect-free
recently, Project EC-2-10 Phase I (Zhu and McGaughy, 2020) completed
P/P2

L_bar=0.5
0.6
L_bar=1
in 2020 numerically quantified the defect width effect on the burst fail- L_bar=2
ure of metal-loss defects. 0.4
L_bar=3
L_bar=4
Based on the numerical results obtained in EC-2-7 and the reference Failure locus defined by L_bar=5

stress obtained in EC-2-6, an alternative corrosion assessment model was Zhu-Leis flow theory for defect-free pipes L_bar=6
0.2 L_bar=8
developed, where the bulging factor is a combined function of the defect L_bar=10
L_bar=15
length, depth and width. This is a significant improvement and different
0.0
from the original bulging factor in Eq. (Keifner and Atterbury, 1971) or 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(Maxey et al., 1972). Fig. 18 compares the EC-2-6 model with RSTRENG, Radial displacement (inch)

PCORRC and Mod-PCORRC, and those model failure predictions with

the full-scale burst data (Kim et al., 2002). This figure shows that the (b)
1.6
proposed alternative model is comparable to PCORRC or Mod-PCORRC Full-wall pipe
L_bar=0.5
and its prediction is less conservatism than RSTRENG. 1.4 L_bar=1
Potential failure
locus for defects
L_bar=2
L_bar=3
1.2
4.2. Constraint effect L_bar=4
L_bar=5
1.0 L_bar=6
As well known, the crack-tip constraint has a significant effect on L_bar=8
L_bar=10
fracture toughness of ductile steels (Zhu and Leis, 2006b). Recently, the 0.8
σ0/σ2

L_bar=15
Failure locus defined by
stress modified true fracture strain criterion (Oh et al., 2007; Oh et al., Thin-wall pipe Zhu-Leis flow theory for
defect-free pipes
0.6
2007; Alang et al., 2012) was developed to characterize the constraint
effect on material strength of a flawed structure. However, the fracture 0.4
strain criterion did not work well for a corrosion defect because the
defect failure is stress controlled. Accordingly, a constraint-based Mises 0.2

stress failure criterion was proposed to predict the plastic collapse of the
0
defect (Leis and Zhu, September 2005). Such a failure criterion develop- 0 0.2 0.4 0.6 0.8 1.0 1.2
ment needs many full-scale experimental tests of well-designed metal- P/P2
loss defects.
Recently, a set of FEA calculations was performed for ten machined Fig. 19. (a) Variation of pressure with radial displacement and (b) Variation of
defects in an X65 line pipe. The pipe has a diameter of 762 mm and a Mises effective stress with pressure.

46
X.-K. Zhu
Fig. 20. Comparison of BF obtained by FEA, B31G, and Mod B31G (Leis et al.,
2017).
pipes and an assumed potential failure locus for metal loss defects are
also included in the figures for illustration.
Experiments showed that local bulging occurs at a defect and in-
creases with defect length or depth. As shown in Fig. 19(a), the potential
failure locus is a curve, and reflects larger radial displacement and thus
higher failure pressure due to bulging. The same potential failure locus
is illustrated in Fig. 19(b), where the Mises stress normalized by the UTS
is not a constant as defined by the Zhu-Leis flow theory for defect-free
pipes, but a curve of increased strength as a function of defect length be-
cause of the constraint effect. Such constraint effect must be calibrated
using reliable full-scale burst test data.
4.3. Bulging factor and shape factor
The conservative corrosion assessment models ASME B31G, or Mod
B31G or RSTRENG remain in use in the oil and gas industry, and thus
it is necessary to assess these mode errors. A general format (Leis and
Zhu, 2013; Leis et al., 2017; Leis et al., 2016; Orth et al., December
2017; Leis and Zhu, 2019; Leis et al., 2019) of B31G criterion can be
expressed for the hoop stress at failure Sf , in the form of:
( )
⎡ 𝑑 ⎤ defect can be reused for nest defect tests so that the material costs can be
⎢ 1 − SF 𝑡 ⎥
𝑆f = 𝑆R ⎢ ( ) ⎥ (37)
𝑑
⎢ 1 − SF ∕BF ⎥
⎣ 𝑡 ⎦
where Sref is the reference stress, SF denotes shape factor, and BF denotes
bulging factor. In B31G, SF=2/3, and BF is given in Eq. (12). In Mod
B31G, SF=0.85, and BF is given in Eq. (14). To improve B31G, a general
function of three parameters Sref , SF and BF needs to be determined.
More recently, Leis et al. (Leis et al., 2017) discussed the effect of SF
and BF on the errors of B31G and Mod B31G models.
For a machined defect with uniform bottom, SF=1, and BF is an
undetermined parameter. Recently, the numerical study (Leis et al.,
2017) showed that the BF is a function of both defect length and
depth, as shown in Fig. 20. For comparison, the BF of B31G, and Mod
B31G are included in the figure. As evident in Fig. 20, the BF of Mod
B31G is close to that for d/t=0.7, and the BF of B31G is overesti-
mated for all depths. This is the root cause of conservatism of B31G
and Mod B31G. Recently, Sun et al. (2020) numerically investigated
the BF and obtained a similar BF as a function of defect length and
depth.
47
Journal of Pipeline Science and Engineering 1 (2021) 36–50
Fig. 21. Full-scale burst test of pipe with a metal loss defect.
4.4. Defect width effect
In all corrosion model discussed above, the defect width is not con-
sidered because its effect is considered insignificant. Although the width
is not a controlling parameter, but it does influence the failure of wide
defects. As shown in the PRCI reports (Leis et al., 2016; Zhu and Mc-
Gaughy, 2020), the PCORRC may underestimate the burst pressure for
narrow defects (or width angle is less than 5o ), and overestimate burst
pressure for wider defects (or width angle is larger than 30o ). The
width effect has been coupled in the BF of the EC-2-6 model. Recently,
Shuai et al. (2017) and Zhang and Zhou (2020) also studied the width
effect on defect failure.
5. Major technical challenges
5.1. Experimental burst tests
Full-scale experimental tests play a dispensable role in measurement
of reliable burst data. Most full-scale burst tests are conducted using
hydrostatic tests, where a pressure vessel is typically made of a long
piece of pipe with two tougher caps welded to both ends, and then water
is slowly pumped into the vessel until failure. Those full-scale tests are
usually expensive, particularly for large diameter pipes due to material
costs. To reduce material costs, investigators may use a short cylindrical
vessel with a length about two times diameter or even shorter. As a
result, test data obtained from a short pressure vessel may be inaccurate
and incapable to represent a long pipe, because its burst pressure is likely
increased due to the end-cap effect. This is particularly true for a shorter
vessel with a long defect. Thus, a short vessel should not be used to
measure burst pressure in a full-scale test. In general, a pressure vessel
should have at least five times diameter to avoid the possible end-cap
effect.
For a long pressure vessel with a corrosion defect, plastic deforma-
tion limits to local defect area, and other pipe body remote to the defect
remains in elastic conditions. As a result, the pipe body remote to the
saved. As shown in Fig. 21, a corroded pipe is designed to have a length
of two to four times diameter, and then welded at each end with an
extended pipe of the same steel with a length about three to five times
diameter. This extended pipe is then welded with two end caps to form
a long pressure vessel specimen for a hydraulic test. The total length
of such a pressure vessel specimen will be eight to ten times diameter,
and thus the end-cap effect will be minimized. Thus, the three-segment
design of a long pressure vessel is a way to compromise the full-scale
test challenge.
In order to save experimental test costs, a ring expansion test ma-
chine shown in Fig. 22 was discovered in the middle 1990s in Europe
to measure burst strength of line pipes with or without corrosion de-
fects, where only a ring or a short segment cut from a full-scale pipe
is tested under expansion. The ring test data were often mixed to use
with the full-scale hydrostatic test data, such as the database collected
in Reference (Chauhan and Brister, 2009) for evaluating existing corro-
sion assessment. However, a ring expansion tests is not equivalent to a
full-scale hydrostatic test, because the ring has no bulging during the
expansion test and is a uniaxial tensile stress state in the hoop direction.
X.-K. Zhu
Fig. 22. Ring expansion test machine.
In contrast, a full-scale pipe experiences a biaxial tensile stress state in
both hoop and axial directions during a hydrostatic test. Thus, a ring
expansion test may be comparable to the uniaxial tensile test, but not
to the full-scale hydrostatic test. Because of this, the ring expansion test
should not be used to replace the full-scale pipe hydraulic test in prac-
tice.
5.2. Numerical simulation and material failure criteria
The FEA numerical simulation of pipeline failure is based on the
continuum theory. However, the continuum theory has not a material
failure criterion, and a prescribed material failure criterion is needed
in the FEA simulation to determine material failure local to a defect.
Consequently, different failure criteria were used in the FEA calculations
of failure pressure, which resulted in different predictions. Thus, the
use of an appropriate material failure criterion in the FEA is a great
challenge.
Because the RIKS model (Riks, 1979) determines a global failure,
different material failure criteria have been used in the FEA simulations
for determining local failure at a defect. Those material failure criteria
were based on the von Mises or Tresca effective stress. Three of them
were often used:
(1) von Mises effective stress equals 1.0 true UTS: 𝜎M FEA = 1.0𝜎 true .
uts
This is the Mises stress failure criterion that was used in
DNV JIP work (DNV, 1997) and by Fu and Kirkwood (1995),
Batte et al. (1997), Fu and Batte (1999) in the FEA simulations
for X52, X60 and X65 during the LPC development. This fail-
ure criterion was then extensively used by many investigators
(Zhou and Huang, 2012; Benjamin and Andrade, 2003; Shuai, 2010;
Karstensen and Smith, 2001; Noronha et al., 2002; Ma et al., 2013;
Shuai et al., 2017; Zhang and Zhou, 2020; DNV, 1997; Jin et al.,
2020; Bao and Zhou, 2020; Zhang, 2016) for various pipeline grades
from X52 to X80.
(2) von Mises effective stress equals 0.9 true UTS: 𝜎MFEA = 0.9𝜎 true . This is
uts
the Mises stress failure criterion that was used by Choi et al. (2003),
Kim et al. (2002), Yeom et al. (2015), Yeom et al. (2016), and
other researchers (Chiodo and Ruggieri, 2009; Mokhtari and Melch-
ers, 2018) for X60 to X70 pipeline steels. This Mises stress failure
criterion may predict better results. Note that the 0.9 true UTS is
approximately equivalent to 1.0 UTS (engineering value) for X60 or
X65 steels.
(3) Tresca stress equals 1.0 true fracture stress: 𝜎TFEA = 1.0𝜎ftrue .
This Tresca stress failure criterion was used by Leis and
Stephens (1997) and Stephens et al. (1999) for X52 to X70 pipeline
steels during development of PCORRC model.
48
Journal of Pipeline Science and Engineering 1 (2021) 36–50
The DNV JIP Report 96-3392 (DNV, 1997) presented four sets of FEA
results of remaining strength for X60 corroded pipes with four depths
√
of d/t=0.15, 0.3, 0.5, and 0.7 and ten lengths up to L/ (Dt)=10, where
the Mises stress failure criterion of 𝜎M
FEA = 1.0𝜎 true was used in all FEA
uts
calculations. Their results show that the remaining strength for short
√
and shallow defects (L/ (Dt) < 1.5 and d/t = 0.15 and 0.3) is larger
than the Tresca strength solution for defect-free pipes. Obviously, the
failure criterion overestimates the burst pressure for shallow and short
defects. As a result, when DNV (Det Norske Veritas, 2015) adopted the
LPC model as an assessment tool, a safety factor of 0.9 was enforced for
single metal-loss defects.
5.3. Application to assess real corrosion defects
Assessing real metal loss defects faces many challenges. First is to
detect defects in pipelines and to characterize defect sizes. This needs
advanced in-line inspection (ILI) technology and tools, such as magnetic
flux leakage (MFL), ultrasonic technology (UT), geometric tool and oth-
ers. Once a corrosion defect is detected and characterized, its assessment
is a great challenge. Any real corrosion defect is not an isolated single
uniform defect, but a cluster of corrosion pits with a river bottom profile
and may be not axially oriented. In addition, separate corrosion clusters
may have an interaction to each other, and assessing corrosion defect
interaction is another challenge. For some very old pipelines, material
properties may not be documented, and this can be an additional chal-
lenge for assessing corrosion defects in ageing pipelines.
6. Conclusions
This paper performed a comparative study of remaining strength as-
sessment models for corroded pipelines. A brief review was first given
to the burst prediction models for defect-free pipes, including strength
solutions and flow solutions of burst pressure and experimental valida-
tions. Followed was a review of corrosion assessment models that were
categorized into three generations in terms of the reference stress used
in each model. The 1st generation was defined based on the flow stress,
the 2nd one was defined based on the UTS, and the 3rd one was defined
based on the UTS and strain hardening rate.
After that, twenty representative corrosion models were critically
evaluated in comparison to full-scale burst data, with a focus on validat-
ing two newly proposed 3rd-G models of Mod-PCORRC and Zhu-2015.
The comparison showed that these new models are the best to predict
more accurate burst pressure for corroded pipelines. Then, recent pro-
gresses were presented, including the PRCI corrosion research projects,
constraint effect, bulging factor, and defect width effect.
Finally, major technical challenges facing the corrosion model im-
provement were discussed with regards to full-scale experiments, nu-
merical modeling, material failure criteria, and real corrosion defects.
It is anticipated that a large experimental database is needed to further
validate the 3rd-G models and the present work will help researchers
for better understanding corrosion models and facilitate further stud-
ies to develop a more accurate corrosion model for practical use in the
pipeline integrity management.
Declaration of Competing Interest
The author declares that this no conflict of interest in this work.
References
ABAQUS Standard, 2018. Analysis User’s Manual. Dassault Systémes Simulia Corporation,
Providence, RI.
Abdelghani, M., Tewfik, G., Djahida, D., Ahmed, S.S., 2018. Prediction of the rupture
pressure of transmission pipelines with corrosion defects. J. Press Vessel Technol.
140, 041701.
Alamilla, J.L., Sosa, E., Sanchez-Magana, C.A., Rndrade-Valencia, R., Contreras, A., 2013.
Failure analysis and mechanical performance of an oil pipeline. Mater. Des. 50,
766–773.
X.-K. Zhu
Alang, N.A., Razak, N.A., Sulaiman, A., 2012. Determination of burst pressure of API steel
pipes using stress modified critical strain model. Mater. Sci. Eng. 36, 012025.
Amaya-Gomez, R., Sánchez-Silva, M., Bastidas-Arteaga, E., Schoefs, F., Muñoza, F., 2019.
Reliability assessments of corroded pipelines based on internal pressure – a review.
Eng. Fail. Anal. 98, 190–214.
American Lifeline Alliance, 2001. Guidelines for the Design of Buried Steel Pipe.
API, 2018. Lin Pipe, API Specification 5L, the 46th Edition American Petroleum Institute.
API/ASME, 2016. Fitness for Service, API 579-1/ASME FFS-1.
ASME, 1991. Manual for Determining the Remaining Strength of Corroded Pipelines,
ASME B31G-1991. American Society of Mechanical Engineers, New York, USA.
ASME, 2009. Manual for Determining the Remaining Strength of Corroded Pipelines,
ASME B31G-2009. American Society of Mechanical Engineers, New York, USA.
Bao, J., Zhou, W., 2020. Influence of the corrosion anomaly class on predictive accuracy
of burst capacity models for corroded pipelines. Int. J. Geosynth. Ground Eng. 6, 45.
Batte, A.D., Fu, B., Kirkwood, M.G., Vu, D., 1997. Advanced method for integrity assess-
ment of corroded pipelines. Pipes Pipelines Int. 42 (1), 5–11.
Benjamin, A.C., Andrade, E.Q., 2003. Modified method for the assessment of the remaining
strength of corroded pipelines. In: Proceedings of 2003 Rio Pipeline Conference. Rio,
Brazil October 22-24.
Besel, M., Zimmerman, S., Kalwa, C., Koeppe, R., Liessem, A. , 2010. Corrosion assess-
ment method validation for high-grade line pipe. Proceeding of the 8th International
Pipeline Conference. Calgary, Canada Sept 27 – Oct 1.
Bhardwaj, U., Teixeira, A.P., Soares, C.G., 2020. Uncertainty quanti-fication of burst pres-
sure models of corroded pipelines. Int. J. Press Vessels Pip. 188, 104208.
Bony, M., Alamilla, J.L., Vai, R., Flores, E., 2010. Failure pressure in corroded pipelines
based on equivalent solutions for undamaged pipe. J. Press Vessel Technol. 132,
051001.
British Standards Institution, 2015. Guide to methods for accessing the acceptability of
flaws in metallic structures, BS 7910:2013 + A1:2015. BSI Standards Limited, Lon-
don, UK.
Chauhan, V., Brister, J., 2009. A Review of Methods for Assessing the Remaining Strength
of Corroded Pipelines US DOT Final Report 153A.
Chen, Z., Zhu, W., Di, Q., Wang, W., 2015. Prediction of burst pressure of pipes with
geometric eccentricity. J. Press Vessel Technol. 137 (6), 061201.
Chen, Y., Zhang, H., Zhang, J., Li, X., Zhou, J., 2015. Failure analysis of high strength
pipeline with single and multiple corrosions. Mater. Des. 67, 552–557.
Chen, Z., Yan, S., Jin, Z., 2020. Dynamic burst pressure of cylindrical explosion contain-
ment vessels. J. Press Vessel Technol. 141, 041203.
Chiodo, M.S.G., Ruggieri, C., 2009. Failure assessments of corroded pipelines with axial
defects using stress-based criteria: numerical studies and verification analyses. Int. J.
Press Vessels Pip. 86, 164–176.
Choi, J.B., Goo, B.K., Kim, J.C., Kim, Y.J., Kim, W.S., 2003. Development of limit load
solutions for corroded gas pipelines. Int. J. Press Vessels Pip. 80, 121–128.
Christopher, T., Rama Sarma, B.S.V., Potti, P.K, et al., 2002. A comparative study on
failure pressure estimation of unflawed cylindrical vessels. Int. J. Press Vessels Pip.
79, 53–66.
Cosham, A., Andrews, R., 2019. The assessment of locally thinned areas: Background to the
guidance given in Annex G of BS 7910:2013. Int. J. Press Vessels Pip. 169, 177–187.
Cosham, A., Hopkins, P., Macdonald, K.A., 2007. Best practice for the assessment of defects
in pipelines. Eng. Fail Anal. 14, 1245–1265.
Coulson, K.W., Worthingham, R.G., 1990. Standard damage assessment approach is overly
conservative. Oil Gas J. 88 (15), 54.
CSA, 2019. Oil and Gas Pipeline Systems, CSA Z662-19. Canadian Standards Association,
Mississauga, Canada.
Dai, L., Wang, D., Wang, T., Feng, Q.S., Yang, X.Q., 2017. Analysis and comparison of
long-distance pipeline failures. J. Pet. Eng. 2017 (2017), 3174636.
Det Norske Veritas, 2015. Recommended Practice DNV-RP-F101 – Corroded
Pipelines.
DNV, 1997. Reliability of Corroded Pipes Finite Element Analysis, DNV JIP Report 96-
3392, Rev 01. Det Norske Veritas.
Folias, E.S., 1964. The Stresses in a Cylindrical Shell Containing an Axial Crack. Aerospace
Research Laboratory Repot ARL 64-174.
Folias, E.S., 1965. An axial crack in a pressured cylindrical shell. Int. J. Fract. Mech. 1,
104–113.
Fu, B., Batte, A.D., 1999. New methods for assessing the remaining strength of corroded
pipelines. EPRG/PRCI 12th Biennial Joint Technical Meeting on Pipeline Research
Paper 28.
Fu, B., Kirkwood, M.G., 1995. Determination of failure pressure of corroded linepipes us-
ing nonlinear finite element method. In: Proceedings of the 2nd International Pipeline
Technology Conference, II, pp. 1–9.
Gao, J., Yang, P., Li, X., Zhou, J., Liu, J.K., 2019. Analytical prediction of failure pressure
for pipeline with long corrosion defect. Ocean Eng. 191, 106497.
Jin, Z., Qiu, C., Chen, Z.F., Yan, S.T., Shen, X.L., 2020. Integrity assessment of pipelines
containing an isolated corrosion pit. Eng. Fail. Anal. 113, 104539.
Karstensen, A.D., Smith, A.T., 2001. Corrosion damage assessment and burst test valida-
tion of 8in X52 linepipe. In: Proceedings of ASME Pressure Vessel and Piping Confer-
ence. Georgia, Atlanta July 22-26.
Keifner, J.F., Atterbury, T.J., 1971. Investigation of the Behavior of Corroded Line Pipe
Project 216 Interim Report.
Kiefner, J.F., Duffy, A.R., 1971. Summary of Research to Determine the Strength of Cor-
roded Areas in Line Pipe Battelle Interim Report to Texas Eastern Transmission Cor-
poration.
Kiefner, J.F., Vieth, P.H., 1989. A Modified Criterion for Evaluating the Remaining
Strength of Corroded Pipe Final report on Project PR 3-805 to the Pipeline Research
Committee of the American Gas Association.
Kiefner, J.F., Maxey, W.A., Eiber, R.J., Duffy, A.R., 1973. Failure stress levels of flaws in
49
Journal of Pipeline Science and Engineering 1 (2021) 36–50
pressurized cylinders. In: Progress in Flaw Growth and Fracture Toughness Testing,
536. ASTM STP, pp. 461–481.
Kiefner, J.F., Vieth, P.H., Roytman, I., 1996. Continuing validation of RSTRENG. Pipeline
Research Supervisory Committee, A.G.A Catalogue No. L51689.
Kiefner, J., 2019. Peer Review of the Plausible Profiles (Psqr) Corrosion Assessment Model
PRCI Project Report for EC-2-9, PR-218-183607.
Kim, W.S., Kim, Y.P., Kho, Y.T., Choi, J.B., 2002. Full scale burst test and finite element
analysis on corroded gas pipeline. In: Proceedings of the fourth International Pipeline
Conference. Calgary, Alberta, Canada Sept 30 – Oct 3.
Law, M., Bowie, G., 2007. Prediction of failure strain and burst pressure in high yield–
to-tensile strength ratio line pipes. Int. J. Press Vessels Pip. 84, 487–492.
Leis, B.N., Stephens, D.R., 1997. An alternative approach to assess the integrity of corroded
line pipes – part I: current status and part II: alternative criterion. In: Proceedings
of the Seventh International Offshore and Polar Engineering Conference. Honolulu,
Hawaii, USA May 25-30.
Leis, B.N., Zhu, X.K., 2013. Assessing Corrosion Severity for High-Strength Steels PRCI
Project Report for EC-2-5, Catalog No. PR003-103603-R01.
Leis, B.N., Zhu, X.K., et al., 2019. Step improvement in metal-loss assessment criteria for
pipelines. APGA-EPRG-PRCI 22nd Joint Technical Meeting on Pipeline Research 29
April - 3 May.
Leis, B.N., Zhu, X.K., September 2005. Corrosion assessment criteria: Rationalizing their
use for vintage vs modern pipelines DOT Research and Special Projects Agency,
DTRS56-03-T-0014.
Leis, B.N., Zhu, X.K., McGaughy, T., 2016. Assessment of Corrosion Model Error for Met-
al-Loss Defects in Pipelines Phase I of PRCI Project Report for EC-2-7, Catalog No.
PR185-143600.
Leis, B.N., Zhu, X.K., Orth, F., Aguiar, D., Perry, L., 2017. Minimize model uncertainty in
current corrosion assessment criteria. PRCI-APGA-EPRG 21th Joint Technical Meeting
on Pipeline Research May 1-5.
Leis, B.N., Zhu, X.K., McGaughy, T, 2017. Reference Stress for Metal-Loss Assessment of
Pipeline PRCI Project Report for EC-2-6, Catalog No. PR185-173600.
Leis, B.N., Zhu, X.K., McGaughy, T., 2019. Applicability of Existing Metal-Loss Criteria for
Low-Hardening Steels Catalog No. PRCI Project Report for EC-2-8, PR185-173611.
Leis, B.N., 2020. Continuing development of metal-loss severity criteria – including width
effects. In: Proceedings of the ASME International Pipeline Conference. Calgary, AB,
Canada Sept 28 - Oct 2.
Ma, B., Shuai, J., Wang, J.Q., Han, K.J., 2011. Analysis on the latest assessment criteria
of ASME B31G-2009 for the remaining strength of corroded pipelines. J. Fail Anal.
Prevent. 11, 666–667.
Ma, B., Shuai, J., Liu, D., Xu, K., 2013. Assessment on failure pressure of high strength
pipeline with corrosion defects. Eng. Fail. Anal. 32, 209–219.
Maxey, W.A., Kiefner, J.F., Eiber, R.J., Duffy, A.R., 1972. Ductile fracture initiation, prop-
agation and arrest in cylindrical vessels. Fract. Toughness 514, 347–362 Part II.
Miller, A.G., 1988. Review of limit loads of structures containing defects. Int. J. Press
Vessels Pip. 32, 197–327.
Mokhtari, M., Melchers, R.E., 2018. A new approach to assess the remaining strength of
corroded steel pipes. Eng. Fail. Anal. 93, 144–156.
Noronha, D.B., Benjamin, A.C., Andrade, E.Q., 2002. Finite element models for the pre-
diction of the failure pressure of pipelines with long corrosion defects. In: Proceedings
of the 4th International Pipeline Conference. Calgary, Canada Sept 29 – Oct 3.
Nyman, D.J., Lee, E.M., Audibert, J.M.E., 2008. Mitigating geohazards for in-
ternational pipeline projects: challenges and lessons learned. In: Proceed-
ings of the 7th International Pipeline Conferences. Calgary, Canada Sept 29-
Oct 3.
Oh, C.K., Kim, Y.J., Baek, J.H., Kim, W.S., 2007. Development of stress-modified fracture
strain for ductile failure of API X65 steel. Int. J. Fract. 143, 119–133.
Oh, C.K., Kim, Y.J., Baek, J.H., et al., 2007. Ductile failure analysis of API X65 pipes
with notch-type defects using a local fracture criterion. Int. J. Press Vessels Pip. 84,
512–525.
Orth, F., Zhu, X.K., McGaughy, T., Leis, B.N., December 2017. Assessment of corrosion
model error for metal-loss defects in pipelines: Phase II - Full-scale experiments PRCI
Project Report for EC-2-7, Catalog No. PR185-163609.
Ossai, C.I., Boswell, B., Davies, I.J., 2015. Pipeline failure in corrosive environments – a
conceptual analysis of trends and effects. Eng. Fail. Anal. 53, 36–58.
Peng, L.-C., 1978. Stress analysis methods for underground pipelines. Pipe Line Ind. 47
(5), 67–71.
Porter, M., Ferris, G., Leir, M., Leach, M., Mario, H., 2016. Updated estimates of frequen-
cies of pipeline failures caused by geohazards. In: Proceedings of the 11th Interna-
tional Pipeline Conferences. Calgary, Canada. Calgary, AB, Canada Sept 26-30.
Riks, E., 1979. An incremental approach to the solution of snapping and buckling prob-
lems. Int. J. Solids Struct. 15 (7), 529–551.
Ritchie, D., Last, S., 1995. Shell 92 – Burst criteria of corroded pipelines – defect ac-
ceptance criteria. EPRG-PRCI 10th Biannual Joint Technical Meeting on Pipeline Re-
search.
Seghier, M.E.A.B., Keshtegar, B., Elahmoune, B., 2018. Reliability analysis of low, mid
and high-grade strength corroded pipes based on plastic flow theory using adaptive
nonlinear conjugate map. Eng. Fail. Anal. 90, 245–261.
Shuai, Y., Shuai, J., Xu, K., 2017. Probabilistic analysis of pipelines based on a new failure
pressure model. Eng. Fail. Anal. 81, 216–233.
Shuai, J., 2010. Evaluating the failure pressure of corroded pipeline. In: Proceedings of
the ASME Pressure Vessels and Piping Conference. Bellevue, Washington, USA July
18-22.
Souza, R.D., Benjamin, A., Vieira, R., Fieire, J., 2007. Rupture tests of pipeline segments
containing long real corrosion defects. Exp. Technol. 31, 46–51.
De Souza, R.D., Benjamin, A.C., Freire, J.L.F., Vieira, R.D., Diniz, J.L.C., 2004. Burst tests
X.-K. Zhu
on pipeline containing long real corrosion defects. In: Proceedings of International
Pipeline Conference. Calgary, Alberta. ASME, Calgary, AB, Canada Oct 4-8.
Stephens, D.R., Leis, B.N., 2000. Development of an alternative criterion for residual
strength of corrosion defects in moderate-to-high toughness pipe. In: Proceedings of
International Pipeline Conference. Calgary, Canada Oct 1-5.
Stephens, D.R., Leis, B.N., Kurre, M.D., Rudland, D.L., 1999. Development of an Alterna-
tive Failure criterion for Residual Strength of Corrosion Defects in Moderate-to-High
Toughness Pipe, PRCI Project Report PR3-9509.
Sun, M., Li, X., Liu, J., 2020. Determination of Folias factor for failure pressure of corroded
pipeline. J. Press Vessel Technol. 142, 031802.
Tresca, H., 1864. Memoire sur Iecoulement des corps solides soumis a de fortes pressions.
C R Acad. Sci. Paris 59, 754–758.
US DOT PHMSA, 2018. Pipeline Incident 20 Year Trends www.phmsa.dot.gov/
data-and-statistics/pipeline .
von Mises, R., 1913. Mechanick der festen korper im plastisch-deformablen zustand. NG-
WG-Math.-Phys. Klass 1, 582–592.
Yeom, K.J., Lee, Y.K., OH, K.H., Kim, W.S., 2015. Integrity assessment of a corroded API
X70 pipe with a single defect by burst pressure analysis. Eng. Fail. Anal. 57, 553–
561.
Yeom, K.J., Kim, W.S., Oh, K.H., 2016. Integrity assessment of API X70 pipe with corroded
girth and seam welds via numerical simulation and burst test experiments. Eng. Fail.
Anal. 70, 375–386.
Zhang, S., Zhou, W., 2020. Assessment of effects of idealized defect shape and width on
the burst capacity of corroded pipelines. Thin-Wall Struct. 154, 106806.
Zhang, S.W., Yan, J., Kariyawasam, S., Huang, T., Al-Amin, M., 2018. A more accurate
and precise method for large metal loss corrosion assessment. In: Proceedings of the
ASME International Pipeline Conference. Calgary, Canada. Calgary, Canada Sept 24 -
28.
Zhang, Y.M., 2016. Failure assessment on offshore girth welded pipelines due to corrosion
defects. Fatigue Fract. Eng. Mater. Struct. 39, 453–466.
Zhou, W., Huang, G.X., 2012. Model error assessments of burst capacity models for cor-
roded pipelines. Int. J. Press Vessels Pip. 99-100, 1–8.
Zhou, W., Huang, T., 2012. Model error assessment of burst capacity models for defec-
t-free pipes. In: Proceedings of International Pipeline Conference. Calgary, Canada
Sept 25-28.
Zhu, X.K., Leis, B.N., 2004. Accurate prediction of burst pressure for line pipes. J. Pipel.
Integr. 4, 195–206.
50
Journal of Pipeline Science and Engineering 1 (2021) 36–50
Zhu, X.K., Leis, B.N., 2005. Influence of yield-to-tensile strength ratio on failure assessment
of corroded pipelines. J. Press Vessel Technol. 127, 436–442.
Zhu, X.K., Leis, B.N. , 2006a. Average shear stress yield criterion and its application to
plastic collapse analysis of pipelines. Int. J. Press Vessels Pip. 83, 663–671.
Zhu, X.K., Leis, B.N. , 2006b. Bending modified J-Q theory and crack-tip constraint quan-
tification. Int. J. Fract. 141, 115–134.
Zhu, X.K., Leis, B.N., 2007. Theoretical and numerical predictions of burst pressure of
pipelines. J. Press Vessel Technol. 129, 644–652.
Zhu, X.K., Leis, B.N. , 2012a. Evaluation of burst pressure prediction models for line pipes.
Int. J. Press Vessels Pip. 89, 85–97.
Zhu, X.K., Leis, B.N. , 2012b. Assessment criteria and burst pressure prediction for
pipelines with long blunt defect. In: Proceedings of the 9th International Pipeline
Conference. Calgary, Canada Sept 24-28.
Zhu, X.K., McGaughy, T., 2020. Development of a Comprehensive Metal-Loss Assessment
Criterion – Phase I on Quantify Defect Width Effect on Failure of Metal-Loss Defects
PRCI Project Report for EC-2-10, Catalog No. PR185-193601-R01.
Zhu, X.K. , 2015a. A new material failure criterion for numerical simulation of burst pres-
sure of corrosion defects in pipelines. In: Proceedings of the ASME pressure vessels
and Piping Conference. Boston, Massachusetts, USA July 19-23.
Zhu, X.K. , 2015b. Corrosion assessment methods for pipelines with long blunt defects. J.
Pipeline Eng. 14, 111–120.
Zhu, X.K., 2016. Strength criteria versus plastic flow criteria used in pressure vessel design
and analysis. J. Press Vessel Technol. 138, 041402.
Zhu, X.K., 2018. Assessment methods and technical challenges of remaining strength for
corrosion defects in pipelines. In: Proceedings of the ASME Pressure Vessels and Piping
Conference. Prague, Czech July 15-20.
Zhu, X.K., 2019. Burst failure models and their predictions of buried pipelines. In: Pro-
ceedings of the Conference on Asset Integrity Management – Pipeline Integrity Man-
agement under Geohazard Conditions. Houston, TX, USA March 25-28.
Zimmermann, S., Marewski, U., Hohler, S., 2007a. Burst pressure of flawless pipes. 3R
Inter, Special Edition 46, 28–33.
Zimmermann, S., Hohler, S., Marewski, U., 2007b. Modeling of ultimate limit states: Burst
pressure and circumferential elongation of flawless pipe. APRA-EPRG-PRCI 16th Bi-
ennial Pipeline Research Joint Technical Meeting.