Journal of Constructional Steel Research 113 (2015) 234–246

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Ratcheting behaviour of corroded steel tubes under uniaxial cycling:

An experimental investigation
M. Zeinoddini a, M. Peykanu a, M. Varshosaz b, M. Ezzati a,⁎, S.J. Zakavi c
a
Department of Civil and Environmental Engineering, K.N.Toosi University of Technology, Iran
b
Department of Geodesy and Geomantic Engineering, K.N.Toosi University of Technology, Iran
c
Department of Mechanical Engineering, University of Mohaghegh Ardabili, Ardabil, Iran

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

Article history: The influence of corrosion defects on the ratcheting behaviour of steel tubes under uniaxial cyclic loadings has
Received 16 September 2014 been experimentally investigated. Different material grades, loading protocols and defect geometries have
Received in revised form 29 May 2015 been considered. Surface defects have been found to be more detrimental to the cyclic response of the tubes,
Accepted 15 June 2015
as compared to their monotonic uniaxial response. The axisymmetric wrinkling mode of failure in intact tubes
Available online 30 June 2015
changes to non-axisymmetric modes in defective tubes. In a defective tube the strains in the damaged and perfect
Keywords:
parts of the tube ratchet at significantly different rates.
Strain ratcheting © 2015 Elsevier Ltd. All rights reserved.
Uniaxial cyclic stressing
Corroded steel pipe
Wrinkling
Defect

1. Introduction shutdown. A question arises as to whether wrinkles formed as a result

Steel tubulars are used in different industrial applications such as
offshore pipelines, offshore platforms and heat exchangers. Axial com-
pression in the tubular members may lead to appreciable plastic strains
in the component. Subsequent axial cycling could cause a gradual
growth of wrinkling in the member which may eventually result in
propagation of plastic strains and failure of the tube. The ratcheting or
cyclic creep effects should be addressed in the safety assessment and fa-
tigue life estimation of these components [1].
Strain ratcheting, for example, can be caused by repetitive start-up/
shutdown and temperature cycles in unburied offshore pipelines. A sub-
marine pipeline is axially restrained, for instance by the frictions from
the pipe bed. As a result, a temperature change, caused by the passage
of hot hydrocarbons coupled with high internal pressure, can plastically
deform the pipe/tube. In some cases the compression is high enough to
initiate axial wrinkling. Imperfections due to small misalignments at
girth welds, in heat-affected regions around the welds, hard spots at
connections with other equipment, etc. can all accelerate the onset of
wrinkling. During its lifetime of around 20–30 years, an offshore pipe-
line usually experiences several hundred cycles of start-up and
⁎ Corresponding author at: K.N. Toosi University of Technology, Vali Asr Street,
Mirdamad Cross, Tehran, Iran.
E-mail addresses: Zeinoddini@kntu.ac.ir (M. Zeinoddini), Mpeykanu@gmail.com
(M. Peykanu), Varshosazm@kntu.ac.ir (M. Varshosaz), Mezzati@sina.kntu.ac.ir (M. Ezzati),
Zakavi@uma.ac.ir (S.J. Zakavi).
of such stress rises can grow (ratchet) from the shutdown and start-
up cycles and if so what are the consequences [2].
Jiao and Kyriakides [2] investigated whether an intact steel tube that
develops small amplitude wrinkles can subsequently lead to total col-
lapse by persistent uniaxial cyclic loads. They then reported experimen-
tal and numerical results involving axial cycling of tubes under internal
pressure. They showed that axial cyclic loading causes a gradual short-
ening of the test section (axial strain ratcheting) while the internal pres-
sure induces ratcheting of the hoop strain [3,4].
The ratcheting response of materials is significantly influenced by
cyclic loading parameters such as maximum stress, stress amplitude,
mean stress, initial axial strain, stress rate and material behaviour such
as cyclic hardening/softening. Effects of stress amplitude, mean stress
and initial pre-strain on the ratcheting response of intact tubes under
uniaxial cyclic loading were investigated by a number of researchers
[1,2,5–9]. It was concluded that the ratcheting strain amplitude and
ratcheting strain rate increase with increase in the initial pre-strain,
stress amplitude or mean stress, correspondingly. Cyclic hardening or
softening has been reported to significantly influence ratcheting [10].
Sea water from outside and fluid being conveyed form inside may
gradually corrode an offshore pipeline. This type of material loss could
appear with different extents and depths on both sides of the pipe
wall. The influence of the corrosion defects on the residual strength
and local buckling of steel pipes under monotonic loading has received
considerable previous attentions in the literature. Netto [11] studied the
effect of narrow and long corrosion defects on the monotonic collapse

http://dx.doi.org/10.1016/j.jcsr.2015.06.007
0143-974X/© 2015 Elsevier Ltd. All rights reserved.
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 235

pressure of offshore pipelines. Results from small-scale experiments

and non-linear finite element numerical analyses were reported in
these studies. Sakakibara et al. [12] and Netto et al. [13] studied the in-
fluence of internal corrosion or erosion defects on the collapse of pipe-
lines under external pressure. Xue [14] presented results from a non-
linear finite element analysis for the steady-state buckle propagation
phenomenon in subsea corroded pipelines subjected to external hydro-
static pressure. Xue and Fatt [15] derived closed form analytical solu-
tions for this problem.
As it was previously mentioned, the strain ratcheting problem and
the corrosion effects on the residual strength of steel tubular have
been separately addressed by different researchers. However, the sub-
ject of ratcheting, wrinkling and collapse of corroded steel tubes under
cyclic uniaxial loadings has not received enough attentions in the liter-
ature. The present paper deals with the influence of rectangular corro-
sion defects on the strain ratcheting behaviour of steel pipelines
subjected to uniaxial cyclic loadings. A small scale experimental model-
ling approach has been chosen. Both monotonic and cyclic uniaxial
loading regimes on specimens with and without the corrosion defect
have been examined.

2. Experimental set-up and procedures

2.1. Specimens' origin and type

Dimensions and the geometry of tubular specimens considered in

the current study were close to those used in cyclic tests carried out
on steel tubes by some researchers such as by Jiao and Kyriakides [1,2]
and monotonic tests conducted by Paquette and Kyriakides [16]. It
should be noted that these researchers employed intact tubes in their
experiments.
Test samples in the current study were cut from the ‘as delivered’
lengths of two seamless carbon steel pipes class X42/B and X56 with
nominal external diameter and thickness of 59.4 mm and 5 mm, respec-
tively. Mechanical properties of the steel material were determined
from standard static tensile coupon tests. For this, four coupon test spec-
imens were prepared using a wire cut machine. The average values ob-
tained from the material tensile tests are presented in Table 1.
The specimens were first cut in 300 mm lengths from the main Fig. 1. Geometry of the tested tubular specimens (all dimensions in mm).
pipes. This was slightly longer than the final desired length. They were
then machined externally and internally in order to eliminate possible
existing distortions. Maximum out of straightness in the specimens adjusted to minimise any stress risers at the discontinuity and maximise
were kept bellow 0.030 mm, which could be regarded as the pieces' the axial stress consistency along the test section in the middle of the
manufacturing error. Final length of the specimens became 280 mm specimen. The thickness of the tube in the test section in the middle of
after the two ends were machined at perfect right angles to the speci- the specimen was chosen based on avoiding local elastic buckling in the
men axis. Geometrical properties of the specimens are shown in Fig. 1 tube wall under plastic compressive stressing. Proper selection of the
with the following variations in thickness: specimen geometry was necessary to ensure that the onset and growth
of wrinkling would approach that expected in a long uniform tube [2].
• Two thicker parts at far ends of the specimen (52 mm long), which
were left at the as-received diameter. 2.2. Artificial rectangular corrosion defects
• The test section in the middle of the specimen (76 mm long), which
was machined down to 2 mm wall thickness. The physical shape caused by the metal loss in a corroded tube is ir-
• Two linear tapers (50 mm long), which connected the test section to regular in depth and in surface. For engineering purposes, such as the
the thicker end segments. evaluation of the residual strength of a corroded pipeline, it is very com-
mon to represent these irregular defects with an equivalent rectangular
The tapers were long enough to minimise the thickness discontinuity shape (Fig. 2). The equivalent defect is expected to provide a conserva-
effects on the axial stresses in the test section. The choice of specimen ge- tive estimation for the residual strength of a corroded pipeline, as the
ometry was based on results of finite element modelling of the test spec- defect size is larger than that from irregular or parabolic shaped defects.
imen under monotonic uniaxial compression. The taper length was Defect depths greater than 85% of the original wall thickness (i.e., when
the remaining ligament is less than 15% of the original wall thickness)
Table 1 are not usually considered in the residual strength evaluation [17]. Cor-
Mechanical properties of the material from standard tensile tests. rosion, erosion or wear grooves wider than 60° are not also common in
offshore pipelines [12].
Material type Yield stress (MPa) Ultimate stress (MPa) Strain at rupture
In the experiments reported in this paper, similar to the majority of
Carbon steel X42/B 297 450 38% previous studies on corroded tubes (see Section 1), the corrosion de-
Carbon steel X56 400 550 32%
fects were simplified by external machine grooves (Fig. 2). They were
236 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246

Fig. 2. Left: Illustration of irregular and the equivalent rectangular corrosion defects [17], right: producing artificial material loss in the specimens.

constant in depth, uniform in axial direction and circumferentially be- It is reiterated the specimen geometry in the current study is close to
tween 45 and 60° wide. It is acknowledged that an equivalent rectangu- that used by Jiao and Kyriakides [1,2] and Paquette and Kyriakides [16].
lar corrosion defect results in sharp geometric discontinuities in the The current experiment is conducted on two types of carbon steel tubes
tube body, causing stress concentration around the defect corners. Influ- but they used a single but and very different type of material (SAF2507
ence of the stress concentrations on the overall static behaviour of the super-duplex seamless tube). The Jiao and Kyriakides [1,2] and Paquette
tube under inelastic loading might be assumed negligible. These, how- and Kyriakides [16] experiments were conducted on intact tubes. The
ever, will almost certainly have an effect on the fatigue strength of the current study includes intact tubes as well as defective tubes. The latter
defective tube as stresses tend to severely intensify near the edges of has an imperfection in the form of a rectangular thinner section ma-
the defective area. The shape, width and depth of the equivalent corro- chined on one side of the test section. There is also a difference in the
sion pits will affect the fatigue strength. current test setup with that used by Jiao and Kyriakides [1,2] and
Specimens' dimensions were carefully measured before and after Paquette and Kyriakides [16]. They used fully encastred boundary con-
carrying out the tests. A coordinate measuring machine (CMM) and a ditions at both ends of the specimens. In the current study the lower
digital three dimensional (3D) surface acquisition and reconstruction end of the specimen is supported by a spherical bearing that allows
technique (will be discussed in details in Section 5) were used for this bending of the specimen (see Fig. 3). The bearing does not seem to
purpose. The mean values for the dimensions of eight defective speci- have much influence on the intact specimens but plays a pivotal role
mens', measured by a coordinate measuring machines (CMM), are for the imperfect specimens where the asymmetry, whether under
given in Table 2. The table shows that the maximum deviation in cir- monotonic or cyclic compression, leads to bending.
cumferential angle for the artificial corrosion defect from the desired As aforementioned, the aim of the current study is to examine the
angle was 1.19%. The corresponding ratios for the defect depth and behaviour of a long steel offshore pipeline. It is acknowledged that it is
the defect length were 0.9% and 0.8%, respectively. The drill's shape very difficult to perfectly replicate the in-place boundary conditions of
had also some effects on the boundaries of the artificial corrosion defect a long pipeline in the laboratory on a short specimen. The authors, how-
causing these edges to be slightly off the vertical. ever, believe that the end conditions considered in the current study
represent most closely the actual restraints in an offshore pipeline
3. Loading system under axial compression. Offshore pipelines are normally installed un-
buried so, in contrast to onshore lines entrenched and buried in a
A servo-hydraulic dynamic testing system (Instron 8502) was trench; an offshore line is very prone to nonuniaxial deformations.
employed for introducing uniaxial monotonic and cyclic loads to Under an axial compressive load, and when this load reaches some crit-
the specimens. Two interface segments were placed between the ical value, the pipe may experience vertical (upheaval buckling) or lat-
specimen's ends and the loading parts to act as load centralisers. Fig. 3 eral (snaking buckling) movements that can jeopardise the structural
shows a schematic view of the testing set-up. The loading history and integrity of the pipeline. The end bearing considered in the current ex-
axial shortening in the specimen were recorded by the testing device. Ex- perimental setup allows for this type of nonuniaxial deformation
tensometers were also used to record the axial shortening along the main when the compression exceeds certain limits, although the potential
section in the middle of the specimen (Fig. 3). Strain gauges were fixed for overall bending deformations is, to some degrees, restricted by the
around the main section to monitor the local strains in the specimen thicker ends of the specimen. This is comparable to the real structure
and to detect possible existing eccentricity in the axial loading. where the adjacent perfect lengths of the pipe impose some restrictions
to bending to the defective part, but the buckling and bending are not
Table 2 totally precluded.
Mean values obtained from coordinate measuring machine (CMM) for the dimensions of
eight defective specimens. 4. Results and discussions
Specimen Overall Outer Inner Corrosion Corrosion Corrosion
ID length diameter diameter length depth width Table 3 lists those tests whose results are reported in this paper. It
(mm) (mm) (mm) (mm) (mm) (degrees) can be noticed that two types of loading (monotonic compression and
CM1(І) 279.968 53.80 49.80 76.64 0.991 58.83 uniaxial cyclic loading), two grades of materials (X42/B and X56), two
CC1(І) 279.934 53.73 49.80 76.24 1.036 58.87 types of specimens (intact and defective) and two widths for the rectan-
CC2(І) 279.983 53.73 49.80 76.10 1.093 59.07
gular corrosion defect (45 and 60°) were considered. In the specimen ID,
CC3(І) 279.983 53.80 49.73 76.10 1.093 59.07
CM2(П) 279.980 53.70 49.86 76.68 0.996 59.83 the first letter (I or C) indicates whether the specimen was intact or de-
CM3(П) 279.985 53.74 49.80 76.54 0.992 44.76 fective. The next letter (M or C) shows if the loading was monotonic or
CC4(П) 279.965 53.81 48.77 76.23 1.001 59.76 cyclic. The ensuing number gives the specimen number in each catego-
CC5(П) 279.987 53.82 49.75 76.56 1.087 59.47 ry. The last letter in parentheses shows the material grades (I for X42/B
Mean values 279.973 53.77 49.66 76.39 1.036 57.46
and II for X56).
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 237

Fig. 3. Testing set-up: a schematic view.

4.1. Monotonic loading: intact specimens
Typical deformed shapes of the intact specimen IM1(I) prior and
after the testing are given in Fig. 4. Fig. 5a shows the axial stress–
strain response of the intact specimen IM1(І) under monotonic
displacement-controlled uniaxial compression. The abscissa in Fig. 5
presents the overall axial strain (axial shortening of the test section of
the specimen divided to its length). The ordinate in Fig. 5 reports the
mean axial stress (the ratio of the axial load in the specimen to the net
cross section area at its mid length).

Table 3
List of specimens and their properties.

Specimen's ID Material grade Specimen condition Load condition Defect depth (mm) Defect width (degrees)

IM1(І) Carbon steel X42/B Intact Monotonic – –

IM2 (П) Carbon steel X56 Intact Monotonic – –
IC1(І) Carbon steel X42/B Intact Cyclic – –
IC2(І) Carbon steel X42/B Intact Cyclic –
IC3(І) Carbon steel X42/B Intact Cyclic – –
IC4(І) Carbon steel X42/B Intact Cyclic – –
IC5 (П) Carbon steel X56 Intact Cyclic – –
CM1(І) Carbon steel X42/B Corroded Monotonic 1 60
CC1(І) Carbon steel X42/B Corroded Cyclic 1 60
CC2(І) Carbon steel X42/B Corroded Cyclic 1 60
CC3(І) Carbon steel X42/B Corroded Cyclic 1 60
CM2 (П) Carbon steel X56 Corroded Monotonic 1 60
CM3 (П) Carbon steel X56 Corroded Monotonic 1 45
CC4 (П) Carbon steel X56 Corroded Cyclic 1 60
CC5 (П) Carbon steel X56 Corroded Cyclic 1 60
238 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246

4.2. Monotonic loading: defective specimens

The specimens studied had an artificial rectangular corrosion defect

Fig. 4. Intact specimen IM1(І) before (left) and after (right) the monotonic test.
With the intact specimen, around overall axial strain of 2% (point B
in Fig. 5a), virtually axisymmetric wrinkles started to take shape in the
middle (main) part of the specimen. At the beginning they were small
in amplitude. The wrinkles amplitude grew larger as the axial compres-
sion increased. Radial amplification of the wrinkles was causing degra-
dation in the axial rigidity. Beyond a limit stress (point D in Fig. 5a)
there was a decrease in the load bearing as the tube deformations local-
ised. Growth of the local plastic buckling, overall buckling and collapse
were the ensuing consequences of the enlarging wrinkles.
Fig. 5b shows the axial stress–strain records for the intact specimen
IM2(П) under monotonic displacement-controlled compression. The
material grade for this specimen was X56 steel. It should be mentioned
that, with this type of material, visible wrinkles initiated around a global
strain of 0.5%.
Local strain time-histories during the monotonic test on specimen
IM1(I), from gauges placed in the middle part of the specimen at ap-
proximately 90° intervals, are reported in Fig. 6a.
on their outer side. The defective area was 1 mm deep, 76 mm long and
circumferentially 45 or 60° wide. Fig. 5a also presents the axial strain–
stress records (over the entire length of the test section) for the
defective specimen CM1(І) tested under monotonic displacement-
controlled axial compression. The material grade for this specimen
was X42/B steel. With the defective specimen CM1(І), around axial
strain of 1.5% (point A in Fig. 5a), first wrinkles started to appear in
the middle part of the specimen. This strain limit was virtually below
the corresponding strain with the intact specimen (point B in Fig. 5a
at around 2%). The wrinkles were non-axisymmetric. They first
emerged in the defective region and gradually expanded to the entire
circumference, as the axial compression increased.
The wrinkles were accompanied by ovalisation in the pipe cross
section. The ovalisation was caused by bending actions due to the
eccentricity of the neutral axis of the defective cross section rela-
tive to the applied load. With further increase in the axial load,
the wrinkling, ovalisation and inward bending of the specimen
grew larger leading to degradation of the axial rigidity in the spec-
imen. As a result, the load bearing was bounded to a limit stress of
410 MPa (point C Fig. 5a). This limit stress was occurring at a global
strain of around 4.2%, lower than the corresponding strain for the
intact specimen (around 6%). The stress limit for the defective
specimen was approximately 2.4% below than that in the intact
specimen.

500
500

400
Axial Stress σ (MPa)

400 D
Axial Stress σ (MPa)

C
300 A B 300

200 200
Defected specimen (60° wide defect)
Defected specimen (45° wide defect)
100 100
Intact specimen
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0 0.002 0.004 0.006 0.008 0.01
Axial Strain ε Axial Strain ε
a) Material grade X42B: b) Material grade X56:
Intact (IM1(I)) and defective (CM1(I)) specimens Intact (IM2(Π)) and defective (CM2(Π), CM3(Π) specimens

Fig. 5. Overall axial strain–stress records for the intact and defective specimens from two different material grades under monotonic displacement-controlled compression. a) Material
grade X42B: intact (IM1(І)) and defective (CM1(І)) specimens. b) Material grade X56: intact (IM2(П)) and defective (CM2(П)), CM3(П) specimens.

0.04 0.03
Damaged part of the
0.035 defected specimen
0.025 Intact specimen
0.03 Perfect part of the
0.02 defected specimen
0.025
Strain

Strain

0.02 Channel1 0.015

0.015 Channel2
0.01
0.01 Channel3
0.005
0.005

0 0
0 50 100 150 200 250 0 10 20 30 40 50 60
Time(s) Time(s)
a) Intact specimen (IM1(I)) b) Defective specimen (No. CM1(I))

Fig. 6. Local strain time-histories during the monotonic tests on the intact and defective specimens recorded by strain gauges placed in the middle part of the specimen at approximately
90° intervals. a) Intact specimen (IM1(І)). b) Defective specimen (No. CM1(І)).
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 239

a) Material grade X42B: Specimen CM1(I) b) Material grade X56: Specimens CM2(Π) and CM3(Π)

Fig. 7. Deformed shapes of two defective specimens after monotonic loading tests. a) Material grade X42B: Specimen CM1(І). b) Material grade X56: Specimens CM2(П) and CM3(П).

Beyond the stress limit (point C Fig. 5a) the tube deformations local-
ised which led to formation of a relatively sharp inward kink in the
defective side of the specimen. At higher strains, the specimen experi-
enced excessive bending, the kink became deeper and the pipe started
to buckle in a dogleg mode. Meanwhile, the ovalisation was amplified
and even extended to the taper parts, up and below the main middle
segment in the specimen.
Fig. 5b presents time series of the axial stress–strain records for the
defective specimens CM2(П) and CM3(П). The material grade for
these specimens was X56 steel. The defective areas were 45 and 60°
wide for the CM2(П) and CM3(П) specimens, respectively. As it can be
noticed in Fig. 5b the stress limit was not seriously affected by the defect
width and even by the defect presence. It is reiterated that stress values
in this figure are mean stresses obtained by dividing the axial loads to
the net cross section at mid height of the specimen.
Local strain time-histories during the monotonic test, from gauges
placed in the middle part of the specimen with approximately 90° inter-
vals, are reported in Fig. 6b. One strain gauge was placed on the
centreline of the defective area and one on the opposite side, on the per-
fect part (Fig. 7). The third gauge was placed, circumferentially, in be-
tween the other two gauges. The time-history diagrams (Fig. 6b)
indicate that the damaged area was initially experiencing higher axial
strains as compared to those in the perfect areas, so the plastic flow oc-
curred earlier in the damaged parts. Apparent differences can be noticed
between the local strain time series for the intact and the defective
specimens.
Post experiment deformed shapes of the defective specimen CM1(І),
CM2(П) and CM3(П) are given in Fig. 7. They show the non-
axisymmetric local wrinkles, the inward kink in the defective side of
the specimen and the overall bending occurred in these specimens.
4.3. Cyclic loading: intact specimens
Five intact specimens were considered for this series of tests. The
material grade for four specimens was X42/B and for one specimen
was X52 steel. The specimens were all first subjected to monotonic
axial compressions to attain an initial pre-defined non-linear axial
strain. The subsequent cyclic axial load was introduced to the specimen
in a load control mode. The mean and amplitude values for the cyclic
stress remained constant in each test but varied with different tests.
The material grade, the initial non-linear axial strain, the mean axial
stress, the axial stress amplitude and the frequency of the cyclic load
for these five test specimens are listed in Table 4.
Fig. 8 shows a typical axial stress–strain curve for one intact speci-
men (IC1(І)) subjected to uniaxial cyclic loading. The figure presents
the (overall) axial strains against the axial stress in the mid-height of
the specimen. The initial axial strain was 2% which corresponded to a
maximum stress of 348 MPa.
Fig. 9a depicts the axial plastic ratcheting strains against the load
cycle number for the IC1(І) specimen. From this figure, a tertiary (sig-
moid) ratcheting type of behaviour can be recognised in which the re-
sponse may be subdivided in three distinct zones. In the early stages
of cycling, the ratcheting strain was following a non-linear decelerating
path (see zone І in Fig. 9a). Afterward it was demonstrating a stabilised
response, with a nearly constant ratcheting strain rate (zone II in
Fig. 9a). The ratcheting strain was then gathering speed and followed
an exponential path (zone III in Fig. 9a). These three zones are also
marked on Fig. 9b. This figure gives the test results for specimen
IC5(П) which was from steel material grade X56. This type of tertiary
ratcheting behaviour was reported by other researchers such as Jiao
and Kyriakides [1,2] and Kang et al. [7].

Table 4
Properties of the intact specimens tested under uni-axial cyclic loads and summary of the results.

Specimen Steel material Initial axial Mean stress σm Stress amplitude σa Cyclic load Mean ratcheting Number of
ID grade strain (MPa) (MPa) frequency (Hz) rate (strain/cycle) cycles to failure

IC1(I) X42/B 2% 192 153 0.3 0.012% 98

IC2(I) X42/B 2% 238 110 0.3 0.0035% 1140
IC3(I) X42/B 0.3% 192 110 0.3 0.001% 4780
IC4(I) X42/B 4% 192 185 0.3 0.3% 4
IC5(П) X56 0.3% 256 180 0.3 0.013% 72
240 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246

400 2.25
IC1
IC2(I)
IC1(I)
IC2

Axial Ratcheting Strain (%)

2.20
Axial Stress σ (MPa)

300
2.15
Trend line
200 2.10

2.05
100
Trend line
2.00

0 1.95
0 0.005 0.01 0.015 0.02 0.025 0.03 0 10 20 30 40
Axial Strain ε Number of Cylces N

Fig. 8. Results from cyclic loading tests for the IC1(І) specimen. Fig. 10. Axial ratcheting strains against the load cycle number for the IC1(І) and IC2(І)
specimens.

The mean plastic ratcheting strain rate and the total number of load
cycles before overall buckling for the five intact specimens examined an apparent amplification of the wrinkling growth under cyclic loads. In
under cyclic axial loads are summarised in Table 4. The mean plastic this test, four load cycles were enough to produce a very rapid growth of
ratcheting strain rate in this table corresponds to the slope of the the wrinkling and to lead to extreme localised buckling and collapse de-
ratcheting strain curve over the linear path of the response (e.g., zone formations in the specimen. Jiao and Kyriakides [2] also showed that
II in Fig. 9). larger initial wrinkles (initial strains) tend to increase the rate of
Results obtained from specimens IC1(І) and IC2(І) indicate that the ratcheting and decrease the number of cycles before collapse.
increase in the cyclic stress amplitude had a major impact on the Fig. 9b gives a plot of the uniaxial ratcheting response for an intact
ratcheting response (Fig. 10). Table 4 and Fig. 10 suggest that the specimen made from X56 grade steel materials (specimen IC5 (П)).
shape of the initial non-linear path (zone І), the constant ratcheting The initial non-linear axial strain was 0.3%, corresponding to a maximum
rate (or the slope of the linear part of the response in zone II) and the axial stress of 436 MPa. The initial strain was just slightly higher than the
number of cycles to the failure were substantially affected by the varia- elastic limit and was relatively low as compared to those used for the
tions in the amplitude of the cyclic stress (σa). Empirical findings from specimens made from X42/B steel grade (see Table 4). The wrinkle
other studies corroborate these results. Jiao and Kyriakides [2] reported growth with the IC5 (П) specimen under cyclic load, however, was rela-
results from some ratcheting tests carried out under uniaxial cycling tively rapid (see Fig. 9b and Table 4), as compared to results on speci-
loads. The intact steel tubular specimens considered in their studies all mens made from X42/B steel grade. This might be attributed to the
had an average compressive initial strain of 2 or 3%, while the cycle am- differences in the mechanical properties of the two materials considered
plitudes were different. They reported that cyclic amplitude played a in the current study. The X42/B steel appears to be a hardening type ma-
significant role on the ratcheting strain rates. terial while the X56 steel seems to exhibit Lüders banding at the onset of
Results obtained from specimens IC2(І) and IC3(І) provide informa- yielding (see Fig. 5). In fact, interaction of the Lüders' bands material in-
tion related to the mean stress effects on the ratcheting response. stability, that nucleates and propagates along the specimens, and the
Table 4 shows that with a higher mean stress the strain ratcheting structural non-linearity, in the form of wrinkling in the tube wall, will
rate increases and the number of cycles to the failure decreases. These complicate the behaviour of the structure as well as the subsequent
results appear to agree with those obtained by other researchers in ratcheting [18,19]. In addition, a low plastic strain to failure characteris-
comparable conditions. tics can be observed in monotonic tests on X52 specimens (Fig. 5).
Test results from specimen IC2(І) and IC3(I) may reflect the effects
from the initial strains on the ratcheting response. Table 4 shows that 4.4. Cyclic loading: defective specimens
the mean ratcheting strain rate and the number of cycles to the failure
were appreciably affected by the increase in the initial strain in the spec- As aforementioned, the defective specimens had an artificial corro-
imen. With specimen IC4(І), this increase in the initial strain resulted in sion on their outer side. The material grade, the initial non-linear axial

3.1 1
Axial Ratcheting Strain (%)

2.9 0.8 Zone I Zone II Zone III

Axial Ratcheting Strain

2.7
Zone I 0.6
Zone II Zone III
2.5
0.4
2.3
0.2
2.1

1.9 0
0 20 40 60 80 100 0 20 40 60 80
Number of Load Cycles N Number of Load Cycles N
a) Material grade X42B: Specimen IC1(I) b) Material grade X56: Specimen IC5(Π)

Fig. 9. Axial ratcheting strains against the load cycle number for the intact specimens from different steel grades. a) Material grade X42B: Specimen IC1(І). b) Material grade X56: Specimen
IC5(П).
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 241

Table 5
Properties of the defective specimens tested under axial cyclic loads and summary of the results.

Specimen Steel material Initial axial Mean stress σm Stress amplitude σa Cyclic load Mean ratcheting Number of
ID grade strain (MPa) (MPa) frequency (Hz) rate (strain/cycle) cycles to failure

CC1(I) X42/B 2.4% 238 110 0. 3 0.9% 4

CC2(I) X42/B 0.2% 192 110 0. 3 0.0021% 2857
CC3(I) X42/B 0.3% 206 110 0. 3 0.0077% 779
CC4(I) X42/B 0.4% 221 110 0. 3 0.0173% 246
CC5(П) X56 0.4% 95 301 0. 3 0.024% 36
CC6(П) X56 0.3% 95 285 0. 3 0.019% 64

strain, the mean axial stress, the axial stress amplitude and the frequen-
cy of the cyclic load, for the six defective test specimens, are listed in
Table 5. The table also reports the mean ratcheting strain rate and the
total number of load cycles to the failure (overall buckling) in the
specimens.
The axial strain–stress curves for one intact specimen (IC2(I)) and
one defective specimen (CC1(I)) are shown in Fig. 11a. Both specimens
were of X42/B steel grade. Up to an initial strain, the specimens were
loaded monotonically under a displacement control mode. The cyclic
load regime, in terms of the mean axial stress and the axial stress ampli-
tude, also remained the same with both the intact and the defective
specimens (see Tables 4 and 5). It should be mentioned that for this,
the average axial stresses at mid-height of both specimens were kept
similar. It means that the axial load and the load amplitude in the defec-
tive specimen were comparatively lower than those in the intact speci-
men. The decrease was proportional to the area loss in the cross section
of the defective specimen at its mid-height.
As it can be seen in Fig. 11, the ratcheting performance of the defec-
tive specimen was remarkably degraded in comparison to that for the
corresponding intact specimen. In contrast, as it was reported in
Section 4.2, the monotonic responses of the defective and intact speci-
mens remained relatively close to each other (see Fig. 5a). Under mono-
tonic loading, the strains for the onset of wrinkling in the intact and
defective specimens were around 0.5% different. The difference for the
stress limits was just around 3% (Fig. 5a). Under cyclic loading the num-
ber of cycles to failure in the intact specimen was around 1140 cycles
but the corroded specimen failed after only 4 cycles (Fig. 11a). The
mean ratcheting rate with the defective specimen was 0.9%/cycle
(Table 5), which is around 260 times higher than that with the intact
specimen (0.0035%/cycle).
In Fig. 11b the axial ratcheting strains in the intact specimen
IC3(I) and the defective specimen CC3(I) are plotted against the load
cycle numbers. These two specimens were of an identical steel grade
and experienced similar cyclic stress histories and conditions. Fig. 11b
indicates that the ratcheting strain rates in the defective specimen
were significantly amplified, in comparison to those in the intact speci-
men. The ratcheting strain rate was around 0.008% for the defective
specimen CC3(I) while the corresponding value for the intact specimen
IC3(I) was around 0.001%.
Local axial strain time-histories in the defective specimen CC1(I),
during the cyclic tests, are given in Fig. 12a. The data was obtained
from strain gauges placed at mid-height of the specimen. One strain
gauge was positioned on the centreline of the damaged part and one
on the opposite side (perfect part) of the specimen. Fig. 12a re-
indicates that the specimen CC1(I) failed after 4 load cycles (similar to
that observed in Fig. 11a).
Fig. 12a demonstrates that the plastic strains in the defective part
were higher than their corresponding values in the perfect part of the
specimen. As a result, the potential for the initiation and development
of wrinkling in the damaged areas is higher than in the perfect areas
of a defective specimen. As the number of load cycles increased, the per-
fect area was experiencing tensile strains due to the specimen's bending.
Fig. 12b shows the local axial strain time-histories during the cyclic
tests on the defective specimen CC3(I) from strain gauges placed
around mid-height of the specimen. It can be noticed that the ratcheting
response in the damaged and perfect parts are distinguishably different.
The strains in the defective area had higher amplitudes which were in-
creasing by time; ratcheting with a higher rate; and had a relatively high
compressive mean value that increased by time. The strains in the intact
area had lower amplitude which remained almost constant with time;
showed a very low ratcheting rate; and had a low mean strain that
was slightly reducing by time (moving slightly from compression to-
wards zero).
Results obtained from specimens CC2(I) to CC4(I) can be used for a
judgement about effects from the initial non-linear axial strain on the
ratcheting response. As indicated in Table 5, the three specimens had
similar stress amplitudes. As the initial axial strain with the CC2(I),
CC3(I) and CC4(I) increased from 0.2% to 0.3% and 0.4%, respectively
the mean ratcheting strain rates increased from 0.0021% to 0.0077%
and 0.0173. Fig. 13a also shows that a relatively small increase in the ini-
tial strain in a defective specimen caused the ratcheting rate to be al-
most doubled. It means that the defective specimens were more
responsive to the initial non-linear strain and to the changes in the cy-
clic load conditions as compared to the intact specimens (see Table 4).

400 0.8
Axial Ratcheting Strain (%)

0.7
300 0.6
Axial Stress σ (MPa)

0.5
200 0.4
0.3
100 0.2
0.1
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 10 20 30 40 50 60
Axial Strain ε
Number of Cycles N
a) Stress-strain results for the intact IC2(I) specimen b) Ratcheting strain results for the intact specimen
and the defective CC1(I) specimen IC3(I) and the defective specimen CC3(I)

Fig. 11. Experimental results from cyclic tests on an intact and defective specimen subjected to similar cycling regimes. a) Stress–strain results for the intact IC2(I) specimen and the
defective CC1(I) specimen. b) Ratcheting strain results for the intact specimen IC3(I) and the defective specimen CC3(I).
242 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246
a) Specimen CC1(I)
Fig. 12. Local axial strain time-histories in the damaged and perfect parts of two defective specimens under different cycling regimes. a) Specimen CC1(I). b) Specimen CC3(I).
Fig. 13b gives a comparison between the local and global (overall)
axial ratcheting strains against the load cycle numbers during the cyclic
tests on the defective specimen CC3(I). Local strains were obtained from
strain gauges placed on the centreline of the damaged part at the spec-
imen mid-height. Global (overall) strains were obtained by dividing the
axial shortening in the specimen to its overall length. The figure, once
again, highlights that the imperfect region in a defective tube ratchets
very differently from those for the global tube and the intact part of
the tube.
A close view of the deformed shape of a defective specimen (CC3(I))
after 300 load cycles is given in Fig. 14a. It shows formation of localised
wrinkles in the defective area. Fig. 14b shows the sequences of forma-
tion and propagation of wrinkles towards failure in defective specimens,
with the increase in the load cycle number.
Table 5 also reports the results from cyclic tests on defective speci-
mens from X56 steel material. It can be seen that defective specimens
from this steel grade were also very responsive to the changes made
in the cyclic load conditions. An increase of around 5.6% in the ampli-
tude of cyclic load in specimen CC5(I) relative to that in specimen
CC6(II) reduced the number of cycles to failure to almost half.
5. Non-contact surface acquisition and reconstruction
Non-contact 3D optical digitising system was used in the current
study for the 3D surface acquisition and reconstruction of the speci-
mens' geometry and for evaluating their wrinkling and ratcheting re-
sponses. This is a camera-based triangulation system. Combined with
digital photogrammetry techniques, a processing unit employs optical
transform equations to automatically and with a great accuracy
2.0
1.8
Axial Ratcheting Strain (%)
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0 50 100
Number of Cycles N
a) Effects from the initial strain on the ratchet strains in two
defective specimens
Fig. 13. Experimental results for the ratcheting responses of defective specimens CC2(I) and CC3(I). a) Effects from the initial strain on the ratchet strains in two defective specimens.
b) Defective specimen (CC3(I)): Global (overall) ratchet strains and local ratchet strains (in the damaged part).
b) Specimen CC3(I)
calculate 3D coordinates for every pixel of the cameras. Details of the
optical 3D surface acquisition and reconstruction programme and its re-
sults will be reported in a separate paper and here a summary is
presented.
The measurement strategy of this type of optical surface contour
sensing systems is triangulation between two image rays. This was
achieved by capturing two images with a distance from the object
surface. For convenience of providing dense and accurate points in
the scan images, a structured light pattern is projected to the speci-
men surface by a digital light processing (DLP) projector. So, the sur-
face of the specimen obtains an appropriate texture making it
possible to trace corresponding points between the two scan images.
Depending on camera resolution a cloud of up to 4 million points was
obtained for every single measurement. Lower cost, higher speed
and higher density of the scan points are the main advantages of op-
tical scanners in comparison to touch based CMMs. Maximum accu-
racy of the measured points by this system was about 5 μm but for
this experimental study the specimens' surface was reconstructed
with 30 micron accuracy.
The 3D surface acquisition and reconstruction system provided high
accuracy digitised measurements on the “As Is” geometry of the speci-
mens. Such a detailed 3D model made it possible to detect the slightest
geometrical imperfections in the fabricated specimens before carrying
out the tests. The optical digitising system was also used to trace the
specimens; deviations from its initial geometry during and after carry-
ing out the tests. Pai and Young [20] employed a similar system, based
on the scanning laser vibrometers, for boundary effect detection (BED)
aiming at pinpointing locations of small damages in beams using oper-
ational deflection shapes (ODS) method.
2.5
Global ratchet strain
Axial Ratcheting Strain (%)
2
Local ratchet strain
1.5
CC2(I)
1
0.5
0
150 200 0 50 100 150 200 250 300
Number of Cycles N
b) Defective specimen (CC3(I)): Global (overall) ratchet
strains and local ratchet strains (in the damaged part)
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 243

a) Close view of specimen b) The sequences of formation and propagation of

CC3(I) after 300 load cycles wrinkles towards failure in defective specimens:
(Left CC3(I), CC2(I) Centre and Right CC1(I))

Fig. 14. Local wrinkle formation and development in defective specimen under axial cyclic loading. a) Close view of specimen CC3(I) after 300 load cycles. b) The sequences of formation
and propagation of wrinkles towards failure in defective specimens: (Left CC3(I), CC2(I) Centre and Right CC1(I)).

5.1. Surface acquisition and reconstruction
Fig. 15 presents some typical 3D surface acquisition and reconstruc-
tion results for a number of intact and defective specimens at different
stages of loading. The figure shows the shaded point cloud fitted to
the surface of the deformed specimens.
Fig. 15a shows the surface reconstruction results for the intact spec-
imen IM1(І), which was subjected to monotonic axial loads. The wrin-
kling patterns in this specimen can be discerned and as will be shown
later they were quantitatively measured and described. Fig. 15b and c
gives two different views of the digitised post buckling surfaces of the
intact specimen IC1(I). This specimen was subjected to cyclic axial
loads up to the failure point. Wrinkling formation in the defective spec-
imen CC2(І) was optically reconstructed and shown in Fig. 15d. This
specimen was subjected to uniaxial cycling and the figure gives the
specimen status after 300 load cycles.
Fig. 15 indicates that with the optical surface contour sensing, accu-
rate 3D representation of the specimen (up to 30 μm) can efficiently and
effectively be generated. This makes utilising traditional contact based
measuring methods both cumbersome and time consuming.

a) Wrinkling b) Post buckling c) Post buckling d) Wrinkling

formation in the status in the intact status in the intact formation in the
intact specimen specimen IC1(I) specimen IC1(I) defective specimen
(IM1(I)) (CC2(I))

Fig. 15. 3D surface acquisition and reconstruction results which show the shaded point cloud fitted to the surface of the intact and defective specimens at different stages of loading.
a) Wrinkling formation in the intact specimen (IM1(І)). b) Post buckling status in the intact specimen IC1(I). c) Post buckling status in the intact specimen IC1(I). d) Wrinkling formation
in the defective specimen (CC2(І)).
244 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246

2.5

Radial Displacement (mm)

2

1.5

0.5
Sec A-A
0

-0.5
0 0.2 0.4 0.6 0.8 1

Longitudinal Distance/L
a) Longitudinal section of the specimen obtained from optical b) Longitudinal profile at Sec. A-A
system

Fig. 16. Longitudinal profiles of the defective specimen IM1(І) obtained from the 3D optical surface contour sensing. a) Longitudinal section of the specimen obtained from optical system.
b) Longitudinal profile at Sec. A–A.

5.2. Longitudinal profiling
Fig. 16a gives a longitudinal section of the specimen IM1(І) obtained
from the optical system. This intact specimen was tested under mono-
tonic axial loads. Three virtually axisymmetric wrinkles can be
recognised in the middle (main) part of the specimen. As an output of
the 3D surface acquisition and reconstruction, the longitudinal profiles
can be easily extracted from these data. Fig. 16b shows the radial dis-
placement in the specimen wall along its middle length.
Fig. 17a gives a longitudinal section of the deformed shape for the
specimen CM1(І), obtained from the optical system. This was a defec-
tive specimen examined under a uniaxial monotonic load regime.
Non-axisymmetric wrinkles can be identified in this figure. As it was de-
scribed in Section 4.2, this non-axisymmetric nature of the wrinkles is a
characteristic of the plastic buckling and ratcheting in defective speci-
mens. The surface mesh outputs for the middle part of the specimen
at axial strain of 0.062 are shown in Fig. 17b. Fig. 17c gives the longitu-
dinal profile for the radial displacements at the centreline of the dam-
aged area. Fig. 17c indicates that, in addition to the wrinkling, the
specimen experienced a bending deformation towards the damaged
part. Fig. 17d gives the longitudinal profile at a 90° interval. These two
figures, once again, point that the local wrinkling and ratcheting re-
sponses in the damaged and perfect parts of a specimen are significantly
different.
5.3. Circumferential profiling
Fig. 18 gives the deformed shapes of the specimen CM1(І) obtained
from the optical system. This was a defective specimen examined under
a monotonic load regime. The circumferential profiles for the radial dis-
placements in the specimen's wall, at three different elevations along its
middle part, are also given in the figure. They highlight that under

Sec B-B

Sec A-A

a) Longitudinal section of the specimen obtained from b) Surface mesh output for the middle part of the specimen
optical system obtained from optical system
3 0
0 0.2 0.4 0.6 0.8 1
2 -0.5
Radial Displacement (mm)
Radial Displacement (mm)

-1
1
-1.5
0
0 0.2 0.4 0.6 0.8 1 -2
-1 -2.5

-2 -3
-3.5
-3
-4
-4
Longitudinal Distance/L Longitudinal Distance/L
c) Longitudinal profile at Sec. A-A d) Longitudinal profile at Sec. B-B

Fig. 17. Longitudinal profiles of the defective specimen CM1(І) obtained from the 3D optical surface contour sensing. a) Longitudinal section of the specimen obtained from optical system.
b) Surface mesh output for the middle part of the specimen obtained from optical system. c) Longitudinal profile at Sec. A–A. d) Longitudinal profile at Sec. B–B.
 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246
SEC 1-1
Sec. 1-1
Fig. 18. Circumferential profiles of the defective specimen CM1(І) at axial strain of 0.062 from the 3D optical surface contour sensing.
monotonic loading the wrinkles in the defective specimens were non-
axisymmetric. They also show that, in addition to the wrinkling and
longitudinal curvature, defective specimens were experiencing
ovalisations in their cross sections.
6. Concluding remarks
In the present paper, wrinkling and strain ratcheting response of the
intact and defective cylindrical steel tubes, under monotonic and cyclic
uniaxial loads, were experimentally investigated. This type of damage,
as an example, might be caused by repetition of start-up/shutdown
and temperature cycles in an offshore pipeline. Small scale steel cylin-
drical specimens with and without an artificial rectangular corrosion
were examined under both monotonic and cyclic uniaxial loads. Differ-
ent material grades, loading configurations and rectangular corrosion
sizes were considered. It was noticed that:
• Under monotonic axial compression, the stress limit for the defective
specimens was slightly (few percentages) below the corresponding
value for the intact specimens. The onset of wrinkling was around
20% lower than that with the intact specimen. It was concluded that
in general, the overall stress–strain response of the specimens under
monotonic loading was not seriously affected by the presence of
those rectangular corrosion defects considered in the current study.
The monotonic response was noticed to be more severely affected
by the steel grade than that from the rectangular corrosion defect
presence and its size.
• The inelastic cyclic response of the specimens was remarkably affect-
ed by the presence of the corrosion defect. In contrast, as mentioned
above, the overall monotonic strain–stress responses of the defective
and intact specimens remained relatively close.
• In the defective tubes the local wrinkling first initiated from the dam-
aged part. This local plastic buckling then gradually proceeded to the
245
SEC 2-2 SEC 3-3
Sec. 2-2 Sec. 3-3
entire circumference. Wrinkles in defective specimens, under both
monotonic and cyclic loadings, in contrast to the intact specimens,
were non-axisymmetric. Defective specimens also experienced some
degrees of overall bending and ovalisation in their cross sections. As
the axial strain grew higher, the ovalisation tended to be amplified
and extended to the thicker tapered parts of the specimen.
• An optical digitising system was used for 3D surface acquisition and
reconstruction of the tested specimen and to evaluate their strain
ratcheting and wrinkling response. Outputs of the optical digitising
system were well able to demonstrate the growth and eventual
localisation of wrinkles in intact and corroded specimens. Vertical,
longitudinal and circumferential cross sections obtained from the op-
tical digitising system provided useful information on the extent and
configuration of the wrinkling and ovalisations in the corroded speci-
mens.
• With the defective tubes the strain ratcheting behaviour in the defec-
tive zone was distinctively different from that in the perfect zones of
the circumference. The ratcheting strains and strain rates in the defec-
tive zone of the specimen were considerably higher than the corre-
sponding values in the perfect zones. The ratcheting strains in the
defective area very rapidly turned exponential, while the ratcheting
strains in the perfect zone still remained in the linear trajectory.
• The cyclic response of the specimens was noticed to be apparently
much more sensitive to the presence of corrosion than that under
monotonic loading.
References
[1] R. Jiao, S. Kyriakides, Wrinkling of tubes by axial cycling, J. Appl. Mech. 77 (3) (2010)
031012.
[2] R. Jiao, S. Kyriakides, Ratcheting, wrinkling and collapse of tubes under axial cycling,
Int. J. Solids Struct. 46 (14) (2009) 2856–2870.
[3] R. Jiao, S. Kyriakides, Ratcheting and wrinkling of tubes due to axial cycling under
internal pressure: part I experiments, Int. J. Solids Struct. 48 (20) (2011) 2814–2826.
246 M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246
[4] R. Jiao, S. Kyriakides, Ratcheting and wrinkling of tubes due to axial cycling under
internal pressure: part II analysis, Int. J. Solids Struct. 48 (20) (2011) 2827–2836.
[5] M. Zeinoddini, M. Peykanu, Strain ratcheting of steel tubulars with a rectangular de-
fect under axial cycling: a numerical modeling, J. Constr. Steel Res. 67 (12) (2011)
1872–1883.
[6] S.C. Kulkarni, Y.M. Desai, T.A.R.U.N. Kant, G.R. Reddy, Y. Parulekar, K.K. Vaze, Uniaxial
and biaxial ratchetting study of SA333 Gr. 6 steel at room temperature, Int. J. Press.
Vessel. Pip. 80 (3) (2003) 179–185.
[7] G.Z. Kang, Y.G. Li, J. Zhang, Y.F. Sun, Q. Gao, Uniaxial ratcheting and failure behaviors
of two steels, Theor. Appl. Fract. Mech. 43 (2) (2005) 199–209.
[8] S.K. Paul, S. Sivaprasad, S. Dhar, S. Tarafder, True stress control asymmetric cyclic
plastic behavior in SA333 C–Mn steel, Int. J. Press. Vessel. Pip. 87 (8) (2010)
440–446.
[9] M. Zeinoddini, M. Ezzati, J. Fakheri, Uniaxial strain ratcheting behavior of dented
steel tubular: an experimental study, Eng. Fail. Anal. 44 (2014) 202–216.
[10] C. Gupta, J.K. Chakravartty, G.R. Reddy, S. Banerjee, Uniaxial cyclic deformation be-
haviour of SA 333 Gr 6 piping steel at room temperature, Int. J. Press. Vessel. Pip.
82 (6) (2005) 459–469.
[11] T.A. Netto, A simple procedure for the prediction of the collapse pressure of pipe-
lines with narrow and long corrosion defects—correlation with new experimental
data, Appl. Ocean Res. 32 (1) (2010) 132–134.
[12] N. Sakakibara, S. Kyriakides, E. Corona, Collapse of partially corroded or worn pipe
under external pressure, Int. J. Mech. Sci. 50 (12) (2008) 1586–1597.
[13] T.A. Netto, U.S. Ferraz, A. Botto, On the effect of corrosion defects on the collapse
pressure of pipelines, Int. J. Solids Struct. 44 (22) (2007) 7597–7614.
[14] J. Xue, A non-linear finite-element analysis of buckle propagation in subsea corrod-
ed pipelines, Finite Elem. Anal. Des. 42 (14) (2006) 1211–1219.
[15] J. Xue, M.S. Hoo Fatt, Symmetric and anti-symmetric buckle propagation modes in
subsea corroded pipelines, Mar. Struct. 18 (1) (2005) 43–61.
[16] J.A. Paquette, S. Kyriakides, Plastic buckling of tubes under axial compression and in-
ternal pressure, Int. J. Mech. Sci. 48 (8) (2006) 855–867.
[17] DNV Offshore Standard, Corroded Pipeline, Det Norske Veritas, 2011. (DNV-RP-F101).
[18] J.F. Hallai, S. Kyriakides, On the effect of Lüders bands on the bending of steel tubes.
Part I: experiments, Int. J. Solids Struct. 48 (24) (2011) 3275–3284.
[19] J.F. Hallai, S. Kyriakides, Underlying material response for Lüders-like instabilities,
Int. J. Plast. 47 (2013) 1–12.
[20] P.F. Pai, L.G. Young, Damage detection of beams using operational deflection shapes,
Int. J. Solids Struct. 38 (18) (2001) 3161–3192.