Professional Documents
Culture Documents
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
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
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
4.1. Monotonic loading: intact specimens displacement-controlled uniaxial compression. The abscissa in Fig. 5
presents the overall axial strain (axial shortening of the test section of
Typical deformed shapes of the intact specimen IM1(I) prior and the specimen divided to its length). The ordinate in Fig. 5 reports the
after the testing are given in Fig. 4. Fig. 5a shows the axial stress– mean axial stress (the ratio of the axial load in the specimen to the net
strain response of the intact specimen IM1(І) under monotonic 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)
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
0.015 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- 4.3. Cyclic loading: intact specimens
ised which led to formation of a relatively sharp inward kink in the
defective side of the specimen. At higher strains, the specimen experi- Five intact specimens were considered for this series of tests. The
enced excessive bending, the kink became deeper and the pipe started material grade for four specimens was X42/B and for one specimen
to buckle in a dogleg mode. Meanwhile, the ovalisation was amplified was X52 steel. The specimens were all first subjected to monotonic
and even extended to the taper parts, up and below the main middle axial compressions to attain an initial pre-defined non-linear axial
segment in the specimen. strain. The subsequent cyclic axial load was introduced to the specimen
Fig. 5b presents time series of the axial stress–strain records for the in a load control mode. The mean and amplitude values for the cyclic
defective specimens CM2(П) and CM3(П). The material grade for stress remained constant in each test but varied with different tests.
these specimens was X56 steel. The defective areas were 45 and 60° The material grade, the initial non-linear axial strain, the mean axial
wide for the CM2(П) and CM3(П) specimens, respectively. As it can be stress, the axial stress amplitude and the frequency of the cyclic load
noticed in Fig. 5b the stress limit was not seriously affected by the defect for these five test specimens are listed in Table 4.
width and even by the defect presence. It is reiterated that stress values Fig. 8 shows a typical axial stress–strain curve for one intact speci-
in this figure are mean stresses obtained by dividing the axial loads to men (IC1(І)) subjected to uniaxial cyclic loading. The figure presents
the net cross section at mid height of the specimen. the (overall) axial strains against the axial stress in the mid-height of
Local strain time-histories during the monotonic test, from gauges the specimen. The initial axial strain was 2% which corresponded to a
placed in the middle part of the specimen with approximately 90° inter- maximum stress of 348 MPa.
vals, are reported in Fig. 6b. One strain gauge was placed on the Fig. 9a depicts the axial plastic ratcheting strains against the load
centreline of the defective area and one on the opposite side, on the per- cycle number for the IC1(І) specimen. From this figure, a tertiary (sig-
fect part (Fig. 7). The third gauge was placed, circumferentially, in be- moid) ratcheting type of behaviour can be recognised in which the re-
tween the other two gauges. The time-history diagrams (Fig. 6b) sponse may be subdivided in three distinct zones. In the early stages
indicate that the damaged area was initially experiencing higher axial of cycling, the ratcheting strain was following a non-linear decelerating
strains as compared to those in the perfect areas, so the plastic flow oc- path (see zone І in Fig. 9a). Afterward it was demonstrating a stabilised
curred earlier in the damaged parts. Apparent differences can be noticed response, with a nearly constant ratcheting strain rate (zone II in
between the local strain time series for the intact and the defective Fig. 9a). The ratcheting strain was then gathering speed and followed
specimens. an exponential path (zone III in Fig. 9a). These three zones are also
Post experiment deformed shapes of the defective specimen CM1(І), marked on Fig. 9b. This figure gives the test results for specimen
CM2(П) and CM3(П) are given in Fig. 7. They show the non- IC5(П) which was from steel material grade X56. This type of tertiary
axisymmetric local wrinkles, the inward kink in the defective side of ratcheting behaviour was reported by other researchers such as Jiao
the specimen and the overall bending occurred in these specimens. 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
400 2.25
IC1
IC2(I)
IC1(I)
IC2
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 (%)
2.7
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
strain, the mean axial stress, the axial stress amplitude and the frequen- specimen CC3(I) while the corresponding value for the intact specimen
cy of the cyclic load, for the six defective test specimens, are listed in IC3(I) was around 0.001%.
Table 5. The table also reports the mean ratcheting strain rate and the Local axial strain time-histories in the defective specimen CC1(I),
total number of load cycles to the failure (overall buckling) in the during the cyclic tests, are given in Fig. 12a. The data was obtained
specimens. from strain gauges placed at mid-height of the specimen. One strain
The axial strain–stress curves for one intact specimen (IC2(I)) and gauge was positioned on the centreline of the damaged part and one
one defective specimen (CC1(I)) are shown in Fig. 11a. Both specimens on the opposite side (perfect part) of the specimen. Fig. 12a re-
were of X42/B steel grade. Up to an initial strain, the specimens were indicates that the specimen CC1(I) failed after 4 load cycles (similar to
loaded monotonically under a displacement control mode. The cyclic that observed in Fig. 11a).
load regime, in terms of the mean axial stress and the axial stress ampli- Fig. 12a demonstrates that the plastic strains in the defective part
tude, also remained the same with both the intact and the defective were higher than their corresponding values in the perfect part of the
specimens (see Tables 4 and 5). It should be mentioned that for this, specimen. As a result, the potential for the initiation and development
the average axial stresses at mid-height of both specimens were kept of wrinkling in the damaged areas is higher than in the perfect areas
similar. It means that the axial load and the load amplitude in the defec- of a defective specimen. As the number of load cycles increased, the per-
tive specimen were comparatively lower than those in the intact speci- fect area was experiencing tensile strains due to the specimen's bending.
men. The decrease was proportional to the area loss in the cross section Fig. 12b shows the local axial strain time-histories during the cyclic
of the defective specimen at its mid-height. tests on the defective specimen CC3(I) from strain gauges placed
As it can be seen in Fig. 11, the ratcheting performance of the defec- around mid-height of the specimen. It can be noticed that the ratcheting
tive specimen was remarkably degraded in comparison to that for the response in the damaged and perfect parts are distinguishably different.
corresponding intact specimen. In contrast, as it was reported in The strains in the defective area had higher amplitudes which were in-
Section 4.2, the monotonic responses of the defective and intact speci- creasing by time; ratcheting with a higher rate; and had a relatively high
mens remained relatively close to each other (see Fig. 5a). Under mono- compressive mean value that increased by time. The strains in the intact
tonic loading, the strains for the onset of wrinkling in the intact and area had lower amplitude which remained almost constant with time;
defective specimens were around 0.5% different. The difference for the showed a very low ratcheting rate; and had a low mean strain that
stress limits was just around 3% (Fig. 5a). Under cyclic loading the num- was slightly reducing by time (moving slightly from compression to-
ber of cycles to failure in the intact specimen was around 1140 cycles wards zero).
but the corroded specimen failed after only 4 cycles (Fig. 11a). The Results obtained from specimens CC2(I) to CC4(I) can be used for a
mean ratcheting rate with the defective specimen was 0.9%/cycle judgement about effects from the initial non-linear axial strain on the
(Table 5), which is around 260 times higher than that with the intact ratcheting response. As indicated in Table 5, the three specimens had
specimen (0.0035%/cycle). similar stress amplitudes. As the initial axial strain with the CC2(I),
In Fig. 11b the axial ratcheting strains in the intact specimen CC3(I) and CC4(I) increased from 0.2% to 0.3% and 0.4%, respectively
IC3(I) and the defective specimen CC3(I) are plotted against the load the mean ratcheting strain rates increased from 0.0021% to 0.0077%
cycle numbers. These two specimens were of an identical steel grade and 0.0173. Fig. 13a also shows that a relatively small increase in the ini-
and experienced similar cyclic stress histories and conditions. Fig. 11b tial strain in a defective specimen caused the ratcheting rate to be al-
indicates that the ratcheting strain rates in the defective specimen most doubled. It means that the defective specimens were more
were significantly amplified, in comparison to those in the intact speci- responsive to the initial non-linear strain and to the changes in the cy-
men. The ratcheting strain rate was around 0.008% for the defective 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
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) calculate 3D coordinates for every pixel of the cameras. Details of the
axial ratcheting strains against the load cycle numbers during the cyclic optical 3D surface acquisition and reconstruction programme and its re-
tests on the defective specimen CC3(I). Local strains were obtained from sults will be reported in a separate paper and here a summary is
strain gauges placed on the centreline of the damaged part at the spec- presented.
imen mid-height. Global (overall) strains were obtained by dividing the The measurement strategy of this type of optical surface contour
axial shortening in the specimen to its overall length. The figure, once sensing systems is triangulation between two image rays. This was
again, highlights that the imperfect region in a defective tube ratchets achieved by capturing two images with a distance from the object
very differently from those for the global tube and the intact part of surface. For convenience of providing dense and accurate points in
the tube. the scan images, a structured light pattern is projected to the speci-
A close view of the deformed shape of a defective specimen (CC3(I)) men surface by a digital light processing (DLP) projector. So, the sur-
after 300 load cycles is given in Fig. 14a. It shows formation of localised face of the specimen obtains an appropriate texture making it
wrinkles in the defective area. Fig. 14b shows the sequences of forma- possible to trace corresponding points between the two scan images.
tion and propagation of wrinkles towards failure in defective specimens, Depending on camera resolution a cloud of up to 4 million points was
with the increase in the load cycle number. obtained for every single measurement. Lower cost, higher speed
Table 5 also reports the results from cyclic tests on defective speci- and higher density of the scan points are the main advantages of op-
mens from X56 steel material. It can be seen that defective specimens tical scanners in comparison to touch based CMMs. Maximum accu-
from this steel grade were also very responsive to the changes made racy of the measured points by this system was about 5 μm but for
in the cyclic load conditions. An increase of around 5.6% in the ampli- this experimental study the specimens' surface was reconstructed
tude of cyclic load in specimen CC5(I) relative to that in specimen with 30 micron accuracy.
CC6(II) reduced the number of cycles to failure to almost half. The 3D surface acquisition and reconstruction system provided high
accuracy digitised measurements on the “As Is” geometry of the speci-
5. Non-contact surface acquisition and reconstruction mens. Such a detailed 3D model made it possible to detect the slightest
geometrical imperfections in the fabricated specimens before carrying
Non-contact 3D optical digitising system was used in the current out the tests. The optical digitising system was also used to trace the
study for the 3D surface acquisition and reconstruction of the speci- specimens; deviations from its initial geometry during and after carry-
mens' geometry and for evaluating their wrinkling and ratcheting re- ing out the tests. Pai and Young [20] employed a similar system, based
sponses. This is a camera-based triangulation system. Combined with on the scanning laser vibrometers, for boundary effect detection (BED)
digital photogrammetry techniques, a processing unit employs optical aiming at pinpointing locations of small damages in beams using oper-
transform equations to automatically and with a great accuracy ational deflection shapes (ODS) method.
2.0 2.5
1.8 Global ratchet strain
Axial Ratcheting Strain (%)
Axial Ratcheting Strain (%)
1.6 2
Local ratchet strain
1.4
1.5
1.2
CC2(I)
1.0
1
0.8
0.6 0.5
0.4
0.2 0
0 50 100 150 200 0 50 100 150 200 250 300
Number of Cycles N
Number of Cycles N
a) Effects from the initial strain on the ratchet strains in two b) Defective specimen (CC3(I)): Global (overall) ratchet
defective specimens strains and local ratchet strains (in the damaged part)
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).
M. Zeinoddini et al. / Journal of Constructional Steel Research 113 (2015) 234–246 243
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 gives two different views of the digitised post buckling surfaces of the
intact specimen IC1(I). This specimen was subjected to cyclic axial
Fig. 15 presents some typical 3D surface acquisition and reconstruc- loads up to the failure point. Wrinkling formation in the defective spec-
tion results for a number of intact and defective specimens at different imen CC2(І) was optically reconstructed and shown in Fig. 15d. This
stages of loading. The figure shows the shaded point cloud fitted to specimen was subjected to uniaxial cycling and the figure gives the
the surface of the deformed specimens. specimen status after 300 load cycles.
Fig. 15a shows the surface reconstruction results for the intact spec- Fig. 15 indicates that with the optical surface contour sensing, accu-
imen IM1(І), which was subjected to monotonic axial loads. The wrin- rate 3D representation of the specimen (up to 30 μm) can efficiently and
kling patterns in this specimen can be discerned and as will be shown effectively be generated. This makes utilising traditional contact based
later they were quantitatively measured and described. Fig. 15b and c measuring methods both cumbersome and time consuming.
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
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 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-
Fig. 16a gives a longitudinal section of the specimen IM1(І) obtained aged area. Fig. 17c indicates that, in addition to the wrinkling, the
from the optical system. This intact specimen was tested under mono- specimen experienced a bending deformation towards the damaged
tonic axial loads. Three virtually axisymmetric wrinkles can be part. Fig. 17d gives the longitudinal profile at a 90° interval. These two
recognised in the middle (main) part of the specimen. As an output of figures, once again, point that the local wrinkling and ratcheting re-
the 3D surface acquisition and reconstruction, the longitudinal profiles sponses in the damaged and perfect parts of a specimen are significantly
can be easily extracted from these data. Fig. 16b shows the radial dis- different.
placement in the specimen wall along its middle length.
Fig. 17a gives a longitudinal section of the deformed shape for the 5.3. Circumferential profiling
specimen CM1(І), obtained from the optical system. This was a defec-
tive specimen examined under a uniaxial monotonic load regime. Fig. 18 gives the deformed shapes of the specimen CM1(І) obtained
Non-axisymmetric wrinkles can be identified in this figure. As it was de- from the optical system. This was a defective specimen examined under
scribed in Section 4.2, this non-axisymmetric nature of the wrinkles is a a monotonic load regime. The circumferential profiles for the radial dis-
characteristic of the plastic buckling and ratcheting in defective speci- placements in the specimen's wall, at three different elevations along its
mens. The surface mesh outputs for the middle part of the specimen 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) 234–246 245
SEC 1-1
SEC 2-2 SEC 3-3
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- entire circumference. Wrinkles in defective specimens, under both
axisymmetric. They also show that, in addition to the wrinkling and monotonic and cyclic loadings, in contrast to the intact specimens,
longitudinal curvature, defective specimens were experiencing were non-axisymmetric. Defective specimens also experienced some
ovalisations in their cross sections. degrees of overall bending and ovalisation in their cross sections. As
the axial strain grew higher, the ovalisation tended to be amplified
6. Concluding remarks and extended to the thicker tapered parts of the specimen.
• An optical digitising system was used for 3D surface acquisition and
In the present paper, wrinkling and strain ratcheting response of the reconstruction of the tested specimen and to evaluate their strain
intact and defective cylindrical steel tubes, under monotonic and cyclic ratcheting and wrinkling response. Outputs of the optical digitising
uniaxial loads, were experimentally investigated. This type of damage, system were well able to demonstrate the growth and eventual
as an example, might be caused by repetition of start-up/shutdown localisation of wrinkles in intact and corroded specimens. Vertical,
and temperature cycles in an offshore pipeline. Small scale steel cylin- longitudinal and circumferential cross sections obtained from the op-
drical specimens with and without an artificial rectangular corrosion tical digitising system provided useful information on the extent and
were examined under both monotonic and cyclic uniaxial loads. Differ- configuration of the wrinkling and ovalisations in the corroded speci-
ent material grades, loading configurations and rectangular corrosion mens.
sizes were considered. It was noticed that: • With the defective tubes the strain ratcheting behaviour in the defec-
tive zone was distinctively different from that in the perfect zones of
• Under monotonic axial compression, the stress limit for the defective the circumference. The ratcheting strains and strain rates in the defec-
specimens was slightly (few percentages) below the corresponding tive zone of the specimen were considerably higher than the corre-
value for the intact specimens. The onset of wrinkling was around sponding values in the perfect zones. The ratcheting strains in the
20% lower than that with the intact specimen. It was concluded that defective area very rapidly turned exponential, while the ratcheting
in general, the overall stress–strain response of the specimens under strains in the perfect zone still remained in the linear trajectory.
monotonic loading was not seriously affected by the presence of • The cyclic response of the specimens was noticed to be apparently
those rectangular corrosion defects considered in the current study. much more sensitive to the presence of corrosion than that under
The monotonic response was noticed to be more severely affected monotonic loading.
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- References
ed by the presence of the corrosion defect. In contrast, as mentioned [1] R. Jiao, S. Kyriakides, Wrinkling of tubes by axial cycling, J. Appl. Mech. 77 (3) (2010)
above, the overall monotonic strain–stress responses of the defective 031012.
and intact specimens remained relatively close. [2] R. Jiao, S. Kyriakides, Ratcheting, wrinkling and collapse of tubes under axial cycling,
Int. J. Solids Struct. 46 (14) (2009) 2856–2870.
• In the defective tubes the local wrinkling first initiated from the dam- [3] R. Jiao, S. Kyriakides, Ratcheting and wrinkling of tubes due to axial cycling under
aged part. This local plastic buckling then gradually proceeded to the 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 [12] N. Sakakibara, S. Kyriakides, E. Corona, Collapse of partially corroded or worn pipe
internal pressure: part II analysis, Int. J. Solids Struct. 48 (20) (2011) 2827–2836. under external pressure, Int. J. Mech. Sci. 50 (12) (2008) 1586–1597.
[5] M. Zeinoddini, M. Peykanu, Strain ratcheting of steel tubulars with a rectangular de- [13] T.A. Netto, U.S. Ferraz, A. Botto, On the effect of corrosion defects on the collapse
fect under axial cycling: a numerical modeling, J. Constr. Steel Res. 67 (12) (2011) pressure of pipelines, Int. J. Solids Struct. 44 (22) (2007) 7597–7614.
1872–1883. [14] J. Xue, A non-linear finite-element analysis of buckle propagation in subsea corrod-
[6] S.C. Kulkarni, Y.M. Desai, T.A.R.U.N. Kant, G.R. Reddy, Y. Parulekar, K.K. Vaze, Uniaxial ed pipelines, Finite Elem. Anal. Des. 42 (14) (2006) 1211–1219.
and biaxial ratchetting study of SA333 Gr. 6 steel at room temperature, Int. J. Press. [15] J. Xue, M.S. Hoo Fatt, Symmetric and anti-symmetric buckle propagation modes in
Vessel. Pip. 80 (3) (2003) 179–185. subsea corroded pipelines, Mar. Struct. 18 (1) (2005) 43–61.
[7] G.Z. Kang, Y.G. Li, J. Zhang, Y.F. Sun, Q. Gao, Uniaxial ratcheting and failure behaviors [16] J.A. Paquette, S. Kyriakides, Plastic buckling of tubes under axial compression and in-
of two steels, Theor. Appl. Fract. Mech. 43 (2) (2005) 199–209. ternal pressure, Int. J. Mech. Sci. 48 (8) (2006) 855–867.
[8] S.K. Paul, S. Sivaprasad, S. Dhar, S. Tarafder, True stress control asymmetric cyclic [17] DNV Offshore Standard, Corroded Pipeline, Det Norske Veritas, 2011. (DNV-RP-F101).
plastic behavior in SA333 C–Mn steel, Int. J. Press. Vessel. Pip. 87 (8) (2010) [18] J.F. Hallai, S. Kyriakides, On the effect of Lüders bands on the bending of steel tubes.
440–446. Part I: experiments, Int. J. Solids Struct. 48 (24) (2011) 3275–3284.
[9] M. Zeinoddini, M. Ezzati, J. Fakheri, Uniaxial strain ratcheting behavior of dented [19] J.F. Hallai, S. Kyriakides, Underlying material response for Lüders-like instabilities,
steel tubular: an experimental study, Eng. Fail. Anal. 44 (2014) 202–216. Int. J. Plast. 47 (2013) 1–12.
[10] C. Gupta, J.K. Chakravartty, G.R. Reddy, S. Banerjee, Uniaxial cyclic deformation be- [20] P.F. Pai, L.G. Young, Damage detection of beams using operational deflection shapes,
haviour of SA 333 Gr 6 piping steel at room temperature, Int. J. Press. Vessel. Pip. Int. J. Solids Struct. 38 (18) (2001) 3161–3192.
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.