You are on page 1of 9

Journal of Testing and Evaluation, Vol. 35, No.

1
Paper ID JTE100090
Available online at: www.astm.org

Sreekanta Das, Ph.D.,1 J. J. Roger Cheng, Ph.D.,2 and David W. Murray, Ph.D.3

A Test Method to Determine Low-Cycle-Fatigue


Behavior of Wrinkled Pipe

ABSTRACT: Field observations of buried pipelines used by energy industries for transporting natural gas and oil indicate that it is not uncommon
for geotechnical movements to impose large displacements on buried pipelines resulting in localized curvature, deformations, and strain in the pipe
wall. Often these local deformation results in local buckling in the pipe wall 共wrinkling兲 and, in its post-buckling range of response, wrinkles develop
rapidly. Subsequent cyclic load histories may produce cyclic plastic strain reversals in the wrinkle region leading to formation of fractures in the
wrinkle region. This paper presents an innovative and simple material test method, called a strip test, which was designed and carried out at the
University of Alberta in order to simulate the complicated behavior of pipe wrinkles subject to such low-cycle-fatigue loading. It is found that the
strip test is capable of replicating the complicated behavior of wrinkled pipe subject to plastic strain reversals at the wrinkle location due to low-
cycle-fatigue loading and provides necessary information that can be used for further studies. For the current project, a total of 16 such strip tests
were carried out, and the test data from these strip tests have been used successfully to develop a fracture life assessment 共FLA兲 model for the
wrinkled energy pipe subject to strain reversal due to low-cycle-fatigue loading. The development of the FLA model will be presented in a future
publication.
KEYWORDS: test method, wrinkled pipeline, low cycle fatigue, plastic strain reversals, fracture

Introduction The motivation for this investigation was to determine the limit
fracture strains, failure modes within wrinkles, and assessment of
Steel pipelines are being used by the energy industries in North remaining fracture life of a wrinkled pipe arose from the diagnosis
America as the primary mode for transporting natural gas, crude and exposure of a wrinkle in the Norman Wells Pipeline, located in
oil, and petroleum products. The majority of these pipelines run the MacKenzie valley of the Northwest Territories of Canada, op-
under the ground. Field observations of buried energy pipelines in- erated by Enbridge Pipeline 共NW兲 Inc. The case history of this
dicate that it is not uncommon for geotechnical movements to im- wrinkle is well documented by 共Wilkie et al. 2000; Oswell et al.
pose large displacements on buried pipelines, resulting in localized 2000; Yoosef-Ghodsi et al. 2000兲. The limiting fracture strains and
curvatures and the associated deformations and strains in the pipe failure modes of wrinkled pipelines are presented elsewhere 共Das
wall 共Yoosef-Ghodsi et al. 1995; Murray 1997; Bai et al. 2000; Jay- et al. 2002, and Das et al. 2000兲.
adevan et al. 2004兲. Such displacements may be associated with
river crossings, unstable slopes, or regions of discontinuous perma-
frost. Often the local deformation of the pipe wall results in local
buckling 共wrinkling兲 and, in its post-buckling range of responses, Fracture under Strain Reversals
local buckles in the pipe wall 共wrinkles兲 develop rapidly and can be A total of twelve full-scale tests on wrinkled pipes were carried out
of significant magnitude. at the Structure Laboratory of the University of Alberta to study
The wrinkling in the field pipeline normally occurs under com- limit fracture strains and fracture modes with wrinkles in wrinkled
binations of internal pressure, monotonic axial and/or bending de- pipe of the NPS12 共i.e., nominal pipe diameter of 12 in.兲 Norman
formations. Subsequent cyclic deformation histories then produce Wells Pipeline, operated by Enbridge Pipelines 共NW兲 Inc. It was
cyclic plastic strain reversals in the wrinkle region leading to dev- found that a monotonic axial stroke and internal pressure produces
astating consequences because fracture forms quickly in the an accordion-type failure, but it could not produce a fracture 共Das
wrinkle region 共Bouwkamp and Stephen 1973; Bai et al. 2000兲. Cy- et al. 2000兲. A cyclic deformation history that produces a history of
clic temperature variations, cyclic freeze-thaw of ground, seismic plastic strain reversals at the wrinkle location was required to pro-
loads, etc., are a few examples of typical sources of cyclic deforma- duce a fracture in the wrinkled pipe 共Das et al. 2002兲. Conse-
tion histories for buried pipelines. Thus, the objective of this study quently, majority of the full-scale tests were conducted under cyclic
is to develop a simple test procedure that is able to simulate the strokes 共deformations兲 after formation of the wrinkle. The wrinkles
complicated plastic strain reversals at the wrinkle region of field in the laboratory test specimens were developed by applying mono-
pipelines. tonic axial deformation and constant internal pressure the way
wrinkle forms in the field pipelines. A detailed discussion on such
Manuscript received October 19, 2004; accepted for publication April 17, test procedure and test results can be found elsewhere 共Das et al.
2006; published online June 2006.
1 2002; Das et al. 2000兲.
Assistant Professor, Department of Civil and Environmental Engineering,
University of Windsor, Windsor, ON N9B 3P4, Canada.
Figure 1 shows a 12-in. nominal diameter wrinkled pipe speci-
2
Professor, University of Alberta, Department of Civil & Environmental En- men that fractured in the laboratory due to strain reversals at the
gineering, Edmonton, AB T6G 2G7, Canada. wrinkle location and a fractured wrinkle pipe specimen of 8-in.
3
Professor Emeritus, University of Alberta, Department of Civil & Environ- nominal diameter that was obtained from the field. Application of
mental Engineering, Edmonton, AB T6G 2G7, Canada. monotonic axial load and stroke with constant internal pressure

Copyright
Copyright by ASTM
© 2006 Int'l (all
by ASTM rights reserved);
International, Mon Harbor
100 Barr Sep 19 11:43:26
Drive, POEDT
Box2016
C700, West Conshohocken, PA 19428-2959. 1
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
2 JOURNAL OF TESTING AND EVALUATION

FIG. 2—Schematic of a strip.

The strip was then bent monotonically at mid-length in the lon-


FIG. 1—Comparison of fractured specimens obtained from field and laboratory gitudinal direction to simulate the formation of outward bend
test.
共wrinkle兲 that forms in the field pipeline under constant internal
pressure and monotonic load/deformation. The bend angle for
produced wrinkle but did not produce a fracture. Subsequent appli- every specimen was set to 45°. The final bend angles obtained are
cation of cyclic tension and compression strokes on the pipe wall shown in Table 1. Internal radius of bend was 15 mm or 20 mm, as
produced plastic strain reversals at the wrinkle region resulting in shown in Table 1. A typical bent strip is shown in Fig. 3. A sche-
fracture. It is believed that the wrinkle in the field pipeline devel- matic of such a bent strip is shown in Fig. 4. The strips that were
oped due to geotechnical movements. Subsequently, the wrinkled bent with 15-mm internal radius 共R兲 are referred to as R15 speci-
location of the pipeline was subjected to cyclic deformation history mens and the strips bent with 20-mm radius are referred to as R20
due to cyclic thermal loads. This produced similar plastic strain re- specimens. The extreme end of the curved portion of the strip is
versals and subsequent fracture in the wrinkle region 共see Fig. 1兲. called the “crest” of the bent strip and it resembles the “crest” of a
pipe wrinkle. Straight portions of the strip specimen are called its
legs and there are two such legs.
Two custom-made loading mounts were then welded to the
Strip Test straight arms 共i.e., legs兲 of the bent strip, as shown in Fig. 5; this is
called a “strip specimen.” The loading mounts were used to secure
each specimen into a servo-hydraulic universal testing machine,
Background
which is referred to as “SHUT machine” in this paper 共see Fig. 5兲.
Full-scale tests on pipe specimens showed that a cyclic deformation A total of 16 strip specimens were made and tested under cyclic
and/or cyclic load history produces fracture in the wrinkle region strokes using a stroke 共displacement兲 controlled method. As a re-
due to strain reversals in the plastic strain. A simple fracture life sult, the loads applied during these tests were also cyclic. Thus, the
assessment 共FLA兲 model that is able to estimate the remaining ser- application of cyclic stroke hysteresis also produced a hysteresis of
vice life of a wrinkled field pipeline is in high demand. Test data plastic strain reversals. Finally, fracture at the crest of strip speci-
from a wide range of full-scale tests would have been necessary to men occurs the way a wrinkled field pipe fails in fracture due to
have sufficient database for developing such a FLA model. Unfor- plastic strain reversals. Discussions on the results are presented in
tunately, this kind of full-scale test is very expensive and time con- subsequent sections.
suming. Consequently, a simple material testing procedure, called a
“strip test,” which is easy to carry out and provides necessary infor-
mation required to develop a FLA model was designed and devel- Tests Setup
oped at the University of Alberta. The following subsections de- The SHUT test machine was used to pry the bent strip open and to
scribe the design of strip specimen and strip test procedure. close it. Two loading mounts attached to the bent strip were gripped
by the SHUT machine with its hydraulic grips. As can be seen in
Fig. 5, each mount had a hinge that allowed the leg of the specimen
Strip Specimen
to rotate as the loading head of the SHUT machine was moved up or
Longitudinal strips 共57-mm wide and about 535-mm long兲 were cut down. This ensured that the specimen was strained only at the crest
from the pipe specimens. It is believed that the effect of any re- of the test strip 共see Figs. 3 and 4兲 but not at other locations. Next, a
sidual stress in extreme low-cycle-fatigue 共LCF兲 behavior would be cyclic stroke hysteresis was applied by the SHUT machine and as a
negligible since the strokes and associated strains applied in strip result, the specimen was pried open and closed; this is illustrated in
specimens are very large. Nevertheless, great care was taken in sec- Figs. 6 and 7, respectively. This produced a hysteresis of plastic
tioning the pipe using a cold-cut method to cut the strips out of the strain reversals at the crest of strip specimen and subsequent frac-
pipe specimens in order to avoid any additional residual stress gen- ture.
eration within the strips. A schematic of such a strip is shown in Instrumentation for strip tests was kept to a minimum. A cable
Fig. 2. The width of specimen was limited to 57 mm so that the transducer was attached to the crest of the wrinkle to record the
curvature of the pipe does not have much effect on the bent geom- eccentricity during the test. The distance eo in Fig. 5 represents the
etry of the strip. A length of 535 mm was chosen such that it is just initial eccentricity; that is, the eccentricity before application of any
enough to produce a shape that simulates a pipe wrinkle, and to cyclic stokes. Eccentricity 共ea兲 at any time is the horizontal distance
provide enough room for attachment of necessary loading mounts between center of the crest of strip specimen and the SHUT ma-
on to the strip at a desirable eccentricity with respect to the crest of chine load line. The eccentricity reduced gradually to et as the
bent strip. specimen is pried open and the same is increased to ec 共where et

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
DAS ET AL. ON LOW-CYCLE-FATIGUE BEHAVIOR 3

TABLE 1—Specimen matrix for strip tests.

Stroke 共mm兲

Bend Maximum Nominal Bend Nominal


Specimen Thickness Radius Change Range Width Angle Length
Series Name 共mm兲 共mm兲 共S兲 共±Sr兲 共mm兲 共degree兲 共mm兲
R15t6 R15t6S50 6 15 50 25 57 44 535
R15t6S60 6 15 60 30 57 41 535
R15t6S70 6 15 70 35 57 41 535
R15t6S80 6 15 80 40 57 45 535

R20t6 R20t6S50 6 20 50 25 57 45 535


R20t6S60 6 20 60 30 57 45 535
R20t6S70 6 20 70 35 57 42 535
R20t6S80 6 20 80 40 57 45 535

R15t9 R15t9S50 8.3 15 50 25 57 45 535


R15t9S60 8.3 15 60 30 57 45 535
R15t9S70 8.3 15 70 35 57 45 535
R15t9S80 8.3 15 80 40 57 45 535

R20t9 R20t9S50 8.3 20 50 25 57 45 535


R20t9S60 8.3 20 60 30 57 45 535
R20t9S70 8.3 20 70 35 57 45 535
R20t9S80 8.3 20 80 40 57 45 535

艋 ea 艋 ec兲 as the specimen is pried closed. This can be observed sistance strain gauge at the crest would not perform well for this
from Figs. 6 and 7. In these figures, a clip gauge 共CG兲 and a cable kind of cyclic deformation history and thus, strain gauges were not
transducer’s wire are also visible. Measuring the eccentricity and installed on the crest of the current specimens. However, a clip
load applied throughout the test allowed one to determine the mo- gauge of 15 mm gauge length that worked well with the previous
ment applied at the crest of the specimen during the test. tests was installed to measure the strain history on the outer face of
Experience from previous tests indicated that an electronic re-
the crest of the strip specimen. Axial load and stroke were recorded

FIG. 3—A bent strip.

FIG. 4—Schematic of a bent strip. FIG. 5—Strip specimen mounted on SHUT machine.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
4 JOURNAL OF TESTING AND EVALUATION

diameter in the center and a bigger wheel that rotates around the
rod. The straight strip was placed in between the center rod and the
wheel. The desired angle of bend in the strip specimen was
achieved by rotating the wheel around the center rod. Conse-
quently, the radius of the center rod became the inner internal ra-
dius 共R兲 of the strip specimens. The center rod is replaceable and
center rods of two different diameters 共30 and 40 mm兲 were used to
achieve two different internal bend radii 共15 and 20 mm兲 of strip
specimens.
Tiny punch marks, as shown in Fig. 8, were made on both sur-
faces on the straight specimens in order to determine the strain in-
troduced at the crest during this forming stage. These marks were
made approximately 5 mm apart and were located on the part of the
FIG. 6—Strip specimen subjected to maximum tension. specimen that would become the crest of the test strip specimen.
Two parallel rows of seven punches along the length of the speci-
men and 5 mm off the center line were made on its both surfaces.
through the SHUT machine controls. An automatic data acquisition Each row, therefore, had six gauges of 5-mm initial gauge length.
system was used to record the data automatically at a desired fre- The punch marks were used for measuring strains on 15-mm
quency. gauge lengths at the crest of the specimen on its both surfaces,
which were induced during the monotonic bending stage. Conse-
Test Procedure quently, only three gauges of each row nearest to the crest of the
specimens were finally considered to compute these strains. The
Each test consisted of two stages: monotonic bending to form the average value of strains obtained from two rows on the same sur-
bend 共wrinkle兲 followed by application of cyclic displacements. In face was taken as the final strain measurement for that surface. The
the monotonic bending stage 共that is the first stage兲, the straight gauge lengths were carefully measured by electronic caliper before
strip was bent to an approximate internal angle of 45° as discussed the specimen was bent and the measurements were recorded for use
in the previous section and shown in Figs. 3 and 4. A machine de- as initial gauge lengths.
signed to bend steel reinforcement bars for reinforced concrete was After the specimen was bent, a strip of transparent adhesive tape
used for this job. This type of machine has a steel rod of smaller was placed on both surfaces of the bent portion of the strip. The
locations of the punches were then marked on the adhesive tape.
The adhesive tapes were then peeled off the surfaces of the strip
specimen and placed on a rigid flat surface. Then linear measure-
ments were taken for gauges between two consecutive punch
marks. These measurements were then used as final gauge lengths.
The initial and final gauge lengths were then used to determine the
strains on extreme tension and compression fibres of the crest of the
specimen. The CG that was used to record the strain hysteresis dur-
ing application of strain reversal cyclic strokes, had a gauge length
of 15 mm. Consequently, these strains were calculated for 15-mm
gauge length, considering three consecutive gauge lengths nearest
to the crest of the bend. Strain values are discussed in the subse-
quent sections.
A total of sixteen strips from two different pipe specimens were
made. These pipes had the wall thickness of 6.0 mm 共Series t6兲 and
8.3 mm 共Series t9兲 with D / t ratios of 85 and 62, respectively. Each
series had eight strips and four of them were bent with an internal
radius of 15 mm 共R15兲 and remaining four were bent with internal
radius of 20 mm 共R20兲.
FIG. 7—Strip specimen subjected to maximum compression. The test specimen matrix is shown in Table 1. Specimen

FIG. 8—Detail of punch marks at mid-length of straight specimen.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
DAS ET AL. ON LOW-CYCLE-FATIGUE BEHAVIOR 5

R15t9S80 indicates that this specimen’s wall thickness was 8.3 mm TABLE 2—Material Properties of strip specimens.
共that is, t9 specimens兲 and the specimen was bent with an internal
Specimens With Specimens With
radius of 15 mm 共R15兲 and maximum stroke change of 80 mm
6 mm Wall 8.3 mm Wall
共S80兲 was applied to this specimen during application of cyclic Property Thickness Thickness
strokes. The curved portion of the strip specimen was intended to Modulus of Elasticity 共GPa兲 202 211
simulate the wrinkle in the compression wall of full-scale pipe Proportional Limit 314 378
specimens. The purpose of using two different radii of curvature Static Yield Stress 共MPa兲 460 479
was to simulate two different internal pressures of full-scale pipe Static Ultimate Stress 共MPa兲 563 546
specimens. Strip specimens with higher internal radius or lower
curvature correspond to a pipe wrinkle with higher internal pres-
sure. observed that they look alike. The photograph on the left in Fig. 9
In the second stage of testing, the loading mounts were spot- shows a wrinkle that developed just before the end of a full-scale
welded to the legs of the strip specimen. Then, the specimen was pipe test on a pipe specimen with D / t ratio of 62 and wall thickness
mounted on to the SHUT machine as shown in Fig. 5. The specimen of 8.30 mm and subject to medium internal pressure of 0.4py. The
was mounted in such a way that the initial eccentricity 共eo兲 mea- py is the required internal pressure to cause yield stress of pipe ma-
sured from center of crest to the line of action of the axial load was terial to be developed in the circumferential direction. For the cur-
approximately 70 mm, as shown in Fig. 5. True value of the eccen- rent work, the high, medium, and low internal pressures indicate
tricity was recorded after mounting the specimens on to the SHUT 0.8py, 0.4py, and 0.0py, respectively. The photograph on the right of
machine and before applications of any cyclic stroke. True mea- Fig. 9 shows the configuration at the same stage but for the strip
sured values for initial eccentricity were within the range of specimen R15t9S70, made out of same pipe specimen. In both
±2 % of 70 mm. The initial eccentricity of 70 mm was chosen cases, a dimple appears at the crest of the wrinkle and fracture was
based on two considerations. First, it had to be small enough to initiated from the inside surface of the wrinkle. The crack in the
prevent bending of the specimen away from its crest. Second, it had pipe specimen is not visible because it is on the inside pipe wall.
to be big enough so that the value of eccentricity would not be ap- The overall shape of the pipe wrinkle is very similar to the overall
proaching a very small or zero value when the specimen was pried shape of the strip specimen. The principal difference between a
open for the maximum stroke range 共±40 mm兲 that was to be ap- pipe wrinkle and a strip specimen is that there were two plastic
plied on these specimens. The ±Sr in Table 1 represents the stroke hinges formed at the feet of the pipe wrinkle during the process of
range for a single stroke cycle. The positive sign with it indicates loading, but for the strip specimen, two mechanical hinges were
tension and negative sign indicates compression on strip specimen. provided instead. However, this difference is considered not to be
The spot-weld between loading mount and the leg of the specimen an issue for this study because the objective of the strip test is to
would have been broken if the eccentricity were reduced to a very simulate and understand the plastic strain reversal behavior of the
small value when the specimen was pried open to its maximum crest of a pipe wrinkle. The exterior dimple and the internal crack at
range. Cyclic strokes that introduced plastic strain reversals at the the crest of the wrinkle are shown in Fig. 9 for both the pipe and the
crest were then applied by the SHUT machine until the specimen strip, and exhibit great similarity. They lend credence to the concept
failed in fracture at the crest the way a wrinkled pipe fails in the that the results from strip testing can provide useful information of
field. crack characteristics for wrinkled pipes.
As can be seen in Table 1, each series had two groups and each
group had four specimens. The first group had internal radius of
15 mm 共that is, R15兲 and the other group had 20 mm of internal Load Hysteresis
radius 共that is, R20兲. The four specimens in each group had four
Figure 10 shows a typical load-stroke hysteresis curve obtained
different stroke ranges: ±40, ±35, ±30, and ±25 mm. The maxi-
from strip Specimen R20t9S50. Cyclic stroke hysteresis was initi-
mum stroke change is the difference between maximum and mini-
ated by applying a tensile stroke from the SHUT machine after the
mum strokes that were applied to the specimen. Consequently, the
specimen was mounted in the SHUT machine and the initial value
maximum stroke change value in a one load cycle 共S兲, that were
of eccentricity 共eo兲 was recorded 共see Fig. 5兲. This tensile stroke
chosen are: 80, 70, 60, and 50 mm, respectively. The stroke range
opens up the specimen and increases the internal bend angle, caus-
magnitude on the compression and tension sides were kept the
same and consequently, the maximum stroke change in one load
cycle, applied to the specimen 共S兲 is the summation of the absolute
values of the maximum and minimum of the stroke range, that is
2兩Sr兩.

Test Results
The strain and load hysteresis loops obtained from these tests are
discussed in this section. Qualitative comparisons between the be-
havior of pipe wrinkles and strip wrinkles are also discussed in this
section. Material properties for the strips are presented in Table 2.

Wrinkle Shape and Fracture


Qualitative comparison between the final shape of a pipe wrinkle
with the final shape of a strip specimen is shown in Fig. 9 and it is FIG. 9—Comparison of pipe wrinkle with strip wrinkle.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
6 JOURNAL OF TESTING AND EVALUATION

FIG. 10—A typical load-stroke hysteresis for a strip specimen. FIG. 12—Single load-stroke hysteresis cycle for a strip specimen.

ing compressive strain on the outside surface of the strip crest 共see
eccentricity 共eo兲 of Fig. 5 changed as the stroke varied with time.
the sketch on the left of Fig. 11兲. The stroke of the machine contin-
How the stroke changed can be seen in Fig. 13. The key points are
ued to be increased until it reached a predetermined maximum level
those marked in Fig. 12 that are also shown in Fig. 13. The eccen-
of tensile stroke 共indicated by +ve abscissa in Fig. 10兲. For ex-
tricity reduced to its minimum value 共et of Fig. 6兲 at Point A and
ample, the maximum stroke level is +25 mm for this specimen.
consequently, a sharp rate of increase and decrease in load is no-
This stroke is shown by load path OA in Fig. 10. Subsequently, the
ticed around this point. The load value changes from tension to
stroke of the SHUT machine was reversed until it reached the pre-
compression at Point B and the maximum compressive load devel-
determined level of minimum stroke 共maximum compression兲 as
ops at Point C. However, the maximum compressive stroke reaches
shown by path ABCDE in Fig. 10. The minimum stroke 共indicated
at Point E, far away from Point C, and compressive load dropped by
by −ve abscissa兲 applied to this specimen was −25 mm. Conse-
a small amount between Point C and Point E 共see Fig. 12兲.
quently, the specimen closed, and the bend became much sharper
The maximum tensile stroke and maximum tensile load occur at
共see the sketch on the right of Fig. 11兲 than its initial configuration
the same point 共see, Point A in Fig. 12兲. However, the maximum
at Point O of Fig. 10. The stroke was then reversed back to its maxi-
compressive load was obtained at Point C, much before maximum
mum tension level 共Point H in Fig. 10兲, as shown by path EFGH in
compressive stroke was applied 共that is, at Point E of Fig. 12兲. The
Fig. 10.
slope of unloading from maximum tensile load 共Line AB兲 is 5.77
The path ABCDEFGH in Fig. 10 constitutes an accumulated
times stiffer than the slope of unloading from maximum compres-
history of complete stroke cycles. The process of changing stroke
sive load 共Line EF兲.
from maximum tensile value to maximum compressive value was
A typical load-stroke cycle for pipe specimen is shown in Fig.
repeated over and over again until a fracture developed at the crest,
14. This pipe specimen was also subject to similar cyclic stroke
and the maximum load capacity of the specimen dropped consider-
history. A quantitative comparison of load-stroke behaviors of Figs.
ably. All the strip tests were conducted under stroke controlled con-
12 and 14 shows a good qualitative correlation. This is an indica-
dition. The loads applied during application of cyclic strokes were
tion that the behavior of the test strip is a good representation of
also recorded. Figure 12 shows a single load-stroke cycle that was
behavior of a wrinkle in the pipe test. The ordinate 共load axis兲 in
applied to the specimen. This plot is asymmetric about both the
Fig. 14 has been shifted to the middle of maximum stroke change
load axis and the stroke axis. The primary reason for this is that the
共S兲 to make it consistent with Fig. 12. The load in Fig. 14, is the
total load applied through the SHUT machine and consequently,

FIG. 11—Schematic of opening and closing of a strip specimen. 共a兲 OPENING


共increasing bend angle兲. 共b兲 CLOSING 共decreasing bend angle兲. FIG. 13—Change in eccentricity with change in stroke in a single cycle.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
DAS ET AL. ON LOW-CYCLE-FATIGUE BEHAVIOR 7

FIG. 14—Typical load-stroke hysteresis cycle for a pipe specimen. FIG. 16—A single moment hysteresis cycle for a strip specimen.

the load magnitude is always compressive 共shown by −ve sign兲. classical fatigue hysteresis plot, but the same is not true for load
However, the true load applied to pipe wall is to be computed with response 共see Fig. 10兲. This is because moment hysteresis curve
respect to the dashed line of Fig. 14. Consequently, load values considers the change in eccentricity value during the test, whereas
above the dashed line represent a tensile load on the pipe wall and the load hysteresis curve does not consider this. It is not quite ob-
those below the dashed line are compressive load on the pipe wall. vious why the stiffness in the region AB is much higher than the
The stroke in Fig. 14 is the variation in total stroke of the pipe speci- stiffness in the region EF 共see Fig. 16兲. The bent geometry of the
men during Cycle No. 3. specimen and prior strain history of monotonic bending may be the
reasons for this kind of behavior. More attention and investigation
is necessary to find out the true reasons for this behavior.
Moment Hysteresis
The moment versus stroke hysteresis for the same strip specimen Strain Hysteresis
共R20t9S50兲 is shown in Fig. 15. A single moment-stroke cycle for
the same response is shown in Fig. 16. The key points of Fig. 12 are As mentioned in an earlier section, only a CG of 15-mm gauge
also shown in Figs. 15 and 16. Unlike the hysteresis loop in Fig. 12, length was mounted at the crest of the strip specimen to record
the maximum negative moment and maximum compressive stroke strain hysteresis. A typical strain hysteresis plot is shown in Fig. 17.
occurred at the same point; that is, at Point E. Consequently, no Only a few cycles are shown to keep this plot relatively clean. The
drop in moment capacities is noticed during compression loading. initial strain applied during monotonic bending was 22.1 %, this is
The magnitude of maximum moments in tension and compression not incorporated into this plot because the whole load path of
are also almost same and these values are 0.78 and 0.76 kN m, re- monotonic bending could not be recorded.
spectively. Like the load-stroke hysteresis curve in Fig. 12, the The salient points marked on Fig. 12 are also indicated in Fig.
slope of unloading from maximum tensile load 共Line AB兲 is stiffer 17. Point H in Fig. 17 indicates the end of first stroke cycle. Initial
than the slope of unloading from maximum compressive load 共Line tensile path OA 共see Fig. 10兲 is omitted in this figure even though it
EF兲. However, the difference between these two stiffness is much was recorded during the test. The CG measured the average strain
smaller in this response 共see Fig. 16兲 and the ratio of these two over a gauge length of 15 mm. Consequently, strain localization
slopes is 2.66 times. and cracks did not affect these strain data and CG did not capture
It is noted that the moment response 共see Fig. 15兲 is more like a the pipe material behavior at the highest strain location. Maximum

FIG. 15—A typical moment hysteresis curve for a strip specimen. FIG. 17—A part of CG strain hysteresis for a strip specimen.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
8 JOURNAL OF TESTING AND EVALUATION

TABLE 3—Maximum strains recorded from strip tests.

Maximum Outside Strains From


Tensile Strain at Outside Cyclic Loading, %
After Monotonic Bend,
Series Specimen Name % Tensile 共+ve兲 Compressive 共−ve兲
R15t6 R15t6S50 19.4 7.1 4.0
R15t6S60 20.2 4.7 4.0
R15t6S70 19.4 11.9 5.6
R15t6S80 17.3 10.2 7.8

R20t6 R20t6S50 20.1 7.3 2.0


R20t6S60 21.5 9.8 3.5
R20t6S70 22.1 12.0 4.7
R20t6S80 18.8 14.5 5.8

R15t9 R15t9S50 29.6 CG did not work properly


R15t9S60 29.2 10.8 6.4
R15t9S70 26.8 11.8 9.7
R15t9S80 26.6 11.2 10.6

R20t9 R20t9S50 25.0 10.2 4.0


CG did not work properly
R20t9S80 24.3 11.3 10.1

tensile and compressive strains recorded from strip tests are 14.50 stalled after monotonic loading was completed. The zero stroke
% and 10.60 %, respectively. This can be observed from the strains value in this figure has been adjusted such that stroke change in
recorded from the strip tests, as reported in Table 3. both tension 共indicated by +ve sign兲 and compression 共indicated by
A similar strain hysteresis plot obtained from a full-scale pipe −ve sign兲 are almost equal magnitude. This adjustment is done to
test is shown in Fig. 18, and the salient points are also marked. This achieve consistency between the stroke values in Figs. 17 and 18.
was obtained from a CG of 12 mm gauge length that was mounted The true location of ordinate 共CG strain axis兲 that was obtained
at the crest of pipe wrinkle of a pipe specimen that was tested under from the pipe test is shown by a dotted line on the right of Fig. 18.
similar cyclic strokes. All the stroke cycles for this pipe specimens The CG strain plots obtained from both strip specimen 共Fig. 17兲
were done at almost same stroke levels and with almost same stroke and pipe specimen 共Fig. 18兲 are quite similar. For pipe specimens,
range, that is, this pipe specimen was cycled almost the same way limitation in maximum internal pressure limited the maximum ten-
the strip specimens were cycled. Consequently, strain hysteresis for sile force that could be applied on the pipe wall. The test setup itself
pipe Specimen 6 was chosen to compare with the strain hysteresis could not apply tensile load on the pipe specimen. However a lim-
of a typical strip specimen. ited tensile load could be applied to the pipe wall by applying in-
This pipe specimen took only three cycles to produce a fracture. creased internal water pressure and letting the pipe specimen elon-
The CG was installed after monotonic axial load was unloaded gate. However, there was no limitation on compressive stroke and
completely 共that is, at a point which is comparable to Point D of compressive load due to the test setup. That is why the maximum
Fig. 10 of a strip specimen兲. Path OABCD for the first stroke cycle compressive strain in Fig. 18 for pipe specimen is much smaller
does not exist in pipe tests. Initial strain obtained from monotonic than its maximum tensile strain. The CG strain is compressive
load history does not appear in Fig. 18 because the CG was in- 共negative兲 when the pipe wrinkle or strip specimen is in tension. For
strip specimen, the difference between these two maximum strain
values is very small. Qualitative correlation between Figs. 17 and
18 indicates that the strain reversals at the crest of the strip speci-
men could simulate the same strain reversal that occurred at the
crest of a pipe wrinkle.

Behavioral Comparison
Many similarities in behavior between a strip specimen and a pipe
wrinkle have been noted and discussed in earlier sections. The other
similarity that was noticed is the number of cycles to fracture 共Nf兲.
It was observed that Nf was higher if the stroke range was chosen
lower and that was true for both strip specimens and pipe speci-
mens.
As discussed in earlier sections, for pipe specimens, the tensile
stroke and load on the pipe wall was applied by removing the exter-
FIG. 18—A part of CG strain hysteresis for a pipe specimen. nal loads and by applying a maximum internal pressure of 0.8py.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.
DAS ET AL. ON LOW-CYCLE-FATIGUE BEHAVIOR 9

Consequently, the maximum internal pressure 共0.8py兲 that was ap- The strip test has a high potential and can be used for further/
plied to the pipe specimens limited the maximum tensile stroke and advanced studies on LCF behavior of wrinkled pipe used in various
maximum tensile stroke on the pipe wall. The maximum compres- applications. The strip test data from the current project have been
sive load and corresponding compressive stroke of pipe wall was used successfully to develop a FLA model of wrinkled energy pipe-
related to the maximum load carrying capacity of the pipe speci- line subject to LCF load history and this will be presented in a fu-
men in compression. Consequently, stroke range values in tension ture publication.
and compression for pipe specimens were different. However for
strip specimens, choice of stroke range in both tension and com-
pression was wider and was primarily controlled by SHUT machine
stroke values and geometry of strip specimen and it was decided to References
maintain the same stroke range in tension and compression for strip
specimens. 关1兴 Bai, Y., Knauf, G., and Hillenbrand, H-G., 2000, “Materials
For pipe specimens, initiation of cyclic stroke was made at dif- and Design of High Strength Pipelines,” Proceedings of Inter-
ferent stroke levels, that is, at different wrinkle amplitudes. In fact, national Offshore and Polar Engineering Conference, ISOPE,
for pipe specimens, cyclic strokes were started when a well- Ed., ISOPE, USA.
developed wrinkle was noticed. However for strip specimens, the 关2兴 Bouwkamp, G. and Stephen, R. M., 1973, “Large Diameter
cycling always started at the same stroke level because the internal Pipe under Combined Loading,” Transp. Eng. J. ASCE, Vol
bend angle was kept constant 共approximately 45°, as shown in Fig. 99. No TE3, pp. 521–536.
4兲 for all the strip specimens. 关3兴 Das, S., Cheng, J. J. R., and Murray, D. W., 2002, “Fracture in
The other difference between a pipe wrinkle and a strip speci- Wrinkled Linepipe Under Monotonic Loading,” International
men stroke history was that usually, a pipe specimen was cycled at Pipeline Conference, OMAE, Ed., ASME, USA.
different stroke levels and consequently with different stroke 关4兴 Das, S., Cheng, J. J. R., Murray, D. W., Wilkie, S. A., and
ranges. However a strip specimen was cycled at the same level of Zhou, Z. J., 2000, “Laboratory Study of Local Buckling,
stroke and within one stroke range. Wrinkle Development, and Strains for NPS12 Linepipe,” In-
In general, it is found that the strip test is capable of simulating ternational Pipeline Conference, OMAE, Ed., ASME, USA.
the behavior of pipe wrinkle when subjected to cyclic deformations 关5兴 Jayadevan, K. R., Ostby, E., and Thaulow, C., 2004, “Strain
and loads. Consequently, strip test behavior and test data can be Based Fracture Mechanics Analysis of Pipelines,” Proceed-
used as an alternative to full-scale test of wrinkled pipe subjected to ings of International Conference on Advances in Structural
plastic strain reversals. For the current project, the strip test data Integrity, ICAS, Ed., ICSA, India.
were used successfully for the development of a fracture criterion 关6兴 Murray, D. W., 1997, “Local Buckling, Strain Localization,
for pipe wrinkles under low-cycle-fatigue load conditions. The de- Wrinkling, and Post-Buckling Response of Line Pipe,” Jour-
velopment of FLA model using the strip test data will be presented nal of Engineering Structures, Vol 19, No 5, pp. 360–371.
in a separate future publication. 关7兴 Oswell, J. M., Hanna, A. J., Doblanko, R. M., and Wilkie, S.,
2000, “Instrumentation and Geotechnical Assessment of a
Pipeline Undergoing Wrinkling,” International Pipeline Con-
Conclusions ference, OMAE, Ed., ASME, USA.
An innovative material test method, called the strip test, has been 关8兴 Wilkie, S. A., Doblanko, R. M., and Fladager, S. J., 2000,
designed and developed in this work. A good correlation between “Case History of Local Wrinkling of a Pipeline,” Interna-
strip test data and full-scale pipe tests data has been obtained. Con- tional Pipeline Conference, OMAE, Ed., ASME, USA.
sequently, it is concluded that the strip test is capable of simulating 关9兴 Yoosef-Ghodsi, N., Kulak, G. L., and Murray, D. W., 1995,
the behavior of pipe wrinkle if it is subject to similar plastic strain “Some Test Results for Wrinkling of Girth-Welded Line
reversal due to LCF load history. Pipe,” Proceedings of OMAE Conference, OMAE, Ed.,
A strip test is easy to carry out and provide required information ASME, USA.
on strain reversal behavior at the wrinkle location if the wrinkled 关10兴 Yoosef-Ghodsi, N., Murray, D. W., Cheng, J. J. R., Doblanko,
pipe is subject to a LCF load history. Consequently, it is concluded R. M., and Wilkie, S. A., 2000, “Analytical Simulation of
that the strip test can be considered as an easy alternative simula- Field measurement of a Wrinkle on the Norman Wells Pipe-
tion for complicated and expensive full-scale pipe test for deter- line,” International Pipeline Conference, OMAE, Ed.,
mining its LCF behavior. ASME, USA.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 19 11:43:26 EDT 2016
Downloaded/printed by
University of British Columbia Library (University of British Columbia Library) pursuant to License Agreement. No further reproductions authorized.