Journal of Constructional Steel Research 121 (2016) 237–252

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Influence of ring-stiffeners on buckling behavior of pipelines under

hydrostatic pressure
R. Shahandeh ⁎, H. Showkati
Dept. of Civil Engineering, Urmia University, Urmia 15311-57561, Iran

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

Article history: Submarine pipelines are important and influential structures in marine engineering because they transport im-
Received 15 May 2015 portant, useful, and common fuels around the world. The pipeline structure is affected by different environmental
Received in revised form 27 January 2016 forces from the surrounding conditions such as the external uniform hydrostatic pressure, hydrodynamic pres-
Accepted 11 February 2016
sure, seismic forces, installation forces, free span forces, and forces scouring the pipeline. These conditions
Available online 26 February 2016
cause the pipeline to express complicated behavior.
Keywords:
Local buckling and buckling propagation along the pipeline are common examples of collapses under these con-
Pipeline ditions and loadings, which can destroy thousands of meters of pipeline. These post-buckling phenomena cause
Buckling great damage and losses to the oil and gas industries and to the environment; the imposed costs for the repair
Post-buckling and protection of pipelines and mitigation of environmental problems are high. The use of ring-stiffeners is
Collapse one way of increasing the buckling capacity of a pipeline in order to prevent and control the buckling propagation
Buckling propagation in pipelines.
Tilting In this study, the buckling and post-buckling behaviors of ring-stiffened pipelines were investigated at a small
scale through experiments and the finite element method (FEM). Two different ring-stiffeners were attached
to the specimens. Only the uniform hydrostatic pressure was considered as the main loading; the axial stress
was neglected in this study. The buckling modes, lateral displacement of the pipeline, ring tilt, formation of yield-
ing lines, and torsion of yielding lines were examined. The results indicated that increasing the number of ring-
stiffeners greatly increases the buckling capacity and lateral displacement of the pipeline. The buckling modes
were changed, the post-buckling region was shortened, and torsion in the yielding lines and tilting in the ring-
stiffeners clearly appeared in the tests. The results of this study were compared with those of recent research
and reliable guidelines and standards.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
The uniform external hydrostatic pressure is the main constant load
that affects subsea pipelines. When it combined with an external or in-
ternal factor such as bending during installation, a drill collar [1,2], or
corrosion on the pipeline wall [3–6], raising the hydrostatic pressure
to the buckling pressure Pcr will cause the pipeline to buckle locally
from a small cavity in the pipeline wall.
After the initial buckling, the buckling propagates if the pressure on
the pipeline is higher than the propagation pressure PP [7]. This is a post-
buckling phenomenon that causes a progressive structural failure. This
occurrence causes the pipeline to flatten in a dog-bone shape (Fig. 1).
There is a transition zone between the buckled and unbuckled regions
(Fig. 1). After the initial buckling, the propagation reaches a steady
state. In this situation, the external hydrostatic pressure is constant.
The buckling propagation only stops when the external hydrostatic
⁎ Corresponding author.
E-mail addresses: r_shahandeh@yahoo.com (R. Shahandeh), h.showkati@urmia.ac.ir
(H. Showkati).
pressure becomes less than the propagation pressure or the bending ri-
gidity of the pipe is suddenly increased by a buckle arrestor [8]. After the
initial buckling, the arrestors limit the spreading damage, which may
destroy hundreds of meters of pipeline [9].
The potential for buckling propagation and the limited availability of
repair tools have encouraged designers to use conservative pipeline de-
sign approaches based on the propagation pressure instead of the col-
lapse pressure. For this reason, pipelines are designed with 1.5–2
times the thickness needed to resist the collapse of intact pipes [7]. Ac-
curately predicting the propagation pressure for deep-water pipelines
will help lead to less conservative design recommendations. A prefera-
ble alternative is to design the pipeline based on the collapse pressure
and install buckling arrestors at regular intervals along the pipeline.
In other words, the buckling and collapse due to external pressure
are important issues that determine the design of subsea pipelines. An-
other important concern is controlling the propagation of the initial
buckling in a pipeline. The elastic stability theory and its applications
can be divided into the categories of pre- and post-buckling [10].
Kyriakedes et al. [11,12] indicated that a pipeline deforms by following
a stable pre-buckling path before the buckling load. Increasing the

http://dx.doi.org/10.1016/j.jcsr.2016.02.006
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
238 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 3. Buckling propagation in post-buckling stage along pipeline wall, between arrestors.

buckling pressure Pcr:

Fig. 1. Steady-state buckling propagation in pipeline.

3Et I
P cr ¼ ð1Þ
R3

pressure causes plastic hinges to form on the pipeline section, and the
where R is the pipeline radius, Et is the tangent modulus of the pipe-
section crumples, respectively. In this step and before buckling, the
line material and I is the moment of inertia of the pipeline cross-
pressure increases rapidly. Along this stable pre-buckling path, the pipe-
section.
line experiences a contraction in the radial direction. When the external
Bryan [18] formulated an expression similar to Eq. (1) for a free-
pressure reaches a critical value (i.e., the buckling load Pcr), buckling oc-
standing long pipeline under hydrostatic pressure. His equation differs
curs. After this, the post-buckling path begins (Figs. 2 and 3). Fig. 2
in terms of (Et), which is replaced with Et/(1 − υ2) to account for the
shows the deformation of pipeline section along buckling, and Fig. 3
buckling pressure under the plane strain condition of the infinitely
shows a plan view of pipeline wall deformation as well as stages of
yielding lines formation along buckling. For elasto-plastic materials,
the post-buckling behavior is highly unstable [13]. A slight increase in
pressure produces ovalization until yielding occurs along the pipeline
wall. Yielding lines appear during this step on the pipeline walls. After
yielding, the load-carrying capacity of the pipeline drastically falls.
Buckling starts in this manner; first, a local collapse begins at a weak
point on the pipeline wall [14,15]. This collapse forms a circular cavity
on the pipeline wall owing to increasing external pressure. The number
of these circular cavities on the pipeline wall then increases with the ex-
ternal pressure. Their spread leads to ovalization on the pipeline wall. As
these oval cavities spread and reach each other, yielding lines appear on
the pipeline wall. The propagation pressure is found to depend on the
diameter-to-thickness ratio of the pipeline geometry and the yielding
stress and tangent modulus of the pipeline material.
Research on pipelines under external pressure started from the
mid-nineteenth century [16]. The pipe length and ratio of the diam-
eter to the wall thickness were found to be important parameters
that determine the buckling pressure. Bresse [17] used the small de-
flection theory to study the stability of a pipeline under hydrostatic
Fig. 4. Shape and section of ring a.
pressure and proposed the following formula to estimate the

Fig. 2. Stages of buckling and post-buckling behavior on cross-section of pipeline. Fig. 5. Shape and section of ring b.
 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 239

Table 1
Geometry of specimens.

Specimen label PE a R 2 PE a R 4 PE a R 7 PE a R 13 PE b R 2 PE b R 4 PE b R 7 PE b R 13

Specimen length (L) (m) 2.46 2.46 2.46 2.46 2.46 2.46 2.46 2.46
Specimen radius (R) (m) 0.051 0.051 0.051 0.051 0.051 0.051 0.051 0.051
Specimen thickness (t) (mm) 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Ring type a a a a b b b b
Ring number 2 4 7 13 2 4 7 13
Ring spacing (LC) (m) 2.46 0.82 0.41 0.205 2.46 0.82 0.41 0.205
Total length/radius ratio 48.24 48.24 48.24 48.24 48.24 48.24 48.24 48.24
Unsupported length/radius ratio 48.24 16.07 8.04 4.02 48.24 16.07 8.04 4.02
Radius/thickness ratio 204 204 204 204 204 204 204 204

long pipe:

3
2Et t
P cr ¼
1−υ2 D
where D is the mean pipeline diameter, Et is the tangent modulus of
the pipeline material, t is the mean pipeline wall thickness and (υ) is
Poisson's ratio.
The buckling propagation of pipelines was initially considered by
Mesloh et al. [19], but the first paper was published on this subject by
Palmer [20], who proposed an equation for the propagation pressure
based on the strain energy of the collapsed cross-section. The results
of this research underestimated the experimental values for low ratios
of the external diameter to the thickness (D/t). This is typical of deep-
water scenarios because of the effects of plastic deformation [7]. The
first experimental studies were performed by Johns et al. and Mesloh
et al. [21,22], who tested different arrestor geometries to stop the buck-
ling propagation. They proposed an empirical formula to evaluate the
propagation pressure PP in terms of the ratio D/t and yielding stress σY:

2:5
PP 2t
¼6
σY D pipeline wall thickness.
where σY is the material yielding stress, D is the mean pipeline diameter,
and t is the mean pipeline wall thickness.
Kyriakedes and Babcock [23] empirically determined that for the
range of pipeline materials and geometric parameters used in the off-
shore industry, the propagation pressure can be anywhere from 0.1 to
0.5 of the pipeline elastic buckling pressure. The buckling pressure is
an important factor in the assessment and design of a pipeline.
Palmer and Martin [20] investigated the buckling propagation prob-
lem theoretically. They developed one of the earliest approximations of
the buckling propagation pressure for pipelines PP, which is given
below. The material was assumed to be rigid and perfectly plastic.
They proposed a model of four plastic hinges to describe the collapse
ð2Þ
mechanism of the pipeline during buckling propagation by using an en-
ergy balance:


πσ Υ t 2
PP ¼ ð4Þ
4 R
where σY is the material yielding stress, R is the pipeline radius, t is the
mean pipeline wall thickness.
Kyriakedes and Babcock [23] studied the dynamic aspects of buck-
ling propagation experimentally and theoretically and developed an
empirical equation for calculating the propagation pressure on alumi-
num and steel:


2:25
PP Et t
¼ 10:7 þ 0:54 ð5Þ
σY σY D
where σY is the material yielding stress, D is the mean pipeline diameter,
Et is the tangent modulus of the pipeline material, and t is the mean
ð3Þ
This paper presents a continuation of previous experimental works
by the authors [13]. The buckling and post-buckling behaviors of ring-
stiffened pipelines under hydrostatic pressure were evaluated. Different
specimens were experimentally tested with different type of arrestors.
Similar to the previous study, the experiments were then simulated
on FEM software (LUSAS) [24] with exact matches for the geometry,
geometric imperfections, loading, and boundary conditions for analysis.
Finally, the obtained experimental and FEM results were evaluated and
compared with the results of the previous study, previous research, and
reputable standards and guidelines.

2. Experiments

2.1. Specimens

In the experimental study, four new specimens were manufactured

to be geometrically similar to the specimens that had been used in pre-
vious research for comparison. In total, eight specimens divided into

Fig. 6. Created geometric imperfections on cross section of specimens. Fig. 7. Meshing on the wall of specimens and created nodes.
240 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 8. Measuring geometric imperfections.

two groups were considered and studied. The first group included four
specimens manufactured with ring type a (previous study). The second
group had four specimens manufactured with ring type b (new re-
search). Figs. 4 and 5 show the geometries of rings a and b. The ring-
stiffeners were vertically attached to the axes of the specimens to in-
crease their buckling capacity [25]. Table 1 lists the geometric properties
of the specimens.
The specimens were labeled as PE b R 7. The second symbol repre-
sents the type of ring-stiffener attached to the specimen, and the end
symbol represents the number of ring-stiffeners attached to the speci-
men. Two rings were attached to the two ends of all specimens, and
the other rings were attached with equal spacing from each other. How-
ever, the intervals between ring-stiffeners differed for each specimen.
For exact simulation of the specimens in FEM software, their geom-
etry, especially the imperfections, and mechanical properties of mate-
rials were carefully measured before the tests, which are being
explained below.
2.1.2. Mechanical properties of materials
The specimens and ring stiffeners were manufactured of steel plates.
Their materials were tested in a universal material test machine. Stress–
strain curve of materials was obtained from the results of three coupon
test specimens in which the average amount of yield stress, Poisson's ra-
tion and tangent modulus of materials were 250 MPa, 0.3 and 2E5 MPa,
respectively. According to these tests, the material of specimens and
ring stiffeners had similar mechanical properties; Fig. 9 shows the
stress–strain graph of this material.
2.1.3. Specimen manufacturing
The test specimens were formed of two parts; the first part was the
pipeline and the second part was the ring stiffeners which were at-
tached to the pipeline after manufacturing.

2.1.1. Measuring specimen geometric imperfection

During the manufacturing process, specimens were caught in geo-
metric imperfections in the form of cavities or knobs on cross section
of specimens as, radius decreasing or increasing, respectively (Fig. 6).
Before experiments, all specimens were meshed to 12 and 24 parts
circumferentially and longitudinally, respectively, as shown in Fig. 7 in
which all geometric imperfections were measured in radial direction
on all nodes of longitudinal meshes of specimens, respectively, by rotat-
ing the specimen via string and ruler system (Fig. 8). Magnitudes of ra-
dius decreasing or increasing on nodes of specimens were transferred Fig. 10. Soldering two edges of rolled plate for forming pipeline.
on nodes of FE models in LUSAS software as radial deformation.

Fig. 9. Stress–strain curve of material. Fig. 11. Attaching ring stiffeners to pipeline by soldering.
 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 241

Fig. 12. General overview of test setup for specimens.

Pipelines were manufactured of steel plates with 0.25 mm thickness.
At first, plates were cut with 32 cm width and 246 cm long, and then
were rolled by rolling machine. After carrying out these steps, for
forming the pipeline with 10.2 cm diameter and 246 cm long, two
edges of rolled plate were soldered together by tin and lead (Fig. 10).
Ring stiffeners (types a and b), which were manufactured by turnery
machine, were attached to specimens at determined places by soldering
(Fig. 11).
2.2. Instruments and experimental setup
Fig. 12 shows a general view of the specimens. The test specimens
were surrounded at both ends with two rigid plates (Fig. 12). In order
to fix the specimens on the plates, circular grooves with 2 mm depth
and 1 mm width were prepared by turnery machine on the side plates
with diameters equal to the specimen diameter. This point must be ex-
plained that boundary conditions for end section of pipeline have six
components which consist of three displacements and three angular de-
formations. Three displacements are along pipeline axis, radial and
circular (circumferential) directions and three angular deformations
which are around three displacement directions, which have been men-
tioned before. The prepared grooves provided a simply-supported
boundary condition at the two ends of the specimens which prevents
radial deformation of the two ends of the pipeline but releases other
five components at the end cross sections of specimens. This should
be explained that, there are not any connections between specimen
and side plates by soldering that could prevent rotation of end cross sec-
tions of pipeline, and for this reason the two ends of the specimen could
easily rotate in the provided grooves on side plates. The only connection
between the two ends of the specimens with side plates was provided
by silicon glue, which was used for preventing air leakage along tests
(Figs. 13a, b and 14). The two side plates were each fixed with a mid
shaft and four nuts, which helped prevent axial loading on the speci-
mens and provided a stable situation for the specimens (Fig. 15).
In order to record the radial displacements and circumferential and
longitudinal strains on the specimen walls during the tests, five trans-
ducers and two circumferential and longitudinal strain gauges were at-
tached to the specimens (Figs. 16 and 17). The equipment was

Fig. 13. a. Inside and outside plan of side plates. b. Inside and outside plan of side plates.
242 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 14. Rotation and angular deformation of specimen's ends and simply-supported
boundary condition at ends of specimens.

connected to a data-logger system that recorded all information on a

computer. The YEFLA-5 strain gauge (series YF/Y) from Tokyo Sokki
Fig. 16. Transducer positions on specimen.
Kenkyujo Co. Ltd. was used. This is a post-yield foil/wire gage that fea-
tures a special plastic carrier base in order to withstand extreme elonga-
tion without creeping or cracking and can measure strains of 10%–20%
with high accuracy.
The uniform external hydrostatic pressure loading on specimens
was simulated with a system that included a vacuum pump and vent
pipe. The system was connected to the specimens by a port on the
side plates (Fig. 12). At the beginning of the test, the vent pipe was
open, and the vacuum pump was operational. In order to load the spec-
imens, the vent pipe was gradually closed; consequently, the air inside
the specimens was evacuated, and the specimens were loaded incre-
mentally. For preventing air leakage along tests, all soldered points on
specimens and connection points between specimen with side plates
and attached instruments on side plates were insulated by silicon glue
(Fig. 18a and b).
In order to record the entrance pressure on the loaded specimen, a Fig. 17. Positions of longitudinal and circumferential strain gauges on specimens.
port was prepared on one of the side plates. A manometer was connect-
ed to the specimen, so that the entrance pressure on the specimen could element can also model warped configurations [26]. The element for-
be recorded for each test (Fig. 12). mulation considers membrane, shear, and flexural deformations. This
element uses an assumed strain field to define the transverse shear,
3. Finite element method which ensures that the element does not lock when it is thin.
For the modeling, the residual stress and slight axial stress were not
In order to verify the accuracy of the experimental study, the FEM considered. A completely simply-supported boundary condition was as-
analysis was performed with the LUSAS software. All specimens that sumed for two end sections of the models. However, this condition may
had been tested in the experimental study were exactly modeled with not completely match that in the experiment. This may have produced a
regard to the geometry, initial geometric imperfections, simply support- few differences between the experimental and FEM results, which are
ed boundary condition, and loading and mechanical properties of the discussed later in this paper.
specimen material (Figs. 9, 19). The yield stress of the material was
250 MPa based on tension tests during the experimental study.
For exact modeling, the models were meshed more than experimen- 4. Buckling and post-buckling behavior and buckling propagation
tal study (four times as much in each directions), with 48 and 96 parts in
the circumferential and longitudinal directions, respectively (Fig. 20). 4.1. Experimental study
All elements had four nodes and four sides; this is a quadrilateral shell
element called QTS4 in LUSAS [26]. The rings were meshed according Figs. 22–28 compare the experimental information of the specimens.
to the pipeline mesh (Fig. 21a, b, and c). QTS4 is a convenient element The graphs show the nonlinear elastic behavior and first hardening of
for thick- and thin-curved shell geometries that include multiple each specimen in the pre-buckling region. The initial local buckling
branched junctions. The quadratic element can accommodate generally started as a cavity on the wall of each specimen during the final stages
curved geometry and varying thickness. This degenerated continuum of the pre-buckling region (Figs. 29 and 30).

Fig. 15. Position of mid shaft and specimen.

 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 243

Fig. 20. Meshing the FEM analysis models.

4.2. FEM study

Figs. 34–38 compare the experimental and FEM study results on

the deformation of the specimen wall. The two sets of results showed
good similarity. The graphs show the nonlinear elastic characteris-
tics for all models, similar to the experimental results, in the pre-
buckling region. However, in the post-buckling region, the graphs
for the models with few arrestors showed a constant or decreasing
gradient after a slight hardening. After this step, failure occurred. In
models with more arrestors, failure suddenly occurred after the
pre-buckling region. Similar to the experimental results, the models
with more arrestors did not have a distinct distance between charac-
Fig. 18. a. Insulating the connection points between specimen with side plates and teristic pressures.
attached instruments. b. Insulating the soldered points of specimen. In the FEM models, when the post-buckling region started, the
initial buckling began through the formation of a cavity at a weak
point of the model wall. Buckling propagation directly began as the
formed cavity developed in the longitudinal and circumferential di-
As the experiment continued, the specimen entered the post-
rections (Fig. 39). Ovalization and then yielding lines were observed
buckling region with nonlinear and inelastic characteristics. Conse-
on the models before failure (Fig. 40). Fig. 41 shows the stages of
quently, the strength of the specimen and the pressure recorded by
buckling on the walls of the models. Table 2 presents all of the infor-
the manometer decreased. At the final point of this region, buckling
mation from the FEM analysis.
started propagating on the wall of the pipeline, and the formed cavity
gradually developed in the longitudinal and circumferential directions
into an oval shape (Figs. 30 and 31), as explained in Figs. 2 and 3.
As the test continued, a poly-wave shape appeared on the specimen
section depending on the number of arrestors attached to the specimen
(Fig. 32a, b and c).
As the experiment continued, the strength of the specimen in-
creased (second hardening in graphs), and yielding lines gradually ap-
peared on the specimens. Consequently, planes with a polygonal
shape formed on the walls of the specimens (Fig. 33). Ultimately, failure
occurred in the final part of this region. Note that the second hardening
of the specimen has been eliminated from the graphs to best compare
the experimental results of the specimen tests (Fig. 30). Table 2 presents
all of the experimental information.
The graphs show that increasing the number of arrestors attached to
the specimen increases the buckling, buckling propagation, and failure
pressures, and these values will be closer together, especially in speci-
mens with 13 arrestors.

Fig. 21. a. Finite element modeling and model meshes. b. Arrestor meshes on finite
Fig. 19. External uniform specimen loading. element model. c. QTS4 element.
244 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 25. Pressure–circumferential strain of specimens PE b R 2 and PE b R 4.

Fig. 22. Pressure–axial strains of specimens PE a R 7 and PE a R 13.

5. Verification of experimental observations

5.1. Circumferential buckling modes

As observed in the experiments, after yielding lines formed on the

walls of the specimens, the circular section of the pipeline changed to
a multi-wave section (Fig. 32a). The number of waves formed on the
specimen section was labeled n. This factor strongly depends on R, 1/t,
and 1/LC. The experimental and FEM analysis results showed good
agreement with regard to n, as presented in Table 3. The results in this
study on n were verified through a comparison with a previous simpli-
fied formulation [27], which is given below:

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffi!
Fig. 26. Pressure–displacement relationships of specimens PE a R 7 and PE a R 13.
u
u R R
n ¼ 2:74t ð6Þ
LC t

Fig. 27. Pressure–displacement relationships of specimens PE b R 7 and PE b R 13.

Fig. 23. Pressure–axial strains of specimens PE b R 2 and PE b R 4.

Fig. 24. Pressure–circumferential strains of specimens PE b R 7 and PE b R 13. Fig. 28. Pressure–displacement relationships of specimens PE b R 2 and PE b R 4.
 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252
Fig. 29. Cavity on pipeline wall as initial buckling begins.
Fig. 30. Stages of buckling and buckling propagation in experimental specimens.
where R, t and LC are the radius and thickness of pipeline, and distance of
the attached ring stiffeners, respectively.
Each set of results showed that increasing the number of attached ar-
restors increased the buckling capacity of the specimens and thus the
value of n.
5.2. Torsion on yielding lines
After yielding lines formed on the walls of the specimens, the presence
of geometric imperfections, geometric properties of the specimens, and
circumferential torsional stress imposed on the pipeline wall caused the
straight yielding lines to change to V-shape lines, especially at higher
pressures on specimens with more arrestors. This is shown in Figs. 42a
and b and 43a–c. This phenomenon was also reported by Showkati for cy-
lindrical shells [25]. This phenomenon strongly appeared in specimens
that had more arrestors and thus a high buckling capacity.
5.3. Formation of inflection point near rings and ring tilting
Rings were attached to the specimens to increase their radial
strength against buckling. The rotation of the pipeline wall generated
circumferential torsional stress, and the interval distance between
rings was reduced because of deformation from the straight yielding
Fig. 31. Ovalization on pipeline wall as buckling propagation begins.
245
lines to V-shape lines from crumpling panels between rings. This caused
rings to be located at the inflection point of axial buckling modes
(Figs. 44a and b). These reasons caused out-of-plane deformation at
the rings. This is a mode of lateral buckling where the rings tilted. This
phenomenon was clearly observed for ring type a. This is because the
moment of inertia around the Y axis and moment of torsional inertia
for ring type a were less than those for ring type b. Thus, ring type b
had a higher strength capacity than ring type a (Fig. 45). This phenom-
enon is shown in Figs. 46 and 47.
5.4. Lateral displacement of specimens
In the experiments, after the yielding lines formed, the specimens
were brought to their failure point. The straight longitudinal axis of
the specimen changed with lateral displacement to a curved line be-
cause of the generated circumferential torsional and axial stresses on
the specimen walls. This phenomenon strongly appeared in specimens
with more attached ring stiffeners, as shown in Figs. 48 and 49.
Thus, increasing the buckling and failure pressures of specimens by
increasing the number of attached ring stiffeners or modifying geomet-
ric properties such as the ratios R/L and t/R was inferred to intensify the
lateral displacement of the specimens.
5.5. Folding on specimen wall
As the tests continued, two series of stresses were created on the
specimen wall. One, was the torsional stress in circumference direction,
which caused rotation of the specimen wall and ring tilting, and the
other was the axial stress, which caused the specimen walls to fold.
This is a post-buckling phenomenon that rarely appeared in the exper-
iments. In fact, it only appeared in two specimens: PE a R 13 and PE b
R13. This means that increasing the number of attached ring stiffeners
and thus the buckling capacity of the specimens increased the amount
of generated axial stress and allowed its effects to be observed. This phe-
nomenon is shown in Fig. 50.
6. Verification of material consumption
In order to consider the weights of the specimens and attached rings,
Tables 4 and 5 were prepared to investigate the efficiency of the ring stiff-
eners (types a and b) with regard to the amount of steel used to improve
246 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 32. a. Three- and four-wave shape on section. b. Three- and four-wave shape on section. c. Five-wave shape on section.

Fig. 33. Hexangular shape on pipeline wall as yielding lines form.

Table 2
Comparison of buckling and buckling propagation pressure.

Specimen label PE a R 2 PE a R 4 PE a R 7 PE a R 13 PE b R 2 PE b R 4 PE b R 7 PE b R 13

Experimental buckling pressure (kPa) 5.96 16.88 27.84 61 6.56 18 36.5 60

FEM buckling pressure (kPa) 6.87 17.59 30.74 50 7.68 22.49 45.54 73.62
Discrepancies between experimental and FEM results 15% 4% 10% −19% 17% 25% 15% 22.7%
Experimental buckling propagation pressure (kPa) 3.4 6.9 13.5 54 3.5 8.3 23 No recorded

the buckling behavior of the pipeline. One specimen of each series (PE a R
2 and PE b R 2) was taken as a representative for the economic evaluation.
Based on Tables 4 and 5, using 13 ring stiffeners of types a and b in-
creased the material weight of the specimens by about 36.5% and 28.8%,
respectively, while the buckling strength was increased about 10.23 and
9.14 times, respectively. Thus, the use of ring stiffeners can be inferred
to strengthen a pipeline economically.
7. Evaluation with standards
The obtained experimental buckling propagation pressures (PP) of
the unstiffened specimens in this study (PE a R 2 and PE b R 2) were
compared against the recommended values for the buckling propaga-
tion pressure (PP) of unstiffened pipelines from API (1999) and ABS
(2005) [28,29], which are given in expressions (7) and (8), respectively.
 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 247

Fig. 38. Comparison of experimental and FEM results for PE b R 13.

Fig. 34. Comparison of experimental and FEM results for PE a R 4.

Fig. 35. Comparison of experimental and FEM results for PE b R 2.

Fig. 39. Ovalization on FEM model wall as buckling propagation begins.

The comparison is presented in Table 6.

2:4
t
P P ¼ 24σ Y ð7Þ
D

Fig. 36. Comparison of experimental and FEM results for PE b R 4.

Fig. 37. Comparison of experimental and FEM results for PE a R 7. Fig. 40. Formation of yielding lines on wall FEM model.
248 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 41. Stages of buckling and buckling propagation in walls of FEM models.

Table 3
Comparison of n.

Specimen label PE a PE a PE a PE a R PE b PE b PE b PE b R
R2 R4 R7 13 R2 R4 R7 13

n, experimental study 2 3 3 5 2 2 4 5
n, FEM analysis 2 3 4 5 2 3 4 5
n, expression 6 2 3 4 5 2 3 4 5

2:5
2t
P P ¼ 6σ Y ð8Þ
D
Fig. 43. a–c. Change from straight yielding lines to V-shape lines before failure.

PP, σY, t, and D are the buckling propagation pressure, material yield-
ing stress, thickness, and diameter of specimen, respectively.
9. Evaluation with Batdorf parameters
Based on Table 6, expression (7) from API (1999) estimated the ex-
perimental buckling propagation pressure almost exactly. Note that
For another evaluation on the buckling pressures from the results of
both expressions underestimated the experimental pressure.
the experiment, the FEM, and expression (9), Batdorf parameters were
used to plot a graph. In this graph, the Batdorf parameters were used as
8. Evaluation with simplified relations
2 2 3
X and Y axes. The X axis was N ¼ ðLRC Þ ðRt Þ, and the Y axis was Q ¼ ðLRC Þ ðRt Þ
A simplified relation was presented [27] to predict the buckling ðPEcr Þ, where P cr ¼
2:5
0:92EðLRC ÞðRt Þ . Fig. 51 shows this comparative graph,
pressure of pipelines, as shown in expression (9).
where R, t, E, LC and Pcr are the radius, thickness of pipeline, elasticity mod-
ulus of the material, distance of the attached ring stiffeners and the buck-


2:5
R t ling propagation pressure, respectively.
P cr ¼ 0:92E ð9Þ
LC R This comparison showed that the experimental results had conve-
nient agreement with the theoretical results than the FEM results, espe-
cially for specimens with few attached ring stiffeners.
In this expression, R, t, E, LC and Pcr are the radius, thickness of pipe-
line, elasticity modulus of the material, distance of the attached ring
10. Conclusion
stiffeners and the buckling propagation pressure, respectively.
Table 7 presents the verification between the experimental and FEM
In this study, experiments were performed on subsea steel pipelines
results and the results obtained with expression (9).
that were reinforced with ring stiffeners under hydrostatic pressure,
Despite some discrepancies between individual specimens, overall
and the results were verified by FEM analysis. The main conclusions
there was convenient agreement between the experimental and FEM
are as follows:
results with the results of expression (9) with regard to the buckling
pressure Pcr. This was especially true for the experimental results and • The use of ring stiffeners increases the amount of steel by 32% while
results of expression (9). increasing the buckling strength 9.7 times. Thus, the use of ring

Fig. 42. Straight yielding lines when initial buckling began.

 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 249

Fig. 44. a, b. Inflection point on rings between two side panels.

Fig. 47. Stages of ring tilting during experiments.

Fig. 45. Moment of inertia around Y axis.

stiffeners is more efficient than increasing the thickness of the pipe-
line wall for enhancing the buckling capacity of the pipeline.
• Increasing the number of ring stiffeners decreased the discrepancy be-
tween the buckling, buckling propagation, and final failure pressures.
This shortened the post-buckling region, and failure suddenly oc-
curred after high-speed buckling.
• The pre-buckling and post-buckling regions in the experimental and
FEM graphs showed nonlinear behavior for most of the specimens.
• After yielding lines formed on the specimens, oval and hexangular
shapes consistently appeared on the specimen walls between the
yielding lines on each panel.
• According to the graphs, the buckling capacity of the specimen in-
creased in two stages: after the initial local buckling began and after
the buckling propagation began.
• When the buckling reached the ring position, all rings were located at
the inflection point between two side panels.
• The torsional pattern of the yielding lines was V-shaped for buckling
modes under high external pressure when the number of attached
ring stiffeners was increased.

Fig. 46. FEM and experimental model of ring tilting.

250 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252

Fig. 48. Lateral displacement of experimental specimens in experiments. Fig. 49. Lateral displacement of models in FEM analysis.
 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252 251

Fig. 50. Folding on specimen walls.

Table 4
Evaluation of material consumption for specimens with ring stiffener type a.

Specimen label Specimen Weight increase by Buckling Pressure

weight (N) stiffeners (%) pressure (kPa) ratio

PE a R 2 16.25 0 5.96 1
(basic specimen)
PE a R 4 17.33 6.6 16.88 2.83
PE a R 7 18.95 16.6 27.84 4.67
PE a R 13 22.19 36.5 61 10.23

Fig. 51. Comparison between results of experiment, FEM, and expression (9) based on
Batdorf parameters.
Table 5
Evaluation of material consumption for specimens with ring stiffener type b.
• Changing the ring section and increasing the moment of inertia
Specimen label Specimen Weight increase by Buckling Pressure
weight (N) stiffeners (%) pressure (kPa) ratio around the Y axis and torsional moment of inertia of the ring stiffeners
increased the resistance of the rings against lateral buckling.
PE b R 2 16.01 0 6.56 1
(basic specimen) • Increasing the number of ring stiffeners intensified the generated cir-
PE b R 4 16.85 5.2 18 2.74 cumferential torsional and axial stresses. Consequently, phenomena
PE b R 7 18.11 13.11 36.5 5.56 such as torsional patterns on the yielding lines, ring tilting, and folding
PE b R 13 20.63 28.8 60 9.14 on the specimen wall appeared more strongly in the test specimens.
• The number of formed waves on the section of the specimens was di-
rectly related to the buckling capacity of the specimens. Higher modes
represented higher buckling pressures.
Table 6
• The lateral displacements of the specimens increased with the num-
Comparison of buckling propagation pressures of unstiffened specimens between experi- ber of ring stiffeners attached to the specimens.
mental study and standards. • There was convenient agreement between the FEM and experimental
Specimen label PE a R 2 PE b R 2
results and the values of recommended equations in the standards, es-
pecially the API standard, with regard to the buckling and buckling
PP for experimental study (kPa) 3.4 3.5
propagation pressures and the buckling modes.
PP for expression (7) (kPa) 3.26 3.26
PP for expression (8) (kPa) 2.52 2.52 • There was convenient agreement between the FEM and experimental
Discrepancy in PP between expression (7) and experimental PP 4% 7% results and previous works with regard to the buckling propagation
Discrepancy in PP between expression (8) and experimental PP 35% 39% when evaluated with Batdorf parameters.

• When the buckling propagation reached the ring stiffeners, it created

a torsional pattern on the yielding lines and increased the circumfer- Acknowledgment
ential torsional stress on the specimen walls. Lateral buckling modes
occurred on the ring stiffeners, which caused out-of-plane displace- The authors acknowledge the financial support from the Iranian Min-
ments for the ring stiffeners. istry of Roads and Urban Development (Project number: 85B8T9P01,

Table 7
Comparison of buckling pressure Pcr from experimental study and expression (9).

Specimen label PE a R 2 PE a R 4 PE a R 7 PE a R 13 PE b R 2 PE b R 4 PE b R 7 PE b R 13

Pcr for experimental study (kPa) 5.96 16.88 27.84 61 6.56 18 36.5 60
Pcr for FEM study (kPa) 6.87 17.59 30.74 50 7.68 22.49 45.54 73.62
Pcr for expression (9) (kPa) 6.09 18.29 36.58 73.16 6.09 18.29 36.58 73.16
Discrepancy in Pcr between expression (9) and experimental study 2% 8.3% 31.3% 20% 8% 1.6% 0.2% 21.9%
Discrepancy in Pcr between expression (9) and FEM study 12% 3.9% 18.9% 22% 26% 22.9% 24.4% 0.6%
252 R. Shahandeh, H. Showkati / Journal of Constructional Steel Research 121 (2016) 237–252
confirmation number: 62-2437). This work was performed in collabora-
tion with the Structural Research Center at Urmia University.
References
[1] T.G. Johns, R.E. Mesloh, R. Winegardner, J.E. Sorenson, Inelastic buckling of pipelines
under combined loads, Proceeding of the Offshore Technology Conference, Houston
TX, OTC Paper 2209, vol. II 1975, pp. 635–646.
[2] G.T. Ju, S. Kyriakedes, Bifurcation buckling versus limit load instabilities of elastic–
plastic tubes under bending and external pressure, ASME J. Offshore Mech. Arct.
Eng. 113 (1991) 43–52.
[3] J. Xue, M.S. Hoo Fatt, Propagating buckles in corroded pipelines, Mar. Struct. 14
(2001) 571–592.
[4] J. Xue, M.S. Hoo Fatt, Y. Liu, Steady-state buckle propagation in corroded pipelines,
Proceeding of the 10th International Offshore and Polar Engineering Conference, Se-
attle, May 28–June 2 2000, pp. 197–204.
[5] M.S. Hoo Fatt, J. Xue, Buckling of a non-uniform shell subjected to external hydro-
static pressure, Eng. Struct. 24 (2002) 1027–1034.
[6] M.S. Hoo Fatt, J. Xue, Symmetric and anti-symmetric buckle propagation modes in
subsea corroded pipelines, Mar. Struct. 18 (2005) 43–61.
[7] I.P. Pasqualino, S.F. Estefen, A nonlinear analysis of buckle propagation problem in
deepwater pipelines, Int. J. Solids Struct. 38 (2001) 8481–8502.
[8] T.A. Netto, S.F. Estefen, Buckle arrestors for deepwater pipelines, Mar. Struct. 9
(1996) 873–883.
[9] S. Kyriakides, T.A. Netto, On the dynamics of propagating buckles in pipelines, Int. J.
Solids Struct. 37 (2000) 6843–6867.
[10] J.W. Hutchinson, W.T. Koiter, Post-buckling theory, Appl. Mech. Rev. 23 (1970)
1353–1356.
[11] S. Kyriakedes, C.D. Babcock, Large deflection collapse analysis of an inelastic
inextensional ring under external pressure, Int. J. Solids Struct. 17 (10) (1981)
981–993.
[12] S. Kyriakedes, E. Arikan, Postbuckling behavior of inelastic inextensional rings under
external pressure, J. Appl. Mech. 50 (1983) 537–543.
[13] H. Showkati, R. Shahandeh, Experiments on the buckling behavior of ring-stiffened
pipelines under hydrostatic pressure, J. Eng. Mech. 136 (4) (2010) 464–471.
[14] S. Kyriakedes, T.D. Park, On the collapse of dented cylinders under external pressure,
Int. J. Mech. Sci. 38 (1996) 557–578.
[15] S. Kyriakedes, C.D. Babcock, D. Elyada, Initiation of propagating buckles from local
pipeline damages. Part a, ASME J. Energy Res. Technol. 106 (1984) 79–87.
[16] W. Fairbairn, On the resistance of tubes to collapse, Philos. Trans. R. Soc. U.K. Lond.
148 (1858) 389–413.
[17] M. Bresse, Cours de méchanique appliquée, second ed., 1866 France, Paris.
[18] G.H. Bryan, Application of the energy test to the collapse of a long pipe under exter-
nal pressure, Proceeding of the Cambridge Philosophical Soc, UK, Cambridge, 6
1888, pp. 287–292.
[19] R.E. Mesloh, J.E. Sorenson, T.J. Atterbury, Buckling and offshore pipelines, Gas Mag. 7
(1973) 40–43.
[20] A.C. Palmer, J.H. Martin, Buckle propagation in submarine pipeline, Nature 254
(1975) 46–48.
[21] S. Kyriakedes, Buckle propagation in pipe-in-pipe system. Part I: experiments, Int. J.
Solids Struct. 39 (2) (2002) 351–366.
[22] S. Kyriakedes, T.J. Vogler, Buckle propagation in pipe-in-pipe system. Part II: analy-
sis, Int. J. Solids Struct. 39 (2) (2002) 367–392.
[23] S. Kyriakedes, C.D. Babcock, Experimental determination of the propagation pres-
sure of circular pipes, J. Press. Vessel Technol. 103 (1981) 328–336.
[24] LUSAS, PDF Manuals, Autoloader User Manual.
[25] Showkati, The Buckling Strength of Cylindrical Shells Under External Pressure, Univ.
of Sydney, Sydney, Australia, 1995 (Ph.D. thesis).
[26] LUSAS, PDF Manuals, Element Reference Manual, 221
[27] H. Showkati, P. Ansourian, Influence of primary boundary conditions on the buck-
ling of shallow cylindrical shells, J. Constr. Steel Res. 39 (2) (1996) 163.
[28] API, API RP 1111, Design, Construction, Operation and Maintenance of Offshore Hy-
drocarbon Pipelines, American Petroleum Institute, Washington, D.C., 1999
[29] ABS, American Bureau, Houston, Guide for Building and Classing Subsea Pipeline
Systems and Risers, 2005 64-01.