6th Int'l Conference on Mechanical, Production & Automobile Engineering (ICMPAE'2014) Nov.

27-28, 2014 Cape Town (South Africa)

Elasto-Plastic Collapse Analysis of Pipe

Bends Using Finite Element Analysis
1*

Tawanda Mushiri and 2Charles Mbohwa

Abstract---When an external load is applied to one of its ends, a

pipes bends cross section tends to deform significantly both in and
out of its end plane. This shell type behaviour characteristic of pipe
bends and mainly due to their curves geometry accounts for their
greater flexibility. This added flexibility is also accompanied by
stressed and strains that are much higher than those present in a
straight pipe.
The primary goal of this research is to study the elastic-plastic
behaviour of pipe bends under out of plane moment loading. It is
also required to study the effects of changing the value of the pipe
bend factor and the value of the internal pressure on that behaviour
and to determine the value of the limit moments in each case.
The results of these analyses are presented in the form of load
deflection plots for each load case belonging to each model. From
the load deflection curves, the limit moments of each case are
obtained. The limit loads are then compared to those computed using
some of the analytical and empirical equation available in the
literature. The effects of modelling parameters are also studied. The
results obtained from small displacement and large displacement
analyses are compared and the effects of using a strain hardened
material model are also investigated.
To better understand the behaviour of pipe elbows under out of
plane bending and internal pressure, it was deemed important to
know how the cross section deforms and to study the distribution of
stresses that cause it to deform in a particular manner. An elbow with
pipe bend factor h=0.1 to h=1 is considered and the results of the
detailed analysis are thereof examined.

released into the atmosphere. Examples of such substances

include steam, oil, sulphur and chlorine gas. Failure in a
piping system could cause problems, like an unscheduled, and
hence costly, plant shut down for maintenance; or even a
catastrophe, like exposing the core of a nuclear reactor.
Therefore, the integrity of pipes in industrial contexts is of
paramount importance. This integrity relies heavily on the
correctness of design codes and practices, which can only be
achieved through a thorough understanding of the behaviour
of piping components and systems under different types of
loads.
Stresses in piping systems are developed as a result of
sustained loads, like the weight of the pipe itself and the
pressure of the fluid running inside it. The effects of this kind
of loads are minimised by selecting an adequate type of
supports and using a sufficient number of them. Dynamic
loads of seismic origin or generated by a defective attached
device (e.g. pump or compressor), and thermal loads, which
cause different pipe segments to expand, also create stresses
within the piping system. Generally, dynamic and thermal
loads are more important and more complex to deal with.
Hence, it is vital that some means or mechanism, for relieving
these stresses, be present in the design of a piping system, to
avoid overloads which might in turn lead to failure of a pipe
segment, or cause damage to an attached device, vessel or
support.

Keywords---Elasto-plastic, finite element analysis, pipe bends,

To investigate the effect of internal pressure on the elasticplastic behaviour of pipe bends using Finite Element
Analysis.

A. Aim

simulation

I. INTRODUCTION

B. Objectives

ARGE pipelines and pipe networks are part of almost

every industrial setup today. These are most commonly
found in petroleum rigs, refineries, factories producing
chemicals and pharmaceuticals, and in power plants. In these
and other industrial applications, pipes are very often used to
carry substances that, by virtue of their pressure, temperature,
physical and chemical characteristics, can have serious
negative effects on health, property and the environment, if
1

To study, with different values of bend factors, the

behaviour of pipe bends
To investigate the limit pressures beyond which the pipe
bend will collapse, using finite element analysis
II. LITERATURE SURVEY
A. Theoretical Analysis Of Pipe Bends
First demonstrated by Bantlin (1910), curved pipes behave
differently under load than predicted by simple beam theory.
The first theoretical explanation of this discrepancy was
presented by Von Karman (1911), and much of the theoretical
work done subsequently was an extension of his work, or at
least based on the same principle of potential energy
minimization. Other closed-form solutions were based on the
mechanics of materials principles, or thin shell theory. All
these solutions acknowledge the fact that pipe bends have

D.Eng. Student; University of Johannesburg, Department of Mechanical

Engineering, P. O. Box 524, Auckland Park 2006, South Africa. Email id :
[email protected] and *Lecturer; University of Zimbabwe,
Department of Mechanical Engineering, P.O Box MP167, Mt Pleasant,
Harare, Zimbabwe.
2
Professor and Supervisor; University of Johannesburg, Auckland Park
Bunting Road Campus, P. O. Box 524, Auckland Park 2006, Room C Green
5, Department of Quality and Operations Management, Johannesburg, South
Africa. Email id: [email protected]

26

6th Int'l Conference on Mechanical, Production & Automobile Engineering (ICMPAE'2014) Nov. 27-28, 2014 Cape Town (South Africa)

higher flexibility and stresses, and a different stresses

distribution, from the ones predicted by simple beam theory.
The general approach followed in piping system flexibility
calculations, and still widely adopted in design codes, relies
on the use of flexibility factors and stress-intensification
factors, which are simply the ratios of actual flexibility and
stress to those predicted by simple beam theory. The different
approaches aim at providing more accurate estimates of these
correction factors, in a form that can adequately be simplified,
to become readily usable in design context.
A common drawback of most theoretical solution however,
is that they are based on elastic analysis concepts, and cannot
be used for the purpose of nonlinear analysis, which is
required to predict the response of a pipe bend correctly,
especially in accident conditions where plasticity behaviour
dominates. Nonlinear theoretical analysis has been explored
by a small number of workers, like Spence (1972). However,
these efforts did not realize enough success to warrant their
use in design. In addition, theoretical solutions are only useful
in the analysis of pipe bends of circular cross-section.
Von Karman (1911) publish the first theoretical stress
analysis of the problem of the problem of the problem of inplane bending of pipe bends, which was later generalised by
Vigness (1943), to include out of plane bending. In both of
these analyses, the deformation of the bends wall was
represented by a trigonometric series whose coefficients were
determined by minimizing the total potential energy.
Major simplifying assumptions were made in these
analyses:
Internal pressure effects are ignored.
The pipe bend is not connected to any straight pipes or
flanges. In other words, end-effects are ignored and a
standalone pipe bend is assumed.
The ratio of the pipe radius to the bend radius (r/R) is
negligible relative to unity, i.e. long radius elbows (R/r>>1)
were assumed.
The pipe bends cross-sections remain plane and
perpendicular to the bends centreline even after
deformation. This precludes warping effects from the
analysis.
Under torsional moment loading, the pipe bend was
assumed to behave like a straight pipe, keeping a circular
cross-section.
The hoop membrane strain in the circumferential direction
was assumed to be nil, which means that the cross-section
ovalizes without any changes in the length of its
circumference.
Both of the analyses, agreed with Bantlins results (1910), by
indicating that longitudinal tensile and compressive stresses in
the tubes wall, produce component forces acting towards and
away from the centre of curvature, respectively. These forces
result in the flattening of the cross-section into an oval shape,
as shown in the figure below which in turn accounts for the
increased flexibility and the different stress distribution
compared to those predicted by simple beam theory and those
found in straight pipes.
They also showed that the bends maximum stresses and its
deformation, namely its end rotation, are higher than those of
a straight pipe of the same size and material. These results,

expressed as flexibility and stress-intensification factors that

are greater than unity, depend on a dimensionless parameter,
the pipe bend factor (h), defined as:

h= R.t
r2

Fig 1. Cross-Sectional Deformation of a

Pipe Bend under In-Plane and Out-of-Plane Loading
(Dodge and Moore, 1972)

III. EXPERIMENTAL INVESTIGATION ON PIPE BENDS

The most comprehensive study, and the one that is referred
to most frequently in the literature, is the one conducted by
Greenstreet (1978), at the Oak Ridge National Laboratory
(ORNL). In this study, load-deflection responses were
determined for twenty 6-inch nominal diameter commercial
pipe elbows, sixteen of which were made of ASTM A-106
grade B carbon steel and the remaining four were made of
ASTM A-312 type 304L stainless steel.
They concluded that in the case of in-plane bending in the
opening direction, the ovalization of the elbow tends to
increase its stiffness, and displace the weak spot in the
assembly towards one of the junctions between the elbow and
the adjacent straight pipe segments. Failure takes the form of
a crease that appears at this weak spot and absorbs all
subsequent deformation, preventing the elbow from
deforming any further. In-plane bending in the closing
direction tends to decrease the stiffness of the elbow, and
failure takes the form of excessive ovalization, or flattening of
the mid-section. It is reported that this mode of loading is the
most critical, since failure takes place at a relative low value
of the applied moment.

27

6th Int'l Conference on Mechanical, Production & Automobile Engineering (ICMPAE'2014) Nov. 27-28, 2014 Cape Town (South Africa)

about finite element analysis of pipe bends were looked into

to see how the tests were done. This was able to provide a
substratum on which to lay the foundation of this numerical
analysis of pipe bends. Already published journals were also
analysed on subjects on elastic-plastic bending and in
particular, pipe bends. This was done as a way to show how
analysis of the elastoplastic behaviour of pipe bends is done
under different parameters of pressures applied and different
materials.
The data was therefore analysed, and trends were put down
on what has been calculated in experiments by those who
carried them out in their journals, articles and papers. All this
gave a guide line on how this project was going to be
undertaken, having analysed the experiments done in the
documents.
The internet was also consulted, which gave quite a number
of experiments carried out on this subject. Much emphasis
was put on the elastoplastic behaviour of different types of
materials, when they undergo some loading. Different
pressures were analysed to see how the different materials
behave under different loads.

Fig 2: Diagram of the Test Setup Used by

Greenstreet (1978)Hilsenkopf, Bonet and Sollogoub (1988)

conducted two series of tests on 900 long radius elbows. The
first series, which consisted of 10 tests, was conducted on
thick-walled TU 42C (equivalent to ASTMA A-106 grade B)
ferrite steel elbows, with an outside diameter to wall thickness
ratio (D0/t=6.7). a second series of 15 tests was conducted on
thin-walled Z2 CN 18-10 (equivalent to ASTM A-312 Type
304L) austenitic stainless steel elbows with Do/t=90.
During the tests, the elbows were subjected to in-plane
bending moments, in either the opening or the closing
direction, or to out-of-plane moments. From the results
obtained, and especially for the thin-walled elbows,
Hilsenkopf et al, reported that the cross-section of the elbow
tends to ovalize as soon as a bending moment is applied, in
any direction, to one of the elbows extremities. This
ovalization keeps increasing applied moment. The tests were
extended sufficiently to observe the ovalization modes under
different types of loading and to reveal the failure mode.
Sobel and Newman (1980, 1983 and 1986), conducted
three experiments on nominally identical 90o long radius pipe
bends, with an outer diameter of 16 inches and a wall
thickness of 0.42 inch. In the first test, which was conducted
at room temperature, and in the second test conducted at
1100oF, (593), an in-plane moment was applied in steps until
plastic collapse occurred. In the third test, conducted at
1100oF, the moment was increased to a value for about 2800
hours, to examine the creep behaviour of the elbow.

B. Experimental designs and system modelling

In the subject topic in question, experimental designs and
modelling was done. ABAQUS software was used to do
experimental design on the pipe bends that are being studied.
The experiments were done and were limited to one material
(type 304 stainless steel).
The model of the pipe bend was drawn using Solid Edge
software. Analysis of the internal pressures was done so as to
come up with a finite element analysis that shows the limits
beyond which the pipe bend will collapse, in an elastic-plastic
manner. Experiments were done and results were given
diagrammatically, at each stage, to show the elastic then
plastic behaviour of the type 304 stainless steel, when
subjected under different loads or pressures. Graphs showing
the elastic-plastic behaviour of the pipe bends were drawn to
enable us to see the trends in the elastoplastic behaviour of
type 304 stainless steel pipe bends. Computed resulted were
used to see the limits to which the bends can hold before they
collapse.
C. Complete system simulation
After having modelled the pipe bends, simulation was done
and results were shown diagrammatically, for how the system
will sustain the pressures they are subjected to before they go
elastic and plastic. ABAQUS was then used to simulate the
model. Results of simulation were analysed and are shown in
the document, diagrammatically. This is the finite element
analysis.

IV. METHODOLOGY

D. Finite element modelling and analysis

One main goal of this study is to investigate the effect of
internal pressure on the behaviour of pipe bends, with
different values of the bend factor (h=Rt/r2), under out-ofplane moment loading. A 90 standalone pipe bend, was
used throughout this study. This yields more conservative
estimates of limit loads, since the stiffening effect of endconstraints (e.g. attached flanges or connected straight pipe
segments) is neglected.

The methodology that this project will follow can be

summarised by three simple aspects:
Research formulation and data analysis
Experimental designs and system modelling
Complete system simulation
A. Research Formulation and data analysis
The tool that was used in this form of methodology is
documents review. Documents containing the information
28

6th Int'l Conference on Mechanical, Production & Automobile Engineering (ICMPAE'2014) Nov. 27-28, 2014 Cape Town (South Africa)
[4]

The pipe bend had a nominal diameter of 16 inches, and the

bends axis had a radius of curvature R=24in , to represent a
long-radius of radius ratio (R/r=3).

[5]

E. Material model
The material used throughout this study was type 304
stainless steel, following the measured response at room
temperature reported by Sobel and Newman (1979). This
material has the following properties:
Youngs modulus
: E = 28.1 106 psi (193.74 GPa)
Yield stress
: y = 39440 psi (271.93 MPa)
: = 0.2642
Poisson s ratio

[6]

[7]
[8]
[9]

[10]

V. CONCLUSIONS AND RECOMMENDATIONS

Shakedown and limit surfaces for pipe bends covering a large
range of bend characteristics are presented. Nondimensionalising the results against limit moments and
pressures for an equivalent straight pipe allows trends to be
identified with changing R/t, r/t and closing and opening
bending. Based on the results presented, the following
observation and design recommendations are proposed:
When subject to internal pressure and cyclic in-plane
bending, the normalised moment corresponding to the
reverse plasticity limit is related to h, for all three modes
of bending. The shakedown envelopes show no further
trends with h; the ratios R/r and r/t must be considered to
ascertain the behaviour.
Decreasing r/t increases the normalised cyclic bending
moment corresponding to the reverse plasticity limit.
Decreasing r/t also decreases the margin between the
shakedown envelope (for cyclic loading) and the limit
load surface (for monotonic loading). Designers must
therefore take care to ensure that a sufficient margin is
present to prevent an unexpected increase in the cyclic
load causing plastic collapse.
Increasing R/t increases the normalised cyclic bending
moment corresponding to the reverse plasticity.
Additionally, increasing R/t decrease the margin between
the shakedown and limit surfaces, and so care should be
taken to ensure sufficient margin is present.
Opening and closing bending show no difference in the
trends displayed, and all those stated above can be
equally applied to either opening or closing bending.
Direct comparison of opening and closing bending shows
that closing bending gives a larger limit load surface. The
normalised bending moment which causes reverse
plasticity is not changed for opening, closing or reversed
bending. Closing bending shows an increased ratchetting
limit over opening bending, with these curves converging
with increasing R/r.

[11]

[12]

[13]
[14]

[15]

Mushiri T is a lecturer at the University of

Zimbabwe from March 2013 to date teaching
Engineering Machine Dynamics and design and is
also a D.Eng student at the University of
Johannesburg in the field of fuzzy logic systems and
maintenance.
Contacted
at
[email protected]
/
[email protected]

Mbohwa C is currently a Full Professor of

Sustainability
Engineering
and
Engineering
Management at the University of Johannesburg,
South Africa. Contacted at [email protected]

REFERENCES
[1]

[2]
[3]

Bathe, K. J. and C. A. Almeida. A Simple and Effective Pipe Elbow

Element Pressure Stiffening Effects. Journal of Applied Mechanics,
49, 1982: 914-918.
Beskin, L. Bending of Curved Thin Tubes. Trans. ASME, 67, 1945:
A-1.
Bolt, S. E. and W. L. Greenstreet. Experimental Determinations of
Plastic Collapse Loads for Pipe Elbows. Trans. ASME, 71-PVP-37,
1971.
Cheng, D. H. and H. J. Thailer. In-Plane Bending of Curved Circular
Tubes. Trans. ASME, 68-PVP-12, 1968.
Clark, R.A. and E. Reissner. Bending of Curved Tubes. Advanced
Applied Mechanics, 2, 1951: 93-112.
Clark, R. A., T.I. Gilroy and E. Reissner. Stresses and Deformations
of Toroidal Shells of Elliptical Cross Section. Journal of Applied
Mechanics, Trans. ASME, 74 1998:
Hilsenkopf, P. , B. Boneh, and P. Sollogoub. Experimental Study of
Behavior and Functional Capability of Ferritic Steel Elbows and
Austenitic Stainless Steel Thin-Walled Elbows. International Journal
of Pressure Vessels and Piping, 33, 1988: 111-128.
Kafka, P. G. and Dunn M. B. Stiffness of Curved Circular Tubes with
Internal Pressure. Journal of Applied Mechanics, Trans. ASME, 78,
1956: 247-254.
Kitching, R. , K. Zarrabi, and M. A. Moore. Limit Moment for a
Smooth Pipe Bend Under In-Plane Bending. International Journal of
Mechanical Science, 21, 1979: 731-738.
Kussmaul, K. , H. Diem, and D. Blind. Investigations of the Plastic
Behavious of Pipe Bends. ASME PVPD, 127, 1987:55-66
Makhutov, N. A., S. V. Serikov and A. G. Kotousov. Increasing the
Design Strength of Piping Fittings. Strength of Materials, 23(4), 1991:
468-472. (Based on a paper that appeared in Russian, Problemy
Prochnosti, 4, April 1991: 77-80.)
Marcal, P. V. Elastic-Plastic Behaviour of Pipe Bends with In-Plane
Bending. Journal of Strain Analysis, 2(1), 1967: 84.

ABAQUS/Standard Ver. 5.8. Users Manual, Example Problems

Manual, and Theory Manual. Hibbit, Karlsson and Sorensen Inc.,
Providence, 1998.
Abo-Elkhier, M. Analysis of Pipe Bends Using Pipe Elbow Element.
Computers and Structures, 37(1), 1990: 9-15.
ASME Boiler and Pressure Vessel Code, Section III, Division 1.
Nuclear Power Plant Components. ASME, New York, 1992.

29