Resistance of Members To Flexural Buckling According To Eurocode 3

Master's Thesis in Mechanical Engineering
RESISTANCE OF MEMBERS TO
FLEXURAL BUCKLING
ACCORDING TO EUROCODE 3
- FOCUS ON IMPERFECTIONS
Authors: Henry Mupeta, George John,

Aliasgar Hirani
Surpervisor LNU: Min Hu
Surpervisor ALSTOM: Sara Rydström
Examinar, LNU: Andreas Linderholt
Course Code: 4MT01E
Semester: Spring 2015, 15 credits
Linnaeus University, Faculty of Technology

Abstract
This work focuses mainly on the resistance of members to flexural buckling
according to Eurocode 3. The work provides the mathematical backgrounds to the
equations and buckling curves presented in Eurocode 3. The work also, attempts to
reveal how different imperfections influence the flexural buckling resistance which is
demonstrated through Finite Element (FE) simulations.
The work presents modeling and analysis on a steel column in ABAQUS 6.14.
Linear and non-linear buckling analyses of the steel column, with the influence of
imperfections, are implemented in this work. Specifically, the imperfections
considered in this study are material plasticity, initial bow and residual stress.
The influence of initial bow imperfection of 0.1% of the length of the column
considering flexural buckling was found to be 45.28% of the Euler buckling load.
The influence of residual stresses, with a magnitude of maximum about 13% in the
flange and 35% in the web, of the yielding strength, on flexural buckling is about
31.9% of the design Euler buckling load. The combined effect of residual stress and
initial bow imperfection on flexural buckling is about 45.34% of the design Euler
buckling load.
Key words: buckling curves, buckling resistance, Eigen-value, Eurocode 3, flexural

buckling, initial bow imperfection, linear-buckling analysis, residual stress, Riks
method.
III
Acknowledgement
This research work was in collaboration between Linnaeus University and Alstom, in
Växjö, Sweden.
We thank Min Hu, who provided insights and expertise that helped greatly in this
work and for being available any time for questions and to clear some of the doubts
that arose along the way.
We also thank Marie Johansson, who provided us with some materials to get started
with the work and for the lecture on Eurocodes which formed the basis of what was
done in this work.
We would also like to show our gratitude to Sara Rydström, from Alstom, whose
calmness and detailed explanation of what was to be done was valuable.
For me, Henry, this work has been produced during my scholarship period at
Linnaeus University, thanks to a Swedish Institute scholarship.
Henry Mupeta, George John & Aliasgar Hirani
13th August, 2015.
Växjö, Sweden.
IV
Table of contents
LIST OF SYMBOLS AND NOTATIONS .................................................................................................................. VI
1. INTRODUCTION ....................................................................................................................................................... 1
1.1 BACKGROUND .......................................................................................................................................................... 1
1.2 AIM AND PURPOSE .................................................................................................................................................... 1
1.3 METHOD, MATERIAL AND LIMITATIONS .................................................................................................................... 2
2. LITERATURE REVIEW AND THEORY ................................................................................................................ 3
2.1 LITERATURE REVIEW ON BUCKLING .......................................................................................................................... 3
2.2 EULER BUCKLING THEORY ........................................................................................................................................ 5
2.3 FLEXURAL BUCKLING ACCORDING TO EUROCODE 3 ................................................................................................ 10
2.3.1 Buckling resistance of members in compression .......................................................................................... 10
2.3.2 Buckling curves ............................................................................................................................................ 11
3. BACKGROUND TO FLEXURAL BUCKLING IN EUROCODE 3. ................................................................... 16
3.1 DERIVATION OF THE SECOND ORDER MOMENT ........................................................................................................ 16
3.2 REDUCTION FACTOR ............................................................................................................................................... 19
4. CASE STUDY-INVESTIGATION OF THE INFLUENCE OF IMPERFECTIONS. ......................................... 25
4.1 FLEXURAL BUCKLING RESISTANCE ACCORDING TO EUROCODE 3 ............................................................................. 26
4.1.1 Cross section classification .......................................................................................................................... 26
4.1.2 Euler buckling load and non-dimensional slenderness ................................................................................ 28
4.1.3 Determination of buckling curve and imperfection factor ............................................................................ 28
4.1.4 Calculation of the reduction factor .............................................................................................................. 30
4.1.5 Buckling resistance ...................................................................................................................................... 30
4.1.6 Buckling effects ............................................................................................................................................ 31
4.1.7 Initial local bow imperfection, 0 ................................................................................................................ 31
4.2 FINITE ELEMENT MODELING ................................................................................................................................... 32
4.2.1 Boundary conditions .................................................................................................................................... 32
4.2.2 Material modeling ........................................................................................................................................ 33
4.2.3 True stress and true strain............................................................................................................................ 33
4.2.4 Mesh and element ......................................................................................................................................... 35
4.2.5 Linear buckling analysis .............................................................................................................................. 36
4.2.6 Non-linear buckling analysis ........................................................................................................................ 36
5. RESULTS AND ANALYSIS .................................................................................................................................... 39
5.1 LINEAR BUCKLING ANALYSIS.................................................................................................................................. 39
5.2 CONVERGENCE STUDY ON THE LINEAR BUCKLING ANALYSIS ................................................................................... 39
5.3 NON-LINEAR BUCKLING ANALYSIS ......................................................................................................................... 40
5.3.1Initial bow imperfections ............................................................................................................................... 40
5.3.2 Residual stresses .......................................................................................................................................... 43
5.3.3 Combined residual stresses and initial bow imperfections ........................................................................... 44
6. COMPARISON OF BUCKLING CURVE C IN EUROCODE 3 AND ESTABLISHED FROM
SIMULATION. .............................................................................................................................................................. 45
6.1 MODEL WITH NON-DIMENSIONAL SLENDERNESS VALUE OF 1.2 ................................................................................ 45
6.2 MODEL WITH NON-DIMENSIONAL SLENDERNESS VALUE OF 2 ...............................ERROR! BOOKMARK NOT DEFINED.
7. DISCUSSION ............................................................................................................................................................. 45
8. CONCLUSIONS ........................................................................................................................................................ 48
REFERENCES ............................................................................................... ERROR! BOOKMARK NOT DEFINED.
APPENDIX .................................................................................................................................................................... 52
V
List of symbols and notations
Cross-sectional area of a steel column

Effective area of a cross-section
Breadth of a cross-section
c Width of the flange or web
Initial bow imperfection amplitude
Modulus of elasticity
Yielding strength
Maximum of axial compressive force applied on a steel column
Height of a cross-section
Radius of gyration of a cross section
Radius of gyration along major axis
Radius of gyration along minor axis
Second moment of area
Second moment of area with respect to major axis
Second moment of area with respect to minor axis
Length of a column
Bending moment at - position along the length of a column
Design bending moment
External moment
Internal moment
Axial compressive load applied on a steel column.
, Design buckling resistance of the compression member
Elastic critical force for the relevant buckling mode
Design value of the compressive force
Design values of the resistance to normal forces
Thickness of the flange of a column along the cross-section.
Thickness of the web of a column along the cross-section.
Elastic section modulus
Predefined position along a column length
Deflection of a column at - position along the length of a column
0 Deflection at starting point of a column
Deflection at the end of a column
VI
Deflection of the column at yield point
α Imperfection factor
Partial factor for member buckling resistance
Coefficient depending on the yield strength
True strain
Intermediate factor
Critical elastic buckling stress
True stress
̅ Non-dimensional slenderness
Geometric slenderness ratio of a column
Reference relative slenderness at which
Reduction factor for relevant buckling mode
VII
1. Introduction
This project was in collaboration between Linnaeus University and Alstom.
Alstom is a global player in the field of energy and transport infrastructure.
In Växjö, Sweden, Alstom works with plants and system for cleaning of
process gases to remove environmentally harmful substances such as
nitrogen oxide, Sulphur dioxide and dust. Alstom has set the benchmark for
innovative and environmentally conscious technologies.
1.1 Background
The design of steel structures has received massive research interests over
the years. One of the main interests when designing steel structures is the
member resistance to flexural buckling. The member slenderness and the
imperfections are influential and play critical roles in the design process.
Eurocode 3, which is a standard for steel structures for European Union
(EU) member states, presents a couple of equations and buckling curves to
help in the design process. However, there is no or little background to the
equations and buckling curves presented on buckling resistance. For the
inexperienced, the code easily becomes a cookbook that can be used without
understanding the underlying problem. Moreover, more need to be done
about the influence of imperfections on the resistance of members to flexural
buckling according to Eurocode 3.
1.2 Aim and purpose

The work focused on understanding the background to the equations and
buckling curves provided in Eurocode 3, simulation and analyzing a steel
column in ABAQUS and studying how imperfections influence the member
flexural buckling resistance. The aim of the study was to reveal the
theoretical background behind the formulae and buckling curves of flexural
buckling in Eurocode 3 and show how imperfections influence member’s
flexural buckling resistance in accordance with Eurocode 3.
The purpose of this work was to present the derivation of the equations and
the establishment of the buckling curves in Eurocode 3 and to provide a FE
model of a steel column and demonstrate the influence of different
imperfections on the buckling resistance.
1
Henry Mupeta, George John, Aliasgar Hirani
1.3 Method, material and limitations
The Euler buckling theory was firstly studied to get an understanding to the
buckling phenomenon. A theoretical study on section 5 and 6 of Eurocode 3
was carried out to get the background of how imperfections have been built
into the formulae and buckling curves dealing with flexural buckling. This
involves deriving the formulae in Eurocode3. The Finite Element (FE)
model was thereafter created in ABAQUS 6.14. On this model, the linear
buckling analysis was first performed followed by a non-linear buckling
analysis using Riks method considering geometrical and residual stresses
separately as well as the combined effect.
To study the effects of imperfections, a case study of a structural support of

an electrostatic precipitator of a flue gas cleaning system was considered.
The structure consists of 6 main support columns and a system of cross-
bracings. This thesis considers the column experiencing the maximum load.
The work investigated mainly on a column having cross-section in class 3

and intermediate slenderness of 0.8. The loading case studied was pure
compression. The load eccentricity was not included in this thesis due to the
limited time. Additionally, only the pure compressive load and pined-pined
boundary conditions were considered.
2
2. Literature review and theory
2.1 Literature review on buckling

Buckling is the deformation developed normal to the applied load on a
member such as a column or beam [1]. Mainly, buckling happens in
members subjected to some compressive forces. The higher the bending
stiffness in a member, the higher the buckling resistance is. In particular, the
buckling resistance decreases when the member length is increased. Thus,
when the member is short with high bending stiffness the bucking resistance
is high. Conversely, when the member is slender, the buckling resistance is
low. Moreover, structural steel, when compared with other construction
materials, has a higher yield strength as well as a higher ultimate strength.
Therefore, compression steel members tend to be slender. Additionally,
buckling is of particular interest when dealing with such members.
Recently, Ungermann et al. investigated local-flexural buckling interaction

with plain channels based on the fact that different buckling failure modes
do occur at the same time [2]. They discovered that a structure (specimen),
that is not rigidly fixed and a load is applied on its centroid of its effective
cross section, experiences a bending moment due to eccentricity leading to a
decrease in local buckling strength. Additionally, for the rigidly fixed
structure (specimen), the bending moment due to eccentricity does not
influence the ultimate local buckling loads. When columns are loaded in
compression as well as in bending, the supplementary bending moment
prompts the flexural buckling collapse of the column [3]. Earlier on, Chan
and Gardner [4] performed column tests to investigate the response of
elliptical hollow sections to flexural buckling if the pin-ended compression
members are experiencing axial loading. The researchers [4] chose their
specimen lengths carefully to provide member slenderness spectrum, and
discovered that the elastic buckling load was below yield load when
buckling about the minor axis and the opposite was true when buckling
about the major axis. Therefore, the ultimate load for slender members
approaches the Euler buckling load. However, a thorough investigation by
Trahair and Rasmussen [5] on what effects the concentrated oblique
restraints have on flexural buckling of columns revealed that restraints may
resist rotation as well as deflection. On one hand, the restraints may be
elastic or rigid, on the other hand their concentration may be at a restraint
point or simply distributed along a portion of the length of the column.
Additionally, Adluri and Madugula [6] performed an experimental
investigation to determine the flexural buckling strength of steel angles,
focusing on residual stresses and material properties. A steel angle is steel
that has been bent lengthwise to a certain angle, usually 90 degrees, and can
be mostly used as braces to help in the reinforcement of structures or as
brackets to provide needed support. Usually, residual stresses are expressed
as a fraction of the yielding strength. However, the two may not be directly
3
related. The researchers [6] concluded that the maximum residual stress
levels in steel angles are below 25% of the yielding strength.
The Euler buckling load equation does not consider imperfections. In reality,
imperfections are always present. Two main types of imperfections are
geometrical and material (mechanical) imperfections. Recently, Lopes and
Real [7] analyzed the influences of initial geometrical and material
imperfections on the determination of ultimate load of steel class 4 cross
section elements at high temperatures. However, a lot still remains to be
done on the influence of imperfection on flexural buckling on elements that
are not subjected to fire or high temperatures. The researchers [7] argued
that the imperfections must be considered according to the expected collapse
mode. Studying the influence of imperfections on flexural buckling [2], the
researchers concluded that residual stresses influence buckling behavior.
However, the influence is small compared to the geometric imperfection. In
their analysis, they discovered that for the specimen used the flexural
buckling failure was a consistent one of the single half wave. Feng et al. [8]
evaluated the sensitivity of column failure strength with regard to initial
imperfections. Initial imperfection plays a critical role in the structural
behavior prediction. The researchers [8] showed that the ultimate strength of
short columns where the local buckling failure is predominant is influenced
significantly by the magnitude of imperfections.
Finite Element Analysis, FEA, has been widely utilized to investigate the
buckling behavior of steel columns, beams and frames. The major theme has
always been that stainless steel columns which have been used in building
massive and strong structures for a long time are subjected, in many ways, to
buckling. Shu et al. investigated a design method for stainless steel column
subjected to flexural buckling using FEA [9]. Their predictions of the finite
element model correlated with the measured imperfection. The
imperfections have a remarkable impact on the behavior and load-carrying
capacity of columns in compression [3]. Additionally, the researchers [9]
developed a finite element model that established the strength curves of
columns failing in flexural buckling. However, no discussion involving
columns with small slenderness, which may have a higher flexural buckling
resistance, was done.
In a more detailed manner, the numerical modeling of flexural buckling of

elliptical hollow section was investigated using ABAQUS [4]. The
researchers chose the four-noded, reduced integration shell elements for
their Finite Element (FE) models. In addition, a mesh convergence study
was performed to choose a uniform mesh density. The sensitivity to
imperfection as was anticipated was confirmed in the numerical results.
A literature study of the previous researches reveals that there is not much of
information about the effects of imperfections on pure flexural buckling.
4
Nevertheless, a lot of study has been conducted on flexural -torsional
buckling over the years.
2.2 Euler buckling theory

In Euler buckling, the elastic behavior of an ideal pin ended column is
considered as shown in Figure 1. In this linear elastic buckling problem, the
following assumptions are made [1]:
1. The material is linearly elastic and homogenous.
2. The column is initially perfectly straight and there are no other

geometrical imperfections.
3. No residual stresses or other inner stresses.
4. The loading is centrically applied to the column
5. The cross section of the column and its support conditions are such
that only plane buckling in one direction is relevant.
Figure 1: Euler Buckling [1]
With reference to the lateral deformation of the column, the column will
remain straight until the axial load reaches the critical buckling value,
, then the column will buckle.
Now, assuming the buckling deformation at a section having a distance

from to be as indicated in Figure 1, then the bending moment, is
expressed as:
5
∙ (1)
According to Euler-Bernoulli beam theory, the bending moment can be

expressed as a second derivative of the deflection as:
(2)
∙
where: is the modulus of elasticity.
is the second moment of area.
Combining equations (1) and (2), the differential equation governing the
deformations, is then expressed as:
(3)
∙ ∙
which is rearranged to:
(4)
∙
This differential equation (4), has the general solution as:
(5)
1 ∙ 2 ∙
Let,
(6)
Therefore, equation (5) is re-written as:
1 2 (7)
where 1 and 2 are constants.
The boundary conditions that apply for a pinned-pinned column are:
0 0; 0
So,
0 (8)
6
Equation (8) is only satisfied when:
0, , 2 , … ,
Therefore:
4
, , … ,
where is any integer or reflector of buckling mode.
The lowest value of the critical buckling load, is:
(9)

Equation (9) is known as the Euler buckling load which is a representation

of the critical load that a column can resist to elastic buckling. It can be
observed that the critical buckling load as indicated in equation (9) does not
depend on the material strength. It is dependent on the dimensions of the
column (i.e. the moment of area, , and the actual length, ) and the stiffness
of the material, . Thus, for a given material, the critical load decreases with
increased length of the column. Therefore, stocky columns will have fewer
tendencies to buckle.
It is useful to control the critical load in terms of stress rather than applied
force. The critical buckling load in terms of stress can be expressed as:
(10)
where: is critical stress.
For columns, the radius of gyration, , is defined as:
(11)
It is a significant parameter because it is considered to indicate the stiffness

of the section based on the cross section shape. Thus, the critical stress can
be reformed as:
(12)
For a more convenient and easy notation, let:
7
(13)

where: is called the column slenderness ratio.
Therefore, equation (12) is re-written as:
(14)
Equation (14), indicates that the stress of a column is inversely proportional

to the square of its slenderness ratio. Figure 2 shows the relation between
slenderness ratio and the critical buckling stress.
Figure 2: Relationship described by the Euler formula between buckling stress and column
slenderness [1]
Steel material exhibits a well-defined yielding point. Therefore, the yielding

strength, , is vital for steel columns. It is used to limit the maximum stress
that can be allowed in the column resisting to buckle. Therefore, a column
under compression can resist a maximum force given by,
∙ (15)
Thus, for a steel column, the stress at buckling cannot exceed the value of
the yield strength. The relationship of yielding strength and strain for ideal
plastic is shown in Figure 3.
8
Figure 3: Stress-strain relationship assuming ideal plastic for steel [1]
Assuming the yield point is reached at the slenderness, , called the

reference relative slenderness, as shown in Figure 4.
Figure 4: Relationship described by the Euler formula between buckling stress and column
slenderness after ideal-plasticity is introduced [1]
Therefore, , is obtained by letting the buckling stress be equal to the yielding

strength of the steel material, which is:
( 16 )
which gives:
( 17 )
∙
The ratio of the slenderness, , to the reference relative slenderness, , is

known as the non-dimensional slenderness and is given by:
9
( 18 )
̅
The failure mode change from plastic yield to elastic buckling failure occurs
when i.e. when ̅ 1.
2.3 Flexural buckling according to Eurocode 3

When considering resistance of members to flexural buckling, different
classifications of cross-section are always considered. Eurocode 3 [10], not
only presents different classes of cross-section, but also the determination of
cross-sectional resistance. Classifying cross-sections may mainly depend on
two critical factors:
1. The yield strength, , of the material, and
2. The width to thickness (c/t) ratio.
There are basically four different cross-section classes and they are defined
as [10]:
“Class 1 cross-sections are those which can form a plastic hinge with the
rotation capacity required from plastic analysis without reduction of the
resistance.
Class 2 cross-sections are those which can develop their plastic moment
resistance, but have limited rotation capacity because of local buckling.
Class3 cross-sections are those in which the stress in the extreme

compression fiber of the steel member assuming an elastic distribution of
stresses can reach the yield strength, but local buckling is liable to prevent
development of the plastic moment resistance.
Class 4 cross-sections are those in which local buckling will occur before
the attainment of yield stress in one or more parts of the cross section.”
2.3.1 Buckling resistance of members in compression

According to Eurocode 3, the ratio of the compressive force to the buckling
resistance of the member may be used to verify a member in compression
against buckling as:
1.0 ( 19 )
,
10
Where, is the design value of the compressive force, and , is the
design buckling resistance of the compression member.
Since there are about four different classes of cross-sections, it follows that
the design buckling resistance of the compression member should be taken
according to which cross-section class is under consideration. For cross-
sections class 1, 2 and 3, the design buckling resistance is given as:
( 20 )
,
and for cross-section class 4:

( 21 )
,
where is the cross-sectional area.
is the reduction factor for the relevant buckling mode.
is the partial coefficient factor for member buckling resistance.
is the effective area of the cross-section when subjected to

uniform compression.
2.3.2 Buckling curves
To compute the value of the reduction factor that appears in equation (20)
and (21) above, equation (22) is utilized.
but 1.0 ( 22 )

where:
= 0.5 1 0,2 ( 23 )
is known as the intermediate factor.
α is an imperfection factor.
The non-dimensional slenderness λ is a parameter that depends on two

properties [11]. These are the geometric and material properties of a
member. The material properties of a member are the modulus of elasticity,
and the yield strength, . Equation (24) and (25) gives the value of the
non-dimensional slenderness λ for different classes of cross-sections. For
cross-sections class 1, 2 and 3, it is given by:
( 24 )
=
11
and for cross-section class 4:
( 25 )
=
As mentioned, is the elastic critical force for the relevant buckling

mode based on the gross cross sectional properties, which is the well-known
Euler buckling load. It is a critical load without considering imperfections.
The imperfection factor α that appears in equation (23) depends on [12]:
1. The cross-section shape of the column under consideration
2. The process of fabrication used
3. The direction in which buckling occurs i.e. the weak or strong

axis plane of buckling
4. The yielding strength.
The imperfection factor must be chosen according to Table 1 and
12
Table 2
Table 1: Imperfection factor for buckling curves [12].
Buckling curve
Imperfection factor 0.13 0.21 0.34 0.49 0.76
Once has been carefully selected, it’s easier to get the reduction factor
represented by equation (22).
13
Table 2. Selection of a buckling curve for a cross-section [10]
When using Eurocode 3, the column design procedures for flexural buckling
may usually consist of six main steps [11]:
1. Trial section selection. This means that the cross-section class

to be considered must be chosen.
2. Determining the buckling length of the column
14
3. Calculate the non-dimensional slenderness,
4. Pick the suitable buckling curve ( , , , , ) and the

equivalent imperfection factor
5. Get the buckling reduction factor,
6. Compute the design buckling resistance using equation (20) or

(21)
For greater values of the non-dimensional slenderness, the influence of

imperfections is small. Therefore, the resistance approaches the Euler
critical value as the slenderness increase. See Figure 5 which shows the
buckling curves.
Figure 5: Buckling curves in Eurocode 3 [10]
The difference between the curves in Figure 5 reflects the influence of the
below factors on the buckling of the column:
1. Manufacturing processes,
2. The cross-sectional dimensions, and,
3. Weak axis about which the buckling occurs.
15
3. Background to flexural buckling in Eurocode 3.
Eurocode 3 presents sets of formulae according to each buckling mode
under consideration. Buckling can be flexural, torsional, torsional-flexural or
lateral-torsional buckling. This work focuses on the flexural buckling.
Flexural buckling is the mode of buckling in which the member deflects
purely along the strong or the weak axis of the column.
3.1 Derivation of the second order moment

Euler buckling theory considers an ideal straight column which, in practice,
cannot be achieved because real columns are usually never perfectly
straight. Geometrical imperfection as initial bow and load eccentricity along
with mechanical imperfection as yielding strength and residual stresses, are
present. This will affect the behavior of the columns and influence their
buckling resistance.
Considering a member under axial compression, , with initial geometrical

imperfections [12] as shown in Figure 6. The initial bow imperfection is
assumed to be sinusoidal having the maximum value at mid-span.
a) b)
Figure 6: Simply supported member loaded with a normal force N and initial bow imperfection
[13]
16
Figure 6 shows a column with initial bow imperfection, where is the
initial deflection in the middle section. The initial deflection together with
the force means that the column is subjected to an external moment . This
moment is given by [13]:
Where:
is the applied axial force
is the initial deflection in the mid-section of the column
is the external moment created by the initial deflection and force .
The bending moment will cause further deflection in mid-length of the

column consequently increasing the external bending moment which can be
expressed as:
(26)
where is the additional deflection in the mid section caused by the axial force on
the initially bowed column.
According to equation (26), the increase in deflection caused due to the

load would cause an increase in external moment . This increase in
moment would eventually cause a further increase in deflection .
When the deflection is increasing, the column resists this through an internal
resistance against bending deformations. This internal moment can be
expressed with the static beam theory as:
(27)
Where:
is the moment caused by the resistance of the column to an external

force.
is the modulus of elasticity
is the moment of inertia
is the deflection
is the coordinate in the length direction of the column
By assuming that the initial deflection and additional deflection

are sinusoidal:
17
. (28)
∙
.
= ∙ (29)
The internal moment caused by the increase in deflection can be

described as:
. ( 30 )
Therefore, the internal moment in the mid-section of the column can then be
described as:
( 31 )
For the internal resisting moment to be able to stop the increase in

deflection, caused by the external moment, it must be at least as large as the
external moment. i.e. rate of change of internal resisting moment must be
greater than the rate of change of the external moment.
The rate of change of the external moment is:
(32)
If the rate of increasing external moment, is higher than or equal to the

rate of increasing internal moment the column will buckle i.e. if
.
However, if the load acting on the column is smaller, the column would
remain stable, i.e. when ≤ . This means that for any given column,
with certain geometry and material, there is a critical value on the force.
(33)
where is the Euler buckling load.
This will mean that for a normal force lower than the critical buckling load,
it is possible to find equilibrium. This equilibrium will occur at a
deformation where the internal moment is equal to the external
moment.
(34)
⇒
18
Simplifying the above expression:
(35)
⇒
Thus the total deflection at equilibrium position is :
(36)
⇒
Equation (36) is called as second order deflection.
The deflection gives the following external moment:
(37)

Equation (37) is called as second order moment.
3.2 Reduction factor
The derivation of the reduction factor, , for a simple case can be made from
second-order analysis done above. Consider the illustration as shown in
Figure 7 of a simply supported member under pure axial compression with
an initial transverse deflection [14].
Figure 7: Column buckling in pure axial compression [14]
Assuming a sinusoidal deflection, the initial geometrical imperfection can be

expressed as:
(38)
When an axial force is applied to the column, the additional deflection

associated to instability can be expressed as:
(39)
19
The classical buckling equilibrium equation for this case then is:
(40)
0
Inserting equations (38) and (39) into equation (40) and evaluating:
(41)
where: is the elastic buckling load i.e. the Euler load.
Thus, equation (39) is re-written as:
(42)
At mid span, the total deflection is expressed as:
(43)
The member cross-section’s resistance criterion at mid-span including

second order effects is then expressed as:
1 ∙ (44)
∙ 1
1
where: is the bending moment.
Let,
(45)
where: is the design values of resistance to normal forces.
The buckling resistance, , , is expressed as:
, ∙ (46)
where: is called the reduction factor.
, is the design buckling resistance
From Equation (18), the relation for Non-dimensional slenderness can be

rewritten as:
20
(47)
̅
The bending moment is the product of the yield force and elastic modulus,
, and is expressed as:
∙ (48)
Where: is called the design bending moment.
At buckling, the maximum applied axial force reaches the actual buckling
resistance, thus:
, ∙ (49)
Inserting equations (45), (46), (47), (48) and (49) into equation (44), then:
∙
∙ 1
∙
.
∙ 1
∙
1 ∙ (50)
∙ 1
1 ∙ ̅
Solving for in equation (50)
Let,
∙ (51)
ƞ
Thus,
∙ƞ 1
∙
After rearrangement, a second order of polynomial is obtained as:
∙ ̅ 1 ƞ ̅ 1 0 (52)
Solving the quadratic equation (52) will give out two solutions. The lower
value is taken to be the reduction factor, therefore:
(53)
1 ƞ ̅ 1 ƞ ̅ 4 ̅
2 ̅
21
Let,
0.5 1 ƞ ̅
Therefore:
̅ (54)
̅
Pre-multiplying equation (55) with Φ Φ λ , thus:

1 (55)
̅
Equation (55) is the reduction factor which is identical to equation (22)

given in Eurocode 3.
The basis to the buckling curves in Eurocode 3 as shown in Figure 8 was

based on an extensive measurement campaign in 1973 on columns to find
out their buckling behaviors [15].
More than 1000 buckling tests on I, H, T, U, circular and square hollow

sections with slenderness between 55 to 160 were studied. During these
experiments, the imperfections taken into account were the residual stresses
(considered according to the manufacturing process and the cross-section)
and an initial out of straightness of the order 1/1000th the length of the
column at the centre point.
These initial out of straightness was calibrated to reproduce the effects of all
the other imperfections found in a column. This is also rightly called the
'Equivalent initial deformed configuration'.
The values of all the experiments plotted were compared to the Euler’s
buckling curve as shown in Figure 8.
22
Figure 8: Results of experimental tests in real members [15].
Figure 9. Columns classified based on slenderness [16].
According to Figure 9, the columns are classified into three major categories
as follows:
a) Columns with large slenderness
These are columns that lie towards the right of the point of inflexion. The
buckling loads for these columns are similar to the Euler buckling load ( ).
Imperfections do not play much of a role in the buckling of these columns as
the buckling of these columns occur in the elastic range.
23
b) Columns with medium slenderness
These are columns that deviate the maximum from the Euler's theory.
At buckling, some of the fibers have already reached yield strength. Hence
the effective area that resists buckling at a point before buckling is less than
the actual cross-sectional area of the column.
Hence, the presence of imperfections greatly affects the load bearing

capacity of these columns.
c) Columns with low slenderness
These columns are also called as stocky columns. Its buckling resistances
are very high as they are short and its load bearing capacity is mainly
governed by the yield strength.
24
4. Case Study-Investigation of the influence of
imperfections.
Alstom in Växjö, Sweden, deals with air pollution control equipment. In
Växjö, Alstom provides service for Växjö Energi AB (VEAB), which is a
thermal power plant. Alstom does this by maintaining the flue gas lines at
VEAB. One of the key pollution control equipments in the flue gas line is
the Electro-Static precipitator (ESP). This equipment separates the
suspended particles in the exhaust gases. It typically collects 99.9 % of the
suspended particles. The size of the equipment can be visualized to be the
size of a 10 or 12 storey building.
In this case study, a steel structure that supports the ESP was considered.
Figure 10 shows the support structure in ABAQUS 6.14. One of the
columns of this structure was used as a specimen to study the effect of
imperfections on the buckling resistance capacity. The structure contains 6
main columns and a network of cross-bracing which are connected together
by pin joints. The main columns were standard hot-rolled with a profile of
HEA 300 and the material is steel S355J2.
Figure 10: The model of the complete structure.
25
Table 3 shows material data and profiles.
Table 3: Material data and profiles
Material Density, Modulus of Poisson’s Yield Strength [MPa] Ultimate Strength

[kg/ ] Elasticity, ratio, [MPa]
[MPa]
S355J2 7850 210000 0.30 355 ( 16mm) 470 ( 16mm)
345 16 40 470 16 40
The material profiles in Table 3 were used in the modeling of a column in

ABAQUS.
The length of the main columns was 9.169 meters. However, the columns
have cross-bracings at 4.597 meters from the bottom indicated by L in
Figure 10. To facilitate the study of the load on the column alone and
considering the effect of the cross bracing, the 4.597 meters length was
considered for the study.
4.1 Flexural buckling resistance according to Eurocode 3

Critical in the design of steel structures are the design loads that the structure
will be subjected to. Table 4 shows the loads that the columns A, B and C
will be subjected to. These loads form the basis for the calculation of the
design loads for each column.
Table 4: Characteristic loads acting on the support points on the structure
Supp. points Dead load [KN] Access liveload [KN] Dust load [KN]
D1, D2 364 38 378
J1, J2 375 63 624
K1, K2 239 38 312
4.1.1 Cross section classification

The coefficient of the yield strength of the cross section under consideration
is given as,
( 56 )
being the yield strength. It should be noted here that, in Eurocode 3, the
definition of employs the base value of 235 / . This is because
26
grade S235 steel is highly regarded as the normal grade throughout Europe
[17]. Thus, 0.8136.
To get the class of the cross section, the class of the flange and of the web
must be determined independently because they have different thicknesses.
Figure 11 shows the cross-sectional dimensions for HEA 300 columns.
Figure 11: Cross-section dimensions of a HEA 300 column. All dimensions in mm [18]
where, is the thickness of the web and is the thickness of the flange.
Thus, for the flange, the width, c, is given by [17]:

2 ( 57 )
2
300 8.5 2 27
c 118.75
2
118.75
8.4821
14
The limit for cross-section class 2 flange: 10 8.1362 [17]. Therefore, the
flange is class 3 since 8.1362.
For the web, the width is given by:

2 2 ( 58 )
c 290 2 14 2 27 208
27
208
24.4706
8.5
The limit for cross-section class 1 web: 33 26.8493 [17]. Thus, the web
is class 1.
The overall cross-section classification is therefore class 3.
4.1.2 Euler buckling load and non-dimensional slenderness

The available data is:
210
6.31 10
4.597
0.0112
. . .
Therefore, the Euler buckling load,
.
6.1887 10
The non-dimensional slenderness is given as:
. . .
̅ √0.6453 0.8033
.
4.1.3 Determination of buckling curve and imperfection factor
The Design loads, , for the columns are calculated as follows:
1.35 1.5 1.05

( 59 )
where 1.35, 1.5 and 1.05 in equation (59) are recommended set of partial
safety factors provided by Eurocode for transient design situations where
there is a risk of loss of static equilibrium [19].
for columns , , 1.35 364 1.5 378 1.05 38

1098.3
for columns , , 1.35 375 1.5 624 1.05 63

1508.4
28
for columns , , 1.35 239 1.5 312 1.05 38
830.55
To help in selecting the buckling curve to be used, the detailed data of the
cross-section as given in Table 5 [20] can be used.
Table 5: Cross-section data
⁄ Radius of gyration [ ] Moment of inertia [ ]

0.9667 127.4 18265 10
74.9 6310 10
From Figure 11 ⁄ 1.2, and since flexural buckling is about the weak axis,
and from the moment of inertia above, Z-axis is the weak axis, and
100 , the buckling curve c is considered, see
29
Table 2. The imperfection factor for this buckling curve, 0.49, from
Table 1.
4.1.4 Calculation of the reduction factor

The available data is:
0.49
Thus, the intermediate factor is given as:
0.5 1 ̅ 0.2 ̅
0.5 1 0.49 0.8033 0.2 0.6453
0.9705
and the reduction factor is:

. √ . .
0.66
and the buckling resistance is:

. . . . .
, 2.6359 10
where: is the partial safety factor
4.1.5 Buckling resistance

A compression member, according to Eurocode 3, should be verified against
buckling using equation (19):
1.0
,
The column experiencing the highest load must be checked against this
criterion. Columns , have the highest load, thus:
.
0.5722
, .
Hence, the criterion is fulfilled. This means that the other columns
experiencing the lowest load than columns , will also fulfill the criteria.
30
4.1.6 Buckling effects
According to Eurocode 3, the buckling effects may be ignored for:
i. ̅ 0.2
ii. 0.04
The columns are HEA300 and the slenderness is 0.8033. Since 0.8033
0.2, it means that the effect of buckling cannot be ignored. However, this
check is not enough because the columns are subjected to different values of
the design compression load, . The column experiencing the lowest load
must be checked against the second equation above. Therefore, for columns
, ,
.
0.1342
.
Since 0.1342 0.04, the buckling effects cannot be ignored. This means
that the columns with the higher loads than columns , , the buckling
effects cannot be ignored either.
4.1.7 Initial local bow imperfection,

The values of the initial bow imperfection for each buckling curve are given
in Table 6. These are reference values as provided in Eurocode 3. These
values take into account all kinds of imperfections.
Table 6: Design value of initial local bow imperfection [10]
Buckling curves Elastic analysis Plastic analysis
1 1
350 300
1 1
300 250
1 1
250 200
1 1
200 150
1 1
150 100
From Table 6, the column in study for which buckling curve c is chosen,
plastic analysis is considered,
31
thus, 4597 150 30.647
Plastic analysis is the analysis in which the inelastic material is studied

beyond the elastic limit. On the other hand, elastic analysis is the analysis in
which the material is studied within the region where the stress is
proportional to the strain i.e. the elastic region.
4.2 Finite Element modeling

ABAQUS 6.14 software was utilized for the finite element modeling of the
column. Only one column was selected for the analysis. The software was
used to perform the buckling and post-buckling analyses of the HEA 300
steel column. The column was modeled as a 3D extruded solid. The detailed
modeling procedures are presented in appendix 1 of this thesis.
In the case of any structure that is being analyzed, the member experiencing
the highest load would define the integrity of the structure. Hence, in this
case, the columns 1 2 as shown in Figure 10 would be the ones that
buckle first as it experiences the highest load. Therefore, one of columns B
was chosen for the analysis.
4.2.1 Boundary conditions

To be able to get a clear and meaningful comparison between numerical and
experimental results, it is important to ensure that the boundary conditions
are adequately and properly modeled [23].The column was modeled as a
pined-pined support condition. Therefore, the boundary conditions
considered for the analysis were that the column was pinned on both ends.
The pined-pined boundary conditions in this work were modeled in such a
way that the bottom and top ends were free to rotate about the axes.
However, no translation was allowed in any direction at the bottom end of
the model. At the top end, where the load was applied, no translation was
allowed in the x and y directions. Translation was allowed in the z direction
because this was the axis along which the load was applied. The load, for
both the linear buckling and non-linear buckling analyses, was applied along
the z-axis as shown in Figure 12 by the red arrow. The surface where the
load was applied was considered the top surface.
32
Figure 12: Column indicating the direction of applied load
Before the application of the load, both the end faces of the columns were
made rigid to prevent an indent caused by a concentrated point load on the
faces. In order to achieve this, two reference points were created at the
center of the cross-sections at both ends. Both surfaces were kinematically
coupled to the reference points in all six degrees-of-freedom.
4.2.2 Material modeling

The HEA300 steel column in this work was modeled as homogeneous and
isotropic. This column was modeled considering linear elasticity and
elasticity-plasticity. The Young’s modulus, , and the Poisson’s ratio, ,
gives the full characteristics of a linear elastic behavior [23]. Implementation
of the elastic-plastic behavior in ABAQUS is easy and straightforward.
Everything is incorporated in the ABAQUS material behavior library.
Implementation simply involves providing the true stress and true strain in
plasticity properties in the material library [24] [25].
4.2.3 True stress and true strain

Using the Young’s modulus, yield strength and ultimate strength, the
material plastic behavior of the steel column were determined. In ABAQUS
software, for the non-linear analysis, presented in section 4.2.5, the plastic
characteristics of the material need to be taken into consideration.
The true strain or the logarithmic strain is a mathematical model which

describes the plastic behavior of the metals which accounts for differences in
the compression and tensile behavior independent of the structure's
geometry or nature of the load applied. [24]
33
In ABAQUS, when plasticity of the material data is defined, true stress and
true strain should be used. These are the values that ABAQUS need to
correctly interpret the data. The nominal stress and strain values are often
supplied by the material test data. However, the material plasticity data must
be converted from the nominal stress and nominal strain values to the true
stress and true strain values respectively [24] [25].
The plastic properties are computed using the following relationships [25]:
1 ( 60 )
1 ( 61 )
where is nominal strain
is nominal stress
σtrue is the true stress
εtrue is the true strain
Figure 13 shows the nominal stress strain curves, for an elastic material and
for the elastic plastic material, which were utilized to calculate the values of
the true stresses and true strains
a) b)
Figure 13: Nominal stress-strain curves, where a) is for purely elastic material and b) is for elastic-
plastic material [25]
To calculate the strain the values of the yield strength and ultimate strength
from Table 3 are used, thus:
From Figure 13a, since the slope is [25], for the elastic region, thus,
355 10 0 0
Therefore: 0.0016
34
For the plastic region, the slope is ⁄100. See Figure 13b. Thus,
470 10 355 10 ⁄100

Therefore: 0.0016 0.0565
Utilizing equation (59) and (60), the true yield strength is:
σ σ 1 ε
σ 355 10 1 0.0016 355.6001
and the true ultimate strength is:
σ σ 1 ε
σ 470 10 1 0.0565 496.5644
Table 7 show the summary of the values calculated.

Table 7: Stress and Strain values
Nominal Strain Engineering True Strain True Stress Strain for

Stress [Mpa] [Mpa] non linear
analysis
1 0.0016 355 0.0016 355.6001 0
2 0.0564 470 0.0549 496.5644 0.0525
The values in the last column of Table 7 were used for the non-linear
analysis in this thesis.
4.2.4 Mesh and element

In finite element analysis, to obtain accurate results, the mesh refinement
levels must be good enough. In this work, the linear hexahedral element was
used. The element size mesh along the length was taken to be 50mm while
along the cross section, the element size was fine-tuned in proximity to
result in equal and uniform elements by seeding the edges. Figure 14 shows
the mesh of the entire column and an enlarged view of the mesh on and
around the cross section.
35
Figure 14: Mesh generation on the model
4.2.5 Linear buckling analysis

The linear buckling analysis requires solving an Eigen-value problem
defined by the column geometric matrices and elastic stiffness. ABAQUS
obtains the solution through the sub-space iteration method. In this work, the
linear buckling analysis was performed to obtain the mode shapes and the
corresponding Eigen-values. The analysis was done by selecting the linear
perturbation buckle step in ABAQUS and selecting the number of modes. 5
modes were chosen in this study. Flexural buckling occurred as the third
mode shape. The displacements of the modes were saved in a node file by
modifying the Keywords of the model. The Eigen-value and mode shape of
the third mode were used as the reference load and initial bow imperfection
in the non-linear buckling analysis will be presented in section 4.2.6.
4.2.6 Non-linear buckling analysis

After the linear buckling analysis was performed the non-linear buckling
analysis, also known as the post-buckling analysis, was conducted to
determine the ultimate loads. In this work, the non-linear buckling analysis
was performed by employing the Riks method, also known as the arc-length
control strategy. ABAQUS implements this with predefined increment and
tolerance parameter. In this work, the number of increments was set to be
30.
36
In this analysis, the mode shape of the elastic flexural buckling shape (i.e.
the third mode from linear buckling analysis) was entered into the model as
the initial bow imperfections by utilizing the edit Keywords of the model.
Four different magnitudes of initial bow imperfections were studied. Two of
the initial bow imperfections studied was 1000 and 1500 which
provide the maximum and minimum tolerance values according to [26]. The
other initial bow imperfection studied was arbitrarily chosen as 1200.
While, the equivalent initial bow imperfection of 150 provided by

Eurocode 3 for curve c was studied as well to verify how well the modified
initial bow represents all kinds of initial imperfections found in real
columns.
Besides, an initial bow imperfection of 1.5 1000 observed to be the

maximum value, allowed by this manufacturer of steel in Sweden, Broderna
Edstrand AB [27] was also considered. This initial bow imperfection would
represent the worst case scenario.
4.2.6.1 Initial bow imperfection
The initial bow imperfection can be incorporated in the model either

manually through ABAQUS script or automatically by defining a linear
combination of buckling mode shapes by editing the Keywords of the model
[23]. These mode shapes are from the preliminary linear buckling analysis.
In this work, the initial bow imperfections were included by editing the
Keywords. Mode 3 was the critical buckling shape because this is the mode
at which the column first experiences flexural buckling and has the most
significant influence on the buckling load.
4.2.6.2 Residual stresses
Residual stresses exist in structural steel members. These residual stresses

are induced in the steel member during the process of manufacturing,
fabrication or refinement by the non-uniform temperature distribution.
Usually, residual stresses are expressed as a fraction of the yield strength.
However, the two may not be directly related [6]. Nevertheless, the presence
of residual stress may impair significantly the stiffness of the compression
member thus influence the buckling capacity and often shorten fatigue life
of steel members under periodic or dynamic load [28].
In this work, residual stresses were incorporated in the FE model using the
reference values shown in Figure 15.
37
Figure 15: Distribution of residual stresses employed in this study [29]
The residual stresses were incorporated in the model to study how much
effect they have on buckling and buckling resistance.
Introducing stresses into the ABAQUS model, sets of elements to assign the
residual stresses values were created. Then the values for the residual
stresses were assigned using the 'Predefined Field' option found in the model
tree in ABAQUS. Appendix 1 explains in details this procedure.
38
5. Results and analysis
This chapter shows the results that were obtained in the FE simulation and
the analysis based on them.
5.1 Linear buckling analysis
The buckling load in the linear buckling analysis was found to be 6.1342
10 and was observed at mode 3 at which the flexural buckling occurs.
Figure 16 shows mode 3 of the buckled mode shape of the column.
Figure 16: The flexural buckling mode shape of the steel column obtained from the linear buckling
analysis
The buckling load of 6.1342 10 obtained in the linear buckling

analysis is close to the theoretical one of 6.1887 10 calculated in
section 4.1.2 in this thesis.
5.2 Convergence study on the linear buckling analysis

A convergence study based on linear buckling was conducted in this work.
The study was conducted on linear element type mesh with element size of
12.5mm, 25mm and 50mm along the length.
The buckling load for a 12.5mm element size linear element type mesh was
6.1328 10 , the buckling load for a 25mm element size linear element
type mesh was 6.1330 10 , while the buckling load for a 50mm
element size linear element type mesh was 6.1342 10 .
Figure 17 shows the results from the convergence study.
39
Convergence
6,1344
Elastic buckling load[MN]

6,1342
6,134
6,1338
6,1336
6,1334
6,1332
6,133
6,1328
6,1326
0 10 20 30 40 50 60
Element size [mm]
Figure 17: Buckling loads for different element size
From Figure 17, it can be clearly seen that the results are converging.
Though the results from each element size would be similar, the 12.5
element size would be better. However, in this thesis, the 50mm element
size was used due to computational time.
5.3 Non-linear buckling analysis
5.3.1Initial bow imperfections

From the non-linear buckling analysis, the buckling resistances, for the three
initial bow imperfections studied, were:
i.
Buckling resistance, ,
Figure 18 shows the Load Proportionality Factor (LPF) curve which was
obtained for this initial bow imperfection.
40
Figure 18: LPF curve for column with imperfection of L/1000
Therefore, the buckling resistance, , 0.5472 6.1342 10
3.3566 10
Similarly, for each and every initial bow presented below, the LPF
graphs were generated and the highest point on the curve was selected
and multiplied with the buckling load to get the buckling resistance.
ii.
For this initial bow imperfection
Buckling resistance, , 0.5595 6.1342 10
3.4321 10
iii.
For this initial bow imperfection,
3.5179 10
41
In order to evaluate the accuracy of the model that has been used for the
analysis, the initial bow imperfection ratio of has been studied:
iv.
2.2580 10
Since the initial bow imperfection of 1.5 1000 represents the worst case
scenario of a column, it's buckling resistance has also been analyzed.
.
v.
3.1658 10
Figure 19 shows the load-displacement curves for all the five initial bow
imperfections that were studied.
Load-displacement curves
4
Load [MN]
3
L/150
2 1.5L/1000
1 L/1000
L/1200
0
0 50 100 150 L/1500
Displacement [mm]
Figure 19: Load-displacement curves for initial bow imperfections
42
5.3.2 Residual stresses
Figure 20 shows the residual stress distribution on the flange input in
ABAQUS.
Figure 20: Input of the residual stress distribution on the flange of the model
Figure 21 shows the residual stress distribution on the web
Figure 21: Input of the residual stress distribution on the web of the model
The presence of residual stresses will result in early partial yielding in the
cross-section. The partial yielding will decrease the effective area on the
cross section and therefore reduce the buckling resistance of the column.
Considering only residual stress,
4.1779 10
Inferring from the results obtained, the effect of residual stress alone is about
32% of the Euler buckling load.
43
5.3.3 Combined residual stresses and initial bow imperfections
Since gives the least buckling resistance concerning hot rolled sections,
it is picked to study the combined effects of the initial bow imperfections
and the residual stresses.
3.3529 10
Taking both the initial bow imperfection and residual stress into account, the
buckling resistance of 3.3529 10 from the simulation, as calculated
above, is significantly higher than 2.636 10 that was theoretically
calculated according to Eurocode 3. Thus, the effect of the combined
residual stresses and initial bow imperfection is about 45.3% of the Euler
buckling load.
.
Since represents the worst-case scenario of any column supplied by
Broderna Edstrand AB, it is also considered to study the combined effect
3.1425 10
Thus, it is observed that in the worst case, a column would buckle at 51.2%
of the Euler buckling load considering the effects of the both the initial bow
imperfections and residual stresses.
44
6. Comparison of buckling curve c in Eurocode 3 and
established from simulation.
To establish the buckling curve from simulation, two separate models were
simulated with different non-dimensional slenderness ratios. Following the
same procedure, firstly the linear buckling analysis was done on each model
to get the buckling load. Thereafter, the non-linear buckling analysis was
conducted incorporating both the initial bow and residual stresses.
6.1 Model with non-dimensional slenderness value of 1.2

The length of the column for this value of non-dimensional slenderness was
6.867 m. 1000 initial bow imperfection is used because it has been
selected for the analysis. The non-linear buckling analysis is then performed
to get the LPF curves in the same way as described in section 5.3.
Thus, buckling resistance, , 0.8020 2.76236 10
2.2154 10
and the reduction factor is calculated by rearranging equation (20) as,
, 2.2154 10

0.55
∙ 0.01125 355 10
6.2 Model with non-dimensional slenderness value of 2

The length of the column for this value of non-dimensional slenderness was
11.445 m. The same ratio of 1000 initial bow imperfection was used
here as well and the analysis done in the same manner as in section 6.1.
Thus, buckling resistance, , 0.9148 9.96618 10
9.1171 10
and the reduction factor is:
, 9.1171 10

0.23
∙ 0.01125 355 10
From Eurocode 3, the non-dimensional slenderness of 0.2 and 3 corresponds to

the reduction factor of 1 and 0.1 respectively. Combining the values from the
simulations and the calculated values from section 4.1, the curve shown in
Figure 22 was generated.
45
Buckling Curves
1,2
1
reduction factor, 0,8
0,6
0,4 FE Simulations
0,2 Eurocode 3
0
0 1 2 3 4
non-dimensional slenderness, ̅
Figure 22: Buckling curve c from simulation and Eurocode
The buckling curve shown in Figure 22 gives higher reduction factor and shows
almost similar trend in comparison to the buckling curve c in Figure 5 from
Eurocode 3.
46
7. Discussion
The boundary conditions of the structure must be well defined. Wrong
boundary conditions will give wrong results and will greatly affect the
analysis. In this work, a pinned-pinned boundary condition was used. This
means the boundary conditions used in this work were close to reality.
Meshing plays a critical role in the finite element modeling. Efforts were
made to pick a mesh that gave equal and uniform elements, see Figure 14.
The value of the buckling load obtained in the linear buckling analysis was
closer to the one calculated according to Eurocode 3. Moreover, the
buckling resistance obtained from the analysis considering the ratio of
150 is close to the value from the calculations according to Eurocode
3.This confirms that the meshing done for this study was accurate enough.
The buckling resistance obtained from the Riks method considering the
residual stress alone was much higher than the one calculated according to
Eurocode 3. The effect of residual stress alone is about 32% of the Euler
buckling load. Hence, it is observed to be higher than the 25% mentioned for
the steel angles as per the literature study. This difference could be due to
the one or a combination of the following reasons concerning this thesis:
 Difference in the cross-section and its area.
 Limitation in accuracy of the result due to the 50mm element size

considered in this thesis.
 The reference value for the residual stresses considered in this study
could be on the higher side.
Similarly, the non-linear buckling analysis, considering only initial bow

imperfection was higher than the one calculated using the formula from
Eurocode 3. This is because Eurocode 3 formula considers all imperfections.
Nevertheless, the influence of the combined imperfections (residual stress
and initial bow) was about 45.3% of the Euler buckling load. While the
value depicted by the worst-case scenario for the combined effect is 51.2%
of the Euler buckling load.
47
8. Conclusions
The linear Eigen-value buckling analysis is used to determine the critical
buckling load. Elastic-plastic behavior and imperfections are not considered
in this analysis. However, imperfections are always present in structures.
Thus, post-buckling analysis, also known as non-linear buckling analysis,
must be performed always. The elastic-plastic material and imperfections are
considered in the non-linear buckling analysis.
From the results obtained in this work, the following conclusions are drawn:
The buckling formulae in Eurocode 3 are based on second order

moment and extensive experimentations on real columns.
Imperfections are well handled by the formulae and buckling curves

in Eurocode 3.
Post-buckling analysis utilizing the Riks method in ABAQUS is an

efficient way to evaluate the effects due to initial imperfections on a
structure.
The influence of residual stresses, with a magnitude of maximum

about 13% in the flange and 35% in the web, of the yielding strength,
on flexural buckling is about 31.90% of the design Euler buckling
load.
The effect of initial bow imperfection with a magnitude of 0.1% of

the column length on flexural buckling is about 45.28% of the Euler
buckling load.
The combined effect of residual stress (magnitude as mentioned

above) and initial bow imperfection (0.1% of the column length) on
flexural buckling is about 45.34% of the design Euler buckling load.
While considering the worst-case scenario, the effect of the initial

bow of 0.15% of the length of the column and the combined effect
were found to be 48.39% and 48.77% of the Euler buckling load
respectively.
Residual stresses influence buckling behavior. However, the

influence is small compared to that of the initial bow imperfection.
48
Reference
[1] [Online]. Available: http://www.steel-
insdag.org/TeachingMaterial/Chapter6.pdf. [Använd 29 March 2015].
[2] D. Ungermann, S. Lubke och B. Brune, ”Test and Design approach for plain
channels in local and coupled local-flexural buckling based on Eurocode 3,”
Thin-walled structures, vol. 81, pp. 108-120, 2014.
[3] L. Palvovcic, B. Froschmeier, U. Kuhlmann och D. Beg, ”Finite Element
Simulation of slender thin-walled box columns by implementing real initial
conditions,” Advances in Engineering Software, vol. 44, nr 1, pp. 63-74, 2012.
[4] T. M. Chan och L. Gardner, ”Flexural Buckling of Elliptical Hollow Section
Columns,” Journal of Structural Engineering-ASCE, vol. 135, nr 5, pp. 546-
557, 2009.
[5] N. S. Trahair och K. J. R. Rasmussen, ”Finite Element analysis of the flexural
buckling of columns with oblique restraints,” Journal of structural
Engineering-ASCE, vol. 131, nr 3, pp. 481-487, 2005.
[6] S. M. Adluri och M. K. S. Madugula, ”Flexural Buckling of Steel
Angles:Experimental Investigation,” Journal of Structural Engineering-ASCE,
vol. 122, nr 3, pp. 309-317, 1996.
[7] N. Lopes och V. P. Real, ”Class 4 Stainless Steel I beam subjected to fire,”
Thin-Walled Structures, vol. 83, nr Special Issue-SI, pp. 137-146, 2014.
[8] M. Feng, Y. C. Wang och J. M. Davies, ”A numerical imperfection sensitivity
study of cold-formed thin-walled tubular steel columns at uniform elevated
temperatures,” Thin-Walled Structures, vol. 42, nr 4, pp. 533-555, 2004.
[9] G. Shu, B. Zheng och L. Xin, ”A new design method for stainless steel
columns subjected to flexural buckling,” Thin-Walled Structures, vol. 83, nr
Special Issue-SI, pp. 43-51, 2014.
[10] EN 1993-1-1. Eurocode 3: Design of Steel Structures-part 1-1: General rules
and rules for buildings., Brussels: European Committee for Standardization
(CEN), 2005.
[11] L. Gardner, ”The Steel Construction Institute,” 2011. [Online]. Available:
http://www.steelconstruction.info/index.php?title=Special:ImagePage&t=Sci+
p360.pdf . [Använd 20 February 2015].
[12] Rules for Member Stability in EN 1993-1-1: Background Documentation and
Design Guidelines, Brussels: ECCS Technical Committee 8-Stability, 2006.
[13] M. Al-Emrani, B. Enstrom, M. Johansson och P. Johansson, Bärande
Konstruktioner, Part 2, Gothernburg: Chalmers University, 2013.
[14] N. Boissonnade, J. P. Jaspart, J. P. Muzeau och M. Villette, ”Improvement of
the interaction formulae for beam columns in Eurocode 3,” Computers and
Structures, vol. 80, nr 27-30, pp. 2375-2385, 2002.
[15] L. S. da Silva, R. Simões och H. Gervásio, Design of Steel Structures:
49
Eurocode 3: Design of steel structures, Part 1-1 - General rules and rules for
buildings, First Edition, JAN 2012.
[16] [Online]. Available: http://www.fgg.uni-
lj.si/~/pmoze/ESDEP/master/wg07/l0510.htm. [Använd 12 May 2015].
[17] L. Gardner och D. A. Nethercot, Designer's Guide to Eurocode 3: Design of
steel structures general rules and rules for buildings, London: Thomas Telfolrd
Publishing, 2005.
[18] [Online]. Available:
http://www.nssmc.com/en/product/stainless/sushshaped_steel_size.html.
[Använd 23 April 2015].
[19] EN 1990: Eurocode-Basis of Structural design, Brussels: European Committee
for Standardization (CEN), 2002.
http://www.constructalia.com/repository/Products/BeamsSections/SectionRang
eFR_EN_DE/HE.pdf. [Använd 11 May 2015].
[21] S. Kumar och S. A. Kumar. [Online]. Available:
http://nptel.ac.in/courses/105106113/2_industrial_building/5_plastic_analysis.p
df. [Använd 15 May 2015].
http://homepages.engineering.auckland.ac.nz/~pkel015/SolidMechanicsBooks/
Part_II/08_Plasticity/08_Plasticity_01_Introduction.pdf. [Använd 11 June
2015].
[23] B. P. Dinis, D. Camotim och N. Silvestre, ”FEM-based analysis of the local
plate/distortional mode interaction in cold-formed steel lipped channel
columns,” Computers and Structures, vol. 85, nr 19-20, pp. 1461-1474, 2007.
[24] ABAQUS 6.11 online Documentation,
http://orange.engr.ucdavis.edu:2080/v6.11/books/gsa/default.htm?startat=ch10s
02.html.
[25] SS-EN 1993-1-5, Swedish Standards Institute, 2006/AC:2009.
[26] S. L. Chan, Guide on second-order and advanced analysis of structures

version 2, www.nida-naf.com, 2009.
[27] Broderna Edstrand AB, Roda Katalogen,Teknisk handbok, Boras: Multitryck,
1999.
[28] Y. B. Wang, G. Q. Li och S. W. Chen, ”Residual Stresses in welded flame-cut
high strength steel H-sections,” Journal of Construction Steel Research, vol.
79, pp. 159-165, 2012.
[29] M. Al-Emrani och B. Akesson, Steel Structures, Course Literature-VSM 191,
Goteborg: Chalmers University of Technology, 2013.
50
51
Appendix
Appendix 1: Modeling procedures
52
APPENDIX 1: Modeling procedures
1. Open a new ABAQUS file.
2. Create new part:
Under  Double click on
The following window opens:
Assign the name ‘Column’ Select

3DDeformableSolidExtrusionassign Approximate size to 1 
Continue.
3. Sketch the cross-section:
In the following window sketch the shape of the cross-section of an HEA

300 beam using the create line: Connected option. Then dimension it
using the Add dimension option and constrain it accordingly using the
constraints option. The required sketch is shown below.
Appendix 1: page1
4. Extrude the sketch:
click Done  Enter the length of the column that we need to extrudeOK.
5. Create radius:
Next using the option, create a radius in the extruded part. Select the
edges by holding down shift.  Done  Enter the values according to the
HEA 300 standards. (0.027)press enter.
Appendix 1: page2
6. Creating datum planes:
Using create datum planes at the center of the flange by selecting 3
points in the center of the flange. Using create a datum plane along the
line where the radius merges with the web. The desired result is shown
below:
7. Partitioning the Model:
Using create a partition Sketch origin : Auto Calculate select the

one of the end faces as shown below:
Appendix 1: page3
Done Select and edge or axis that will appear: Horizontal and on the
top select the top edge. In the following sketch window, draw 2 lines
along the centers of the cross-section as shown below:
 Click Done.
Repeat the same on the other side of the column. The partition sketches
are made in order generate the required points to help in partitioning the
column into equal cells.
We get the below result on both the ends:
Appendix 1: page4
8. Partitioning the model into cells:
Using partition the model into cells using the 3 points option. First
make sure to partition the Model into 4 large cells along the x-y direction.
i.e. we get 4 partitions of the cross-section shown below:
Then continue to partition using the 3 points command so as to obtain a

partition of the below cross-section on all of the 4 sides:
Appendix 1: page5
9. Assigning of material property:
Double click on  Assign name: Steel S355J2 In General

tab Density enter the following value:
Under Mechanical Elasticity Elastic enter the following valuesOK
Double click on  Assign name: I Section Solid 

Homogeneous Continue
In Edit Sections window Material: Steel S355J2 OK
From tool bar: Assign SectionSelect the column in the graphic

window DoneSection: I Section OK Select the column in the
graphic windowDone
Appendix 1: page6
In the Edit Section Assignment window Section: I SectionOK Done.
10. Assembly:
Select instances under assembly in the model tree
Under Create Instance window Partsselect Column DependentOK
Appendix 1: page7
11. Creating Reference points:
Under the Interaction Module  create a reference

point at the center of the cross-section at the top and the bottom using
as shown below:
Under ‘Column’ in the model tree Double click on Sets Name: Upper
NodesSelect whole surface on the upper cross-section holding down shift
Done.
Repeat the same for the lower nodes, this time giving the Name: Lower
Nodes.
12. Making the top and bottom surfaces rigid:
Under in the model treedouble click on Rigid

body  Continue
Appendix 1: page8
select pin(nodes)click on select ‘Lower Nodes’ from the sets
For the reference point , select and pick on RP-1 from the graphic
window. OK.
Repeat the same for the upper nodes.
13. Meshing the part:
Under the ‘column’ in the model tree double click on ‘Mesh’.
In toolbar, click on SeedPart
In the Global Seed window give Approximate Global seed to 0.05 OK.
This will seed that column along the length for element sizes of 50mm.
In toolbar, click on Seed Edges select the outline of the cross-section

line by line.
Appendix 1: page9
In the Basic tabSelect ‘By size’ approximate element size: 0.008 tick
Curvature control  Maximum deviation factor: 0.1OK.
In the tool bar click on ‘Mesh’PartOK to mesh the part?: Yes.
This completes the mesh for the column.
Appendix 1: page10
14. Creating Step:
Double click on ‘Step’ in the model tree Name: Linear Buckling Insert
New Step after: InitialProcedure type: Linear Perturbation Buckle
Continue.
In Edit Step window Number of eigenvalues requested:5 Maximum

number of iterations:1500OK.
Appendix 1: page11
15. Assigning Boundary Conditions:
Double click on in the model tree
In the Create Boundary Condition window Name: Top PinnedStep:

Linear Buckling MechanicalDisplacement/RotationContinue.
In the graphic window select RP-2(reference point 2) Done.
Appendix 1: page12
In the ‘Edit boundary Condition’ window Tick off U1 and U2 OK.
Follow the same for procedure Name: Bottom pinned, but this time
select RP-1 tick off U1, U2 and U3 as shown below:
Appendix 1: page13
Appendix 1: page14
16. Applying load:
In the Create load window Name: Load Mechanical Concentrated

force Continue.
In the graphic window select RP-2 done.
Appendix 1: page15
In the Edit Load window Assign CF1:0, CF2:0, CF3: -1OK.
17. Editing Keyword:
Right click on click on Edit Keyword.
In the Edit keywords window click at the end of the last
line….PRESELECTclick  type ” *NODE FILE”  press

Enter in the next line type ”U”OK.
18. Creating a Job:
Double click on Jobs
Appendix 1: page16
In the ‘Create Job’ window Name: Linear_Buckling_Analysis
Continue
In the Edit Job window OK.
Right click on the created job Submit.
Appendix 1: page17
NON LINEAR BUCKLING ANALYSIS:
Right click on  Copy Model Copy Model-1 to: Non-Linear

Buckling AnalysisOK.
1. Creating the Non-Linear Buckling Step:
Expand the created model tree under steps delete the Linear Buckling
step.
Double click on
In the Create Step WindowName: Non-Linear BucklingProcedure

type General Static, Riks Continue
Appendix 1: page18
In the Edit Step window Nlgeom: ON.
Appendix 1: page19
In the Incrementation tab Maximum number of increments: 50 Arc
length increment Initial: 0.01  Arc length increment Minimum: 1E-045
Arc length increment Maximum: 1E+036OK.
2. Applying load:
Double click on
In the Create Load window Name: Eigen Value Load Category:

Mechanical Type for Selected Step: Concentrated Load Continue.
In the graphic window Select RP-2Done.
Appendix 1: page20
In the Edit Load Window CF1: 0CF2: 0CF-3: -6.2267E+006(value
corresponding to Eigen mode at which buckling starts)
3. Boundary conditions:
Repeat the same as in the Linear Buckling step.
4. Editing Keywords:
Right click on  Edit key words
Appendix 1: page21
IN the Edit Keyword window After  click
on  and type
”*IMPERFECTION,FILE=Linear_Buckling_Analysis, STEP=1(Press
Enter) 1, 0.0306OK
(Make sure to remove the modifications in the keywords created in the

Linear buckling stage before creating the job.)
Appendix 1: page22
5. Creating the Job:
Double click on ’Jobs’ under Analysis
In the Create job windowName: NonLinear_Analysis Continue
In Edit Job WindowOK.
Appendix 1: page23
RESIDUAL STRESSES ANALYSIS:
Right click on  Copy Model Copy Model-1 to:

Residual Stress Analysis OK.
1. Creating sets of elements to assign Residual stress values:
Select
From the ‘Tools’ from the tool bar Sets Create
In the Create Set window Name: Max_Tension_FlangeType:

Element Continue.
In the graphic window,
Appendix 1: page24
Select the two sets of central stack of 3 elements as shown below:
Continue doing the same procedure for the stack of elements on the far end
of the flange
This time, Name: Max_Comp_Flange
The selection of the elements are as shown below:
Appendix 1: page25
Continue creating sets with the following names and in the below shown
pattern:
Name: Max_Comp_Web
Appendix 1: page26
Observe that for selection of the elements along the web, a stack of 4
elements are selected.
Now continue selecting the stack of elements center-outwards, each time

selecting the stacks along the web and flange separately. Name the sets of
elements on the web center-outwards as Web1, Web2, Web3……..etc.
Name: Web1
Appendix 1: page27
Name: Web2
Appendix 1: page28
Name: Web3
Similarly, create the sets along the flange. Name them as Flange1, Flange2,
Flange3 …..etc.
Appendix 1: page29
Name: Flange1
Appendix 1: page30
Name: Flange2
Appendix 1: page31
Name: Flange3
Appendix 1: page32
2. Assigning Residual stresses to the sets of elements:
In the Create Predefined FieldName:Max_Tension_Flange_Load

Category: MechanicalStressContinue.
Click on
In the Region Selection window select ‘Max_Tension_Flange
Appendix 1: page33
IN the Edit Predefined Field Sigma11: 0, Sigma22: 0, Sigma33: -1E+007,
Sigma12: 0, Sigma13: 0, Sigma23: 0 OK.
Continue assigning the stress values for the created sets using the values
obtained for the stress distribution calculated using the formulae generated
in excel.
3. Create Job and run the analysis:
Double click on jobs’
19. Creating a Job:
Double click on Jobs
In the ‘Create Job’ window Name: Residual_stress_analysis Continue
Appendix 1: page34
In the Edit Job window OK.
Right click on the created job Submit.
Appendix 1: page35
Faculty of Technology
351 95 Växjö, Sweden
Telephone: +46 772-28 80 00, fax +46 470-832 17

Resistance of Members To Flexural Buckling According To Eurocode 3

Uploaded by

Copyright:

Available Formats

Resistance of Members To Flexural Buckling According To Eurocode 3

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Resistance of Members To Flexural Buckling According To Eurocode 3

Uploaded by

Copyright:

Available Formats

Master's Thesis in Mechanical Engineering

Authors: Henry Mupeta, George John,

Examinar, LNU: Andreas Linderholt

Course Code: 4MT01E

Semester: Spring 2015, 15 credits

Linnaeus University, Faculty of Technology

Key words: buckling curves, buckling resistance, Eigen-value, Eurocode 3, flexural

Henry Mupeta, George John & Aliasgar Hirani

13th August, 2015.

Cross-sectional area of a steel column

1.2 Aim and purpose

To study the effects of imperfections, a case study of a structural support of

The work investigated mainly on a column having cross-section in class 3

2.1 Literature review on buckling

Recently, Ungermann et al. investigated local-flexural buckling interaction

In a more detailed manner, the numerical modeling of flexural buckling of

2.2 Euler buckling theory

1. The material is linearly elastic and homogenous.

2. The column is initially perfectly straight and there are no other

3. No residual stresses or other inner stresses.

4. The loading is centrically applied to the column

Figure 1: Euler Buckling [1]

Now, assuming the buckling deformation at a section having a distance

According to Euler-Bernoulli beam theory, the bending moment can be

where: is the modulus of elasticity.

is the second moment of area.

which is rearranged to:

This differential equation (4), has the general solution as:

Therefore, equation (5) is re-written as:

where 1 and 2 are constants.

The boundary conditions that apply for a pinned-pinned column are:

where is any integer or reflector of buckling mode.

The lowest value of the critical buckling load, is:

Equation (9) is known as the Euler buckling load which is a representation

where: is critical stress.

For columns, the radius of gyration, , is defined as:

It is a significant parameter because it is considered to indicate the stiffness

For a more convenient and easy notation, let:

where: is called the column slenderness ratio.

Therefore, equation (12) is re-written as:

Equation (14), indicates that the stress of a column is inversely proportional

Steel material exhibits a well-defined yielding point. Therefore, the yielding

Assuming the yield point is reached at the slenderness, , called the

Therefore, , is obtained by letting the buckling stress be equal to the yielding

The ratio of the slenderness, , to the reference relative slenderness, , is

2.3 Flexural buckling according to Eurocode 3

1. The yield strength, , of the material, and

2. The width to thickness (c/t) ratio.

Class3 cross-sections are those in which the stress in the extreme

2.3.1 Buckling resistance of members in compression

and for cross-section class 4:

where is the cross-sectional area.

is the reduction factor for the relevant buckling mode.

is the partial coefficient factor for member buckling resistance.

is the effective area of the cross-section when subjected to

2.3.2 Buckling curves

is known as the intermediate factor.

The non-dimensional slenderness λ is a parameter that depends on two