parametric study of
TRANSCRIPT
Author:
Javier Pascual Ramos
Advisors:
Arcadi Sanmartín Carrillo
Nativitat Pastor Torrente
Degree in:
Construction Engineering
Barcelona, 24 of may of 2013
Department of Applied Mathematics III
FIN
AL
DEG
REE
PR
OJE
CT
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
1
Contents
List of figures ................................................................................................................... 3
Acknowledgments ........................................................................................................ 5
Abstract ......................................................................................................................... 6
0. Introduction .................................................................................................................. 7
0.1. Motivation ............................................................................................................. 7
0.2. Outline of this document ....................................................................................... 8
1. Lateral-Torsional Buckling Overview .......................................................................... 9
1.1. History of Lateral-Torsional Buckling .................................................................. 9
1.2. Physical Vision of Lateral-Torsional Buckling and Phenomenological
Description ................................................................................................................. 12
1.3. Influencing Factors .............................................................................................. 14
1.3.1. End Restraints ............................................................................................... 14
1.3.2. Type of Loads ............................................................................................... 16
1.3.3. Load Application Point ................................................................................. 17
1.3.4. Cross-Section ................................................................................................ 18
2. Overview of Lateral-Torsional Buckling in Standards and Guides ........................... 19
2.1. Eurocode 3: Design of Steel Structures ............................................................... 19
2.1.1. Background of the European Standards ....................................................... 19
2.1.2. EN 1993 - Eurocode 3: Design of Steel Structures ...................................... 20
2.1.3. Buckling Resistance of Members According to EN 1993-1-1 ..................... 20
2.2. Specification for Structural Steel Buildings – American Institute of Steel
Construction (AISC) & American National Standard Institute (ANSI) ..................... 26
2.2.1. Studying of Buckling according to Specification for Structural Steel
Buildings 2010 (AISC/ANSI) ................................................................................ 26
2.3. British Standard (BS 5950-1:2000) ..................................................................... 31
2.4. Committee International for the Development and the Study of Tubular
Construction (CIDECT) ............................................................................................. 31
3. Parametric Study of Lateral-Torsional Buckling with Visual Basic for Applications 32
3.1. Calibrating the VBA Calculation Program .......................................................... 32
3.1.1. Variable Length ............................................................................................ 34
3.1.2. Variable Height............................................................................................. 35
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
2
3.1.3. Variable Base ................................................................................................ 36
3.1.4. Variable Web Thickness ............................................................................... 37
3.1.5. Variable Flange Thickness ........................................................................... 38
3.1.6. Variable Web and Flange Thickness ............................................................ 39
3.1.7. Summary Table ............................................................................................ 40
3.2. Parametrical Study of Closed Cross-Sections (Rectangular and Square) ........... 41
3.2.1. Parametrical Study: Case L/b ....................................................................... 41
3.2.2. Comparison EN-1993-1-1 with CIDECT ..................................................... 52
3.2.3. Comparison EN-1993-1-1 with British Standard ......................................... 57
4. Parametric Study of Lateral-Torsional Buckling with a Finite Element Method
Program .......................................................................................................................... 66
4.1. Linear Buckling Analysis with Finite Element Method ...................................... 66
4.2. Analysis of Different Cases with SAP2000 v14 ................................................. 68
4.2.1. Boundary Conditions .................................................................................... 68
4.2.2. Particular Cases to Study .............................................................................. 70
5. Conclusions ................................................................................................................ 76
5.1. Conclusions of the calibration of the VBA model .............................................. 76
5.2. Conclusions of the parametric study ................................................................... 76
5.3. Conclusions on comparison with British Standard and CIDECT ....................... 77
5.4. Conclusions of the SAP2000 v14 study .............................................................. 77
5.5. Future Works ....................................................................................................... 78
6. References .................................................................................................................. 79
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
3
List of figures
Figure 1: sample of lateral-torsional buckling ................................................................ 12
Figure 2: Buckling curves. [EN-1993-1-1, 2005] .......................................................... 25
Figure 3: Sample of studied beam .................................................................................. 32
Figure 4: χ vs L/b for ratios of h/b 1 to 3 and b=0,1m ................................................... 42
Figure 5: χ vs L/b for ratios of h/b 1 to 3 and b=0,15m ................................................. 42
Figure 6: χ vs L/b for ratios of h/b 1 to 3 and b=0,2m ................................................... 43
Figure 7: χ vs L/b for ratios of h/b 1 to 3 and b=0,25m ................................................. 43
Figure 8: χ vs L/b for ratios of h/b 1 to 3 and b=0,3m ................................................... 44
Figure 9: χ vs L/b for ratios of h/b 1 to 3 and b=0,35m ................................................. 44
Figure 10: χ vs L/b for ratios of h/b 1 to 3 and b=0,4m ................................................. 45
Figure 11: χ vs L/b for ratios of h/b 1 to 3 and b=0,45m ............................................... 45
Figure 12: χ vs L/b for ratios of h/b 1 to 3 and b=0,5m ................................................. 46
Figure 13: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1 .......................................... 47
Figure 14: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,22 ..................................... 47
Figure 15: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,44 ..................................... 48
Figure 16: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,67 ..................................... 48
Figure 17: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,89 ..................................... 49
Figure 18: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,11 ..................................... 49
Figure 19: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,33 ..................................... 50
Figure 20: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,56 ..................................... 50
Figure 21: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,78 ..................................... 51
Figure 22: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=3 .......................................... 51
Figure 23: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1 .......................................... 54
Figure 24: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,11 ..................................... 54
Figure 25: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,25 ..................................... 55
Figure 26: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,42 ..................................... 55
Figure 27: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,66 ..................................... 56
Figure 28: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=2 .......................................... 56
Figure 29: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,25 .................................... 59
Figure 30: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,33 .................................... 59
Figure 31: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,4 ...................................... 60
Figure 32: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,44 .................................... 60
Figure 33: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,5 ...................................... 61
Figure 34: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,67 .................................... 61
Figure 35: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,75 .................................... 62
Figure 36: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,8 ...................................... 62
Figure 37: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=2 ......................................... 63
Figure 38: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=3 (EC-3 and BS limits
matching) ........................................................................................................................ 63
Figure 39: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=4 ......................................... 64
Figure 40: Bifurcation Graphic ...................................................................................... 66
Figure 41: square cross-section (0,2x0,2x0,15 with L=20) ............................................ 70
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
4
Figure 42: First buckling mode for RHS 0,2x0,2x0,015 with L = 20 m. αcr = 3,13 ....... 71
Figure 43: Longitudinal view of the first buckling mode ............................................... 71
Figure 44: Front view of the first buckling mode ........................................................... 71
Figure 45: rectangular cross-section (0,2x0,1x0,15 with L=20) .................................... 72
Figure 46: First buckling mode for RHS 0,2x0,1x0,015 with L = 20 m. αcr = 0,724 ..... 73
Figure 47: Longitudinal view of the first buckling mode ............................................... 73
Figure 48: Front view of the first buckling mode ........................................................... 73
Figure 49: rectangular cross-section (0,3x0,1x0,15 with L=20) .................................... 74
Figure 50: First buckling mode for RHS 0,2x0,1x0,015 with L = 20 m. αcr = 0,724 ..... 75
Figure 51: Longitudinal view of the first buckling mode ............................................... 75
Figure 52: Front view of the first buckling mode ........................................................... 75
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
5
Acknowledgments
Honestly, I want to thank my advisors, Nati Pastor and Arcadi Sanmartín for the help
and the support they have provided me in the development of this work during the last 4
months. Also, the great patience they have had with my endless questions and anything
I needed, they provided me.
Despite of their youth, they have managed me to instill their extensive knowledge in the
matter and they have made me see beyond as well as feed my curiosity to know more
and not settle with the first thing you find.
It has been a pleasure. I stand by the decision I took almost a year ago.
Now that I am about to finish my degree, I am still thinking that these kind of teachers
make students improve.
THANK YOU VERY MUCH!
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
6
Abstract
In this work, out a Parametric Study of Lateral-Torsional Buckling in Closed Cross-
Sections has been carried. The aim is to assess the extent of this phenomenon
comparing different standards (Eurocode 3, AISC, Brisith Standard) and design guides
(CIDECT) and doing an exhaustive parametric analysis in order to assess the need of
considering the lateral-torsional buckling in these sections in some cases.
First, we present historical facts which show how the phenomenon started to be studied
centuries ago, what scientists have found out about lateral-torsional buckling and
examples of collapse of some structures because of this matter.
Secondly, the formulation of this phenomenon explaining their influencing factors, such
as end restraints, type of loads and so on is carried out.
After that, the treatment of different standards and design guides is carried out; how
they face the problem and what recommendations they give to study or not the problem,
specifically in square and rectangular cross-sections, as they are the ones studied in this
work.
Once the state of the art is analysed, a parametrical analysis with Visual Basic for
Applications (VBA) has been developed to try to assess the behaviour of the different
parameters of the cross-section and how they affect to the analysis of lateral-torsional
buckling. Also a comparison, between the parametric study carried out with VBA
(using the formulation of Eurocode 3), and other standards such as the British Standad
and Design Guide of the CIDECT is developed.
In addition, a few examples with a finite element method program (SAP2000 v14) have
been carried out, to compare with the results obtained with the formulation of Eurocode
3.
Conclusions of this work and interesting future works are exposed on.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
7
0. Introduction
The lateral-torsional buckling is a phenomenon of collapse in steel construction. It is
well-known that a certain type of cross-sections, such as “I” beams tend to suffer this
problem and must be braced properly in order to limit the undesirable out-of-plane
deformation and then the collapse. This is due to their low torsional inertia and the
lateral restraint provided by the web is low and need a physical displacement to avoid
the failure. However it is not critical in closed cross-sections such as square or
rectangular which have high torsional inertia, this is because the webs are connected on
the extreme of the flanges making the cross-sections closed and that plays against
lateral-torsional buckling, so this phenomenon is unlikely to happen.
The statement above is confirmed by two main standards used for the design of steel
structures: Eurocode-3 and American Standard (AISC). Both standards state that open
cross-sections must be checked against lateral-torsional buckling, but closed-cross
sections have a special treatment. Eurocode-3 states that squared cross-sections are not
susceptible to this phenomenon but it can be checked anyhow with the general
formulation. However, the American Standard provides a table of all kind of cross-
sections, and the specific problems they can suffer, and square & rectangular cross-
sections must not be checked against the phenomenon. In addition, no guidelines are
provided to check these cross-sections against lateral-torsional buckling, contrary to
what happens in Eurocode 3.
The aim of this work is to assess the extent of lateral-torsional buckling in closed cross-
sections. The point is to study commercial square & rectangular profiles and also bigger
profiles that can be prefabricated, welded or composite structures. The best way to
analyse how lateral-torsional buckling is affected by different geometric characteristics
is doing a parametrical study with the formulation given by the standards mentioned
above and besides check them with a commercial program of structures.
0.1. Motivation
First of all, this matter was of interest for the advisors, because despite of the fact that
standards state that it is not necessary to check closed sections, they have worked with
some special structures with hollow slender sections that had reduction for lateral-
torsional buckling. This is the case, for example, of a monorail with a rectangular
welded hollow section with a height about 0,7m and length of 35m.
In addition, the Design Guide of the CIDECT [CIDECT, 1996], which is not a standard,
but is widely used in the design of tubular structures, shows limit values of slenderness
of members below of which is not necessary to check against lateral-torsional buckling.
Implying beams that for the same slender ranges out of their limits necessary to check
it. Besides a design example provided in the Designers Handbook to Eurocode 3
[Gardner and Nethercot, 2005] shows the calculation of a reduction factor for lateral-
torsional buckling in a RHS is performed.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
8
Therefore, our view was as that in ordinary structures it is not an important factor, but
on special structures (high slenderness) could be significant and we would like to find
out these limit values in which is necessary to take into account the lateral-torsional
buckling or not.
Furthermore, while we were working on the state of art, we found that in the British
Standard (BS 5950) [British Standard, 2001] limit values of slenderness appear (as it is
explained in chapter 3) and there is a recent published article about lateral-torsional
buckling in elliptical hollow sections [Law and Gardner, 2012]. This seems confirm the
point that for slender elements or/and special structures it could be a factor to consider
in the design of closed cross-sections, although probably, it will not decisive.
So in this way, with the objective to consider which cases we have to take into account
the lateral-torsional buckling and which was its influence on the resistance of the beams
in that cases, we developed the study described on chapters 3 and 4 of this work.
0.2. Outline of this document
This document is organised as follows. After this introduction about the aim of the
work, an overview of lateral-torsional buckling phenomena is carried out in chapter 1,
including a historical review, the formulation as well as the main influencing factors for
this phenomenon. In chapter 2, we present the treatment of this mode of failure for
closed sections in different codes and/or design guides such as Eurocode 3 (European
standard), AISC (American standard), British Standard (United Kingdom standard) or
CIDECT (Design guide). In chapter 3, a parametric study of lateral-torsional buckling
for rectangular hollow sections is carried out, including a comparison with standards
depicted in section 2. In chapter 4 a finite element buckling analysis is performed for
some selected sections and compared with the results obtained in chapter 3. Finally,
main conclusions of this work can be found in chapter 5, including a provision for
possible future works.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
9
1. Lateral-Torsional Buckling Overview
1.1. History of Lateral-Torsional Buckling
Several well-known scientists have noticed on their experiments the curious behaviour
of the slender beams under certain load conditions [Timoshenko, 1953]. The beginning
of lateral-torsional buckling was noticed a few centuries ago, in seventeenth century,
when the illustrated Italian mathematician Galileo (1564-1642) started to experiment
and investigate the resistance to fracture of the same bar if it
works as a cantilever with the load at the end. Assuming
that the bar has a rectangular cross-section and that material
follows Hooke’s law up to failure, we obtain the stress
distribution. Actual materials do not follow Hooke’s law
until the failure, but on this basis of his theory Galileo
draws several important conclusions. “How and in what
proportion does a prism, whose width is greater than its
thickness, offer more resistance to fracture when the force
is applied in the direction of its breadth than in the
direction of its depth” [Crew and Macmillan, 1914]. So he
arrived at the conclusion that any given prism, whose thickness exceeds its width, will
offer greater resistance. Also, he studied the cantilever-beam problem keeping a
constant cross-section and concluded that the bending moment, due to the weight of the
beam, increases as the square of the length. Finally, Galileo discusses the strength of
hollow beams and states that “such beams are seen in the bones of birds which are light
and highly resistant both to bending and breaking”.
Then arrived Euler (1707-1783), an important talented mathematician, whose sentence
“Under a large enough compressive axial load an elastic
beam will buckle.” gave rise to the modern study of the
phenomenon as it is known nowadays as buckling. It is one
of the most well-known instabilities of classical elasticity.
The critical load for buckling was first derived by Euler in
1744 and further refined for higher
modes by Lagrange in 1770. In 1744
Euler’s book “Methodus inveniendi
líneas curvas...” was the first book on
variational calculus and it also contained
the first systematic treatment of elastic
curves. Euler, as a mathematician was
interested principally in the geometrical forms of elastic curves. He
accepted Jacob Bernoulli’s theory that the curvature of an elastic beam at any point is
proportional to the bending moment at that point. In addition, he investigated the shapes
of the curves which slender elastic bar will take up under various loading conditions.
Euler considered the various cases of bending and classified the corresponding elastic
curves according to values of angle between the direction of the force P and the tangent
Galileo Galilei
Leonhard Paul Euler
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
10
of at the point of application of the load, “When this angle is very small, we have the
important case of columns buckling under the action of an axial compressive force”
[Timoshenko, 1953]. Euler established a formula for the compression force, modified
later by Young, obtaining:
for the buckling columns which now has such
a wide application in analyzing the elastic stability of engineering structures. Since then,
Euler buckling has played a central role in the stability and mechanical properties of
slender structures in engineering.
At the end of seventeenth and beginning of eighteenth century, Lagrange (1736-1813),
contributed to elastic curves as well. The most important
contribution made by Lagrange to the theory of elastic curves is
his memoir “Sur la figure des colonnes” [Lagrange, 1781]. He
started with a discussion of a prismatic bar having hinges at the
ends and assumed there to be a small deflection under action of
the axial compressive force P, under this configuration it is
possible to have an infinite number of buckling curves.
Lagrange did not limit himself to a calculation of critical values
of the load P but went on to investigate the deflections which
exist if the load P exceed the critical value. Larange concluded that the column of
greatest efficiency has a cylindrical form. He arrived at the same conclusion by
considering curves which pass through four points taken at equal distances from axis.
He did not succeed in getting a satisfactory solution to the problem of the shape of the
column maximum efficiency. Later on the same problem was discussed by several other
authors. He made the usual assumption that the curvature is proportional to the bending
moment and discusses several instances which might be of some interest in studying.
The shape of Lagrange’s solution is too complicated for practical application.
Finally, Prandtl (1875-1953) a German physicist, in his PhD thesis (Kipperscheinungen)
[Prandtl, 1899] wrote a paper on lateral-torsional buckling of
beams with narrow rectangular cross-section bent in the plane
of maximum flexural rigidity. He obtained solutions for several
particular cases of practical importance. In making these tests,
Prandtl developed a technique of accurately determining the
critical load which was later used by many investigators in
elastic stability. This paper on lateral buckling initiated
numerous investigations on the lateral stability of beams and
curved bars. This is the first truly scientific work aimed
exclusively towards lateral-torsional buckling of beams. Before
Prandtl presented his work, an important and an early case of failure due to lateral-
torsional buckling, was the cast iron Dee Bridge in 1847 [Delatte, 2009] designed by
Robert Stephenson. The Dee Bridge failed because torsional instability was ignored a
failure mode.
The cast iron is weakest in tension than in compression, the tension flanges of girders
were larger than compression flanges by ratio of 16:3 (see on picture below), following
Joseph Louise de Lagrange
Ludwig Prandtl
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
11
the ratio material strengths. The Dee Bridge had a relatively long simple span for the
day of roughly 29m. The bridge was also designed with relatively low factor of safety
for the era of 1,5 [Petroski, 1994]. The most likely cause of failure was a torsional-
buckling instability to which the bridge girders were predisposed by compressive loads
introduced by eccentric diagonal tie
roads on the girder. Indeed, the top
flange went out of plane, causing
collapse. The lengthening of cast iron
spans caused the governing mode of
failure to change from bending to
lateral-torsional buckling. Subsequently,
Stephenson pointed out the need to
study failures to learn from them.
Lateral-trosional buckling remains an
important failure mode for steel beams,
unless the compressed flanges are
braced. The problem is generally less
severe than it was with iron because steel is equally strong in tension and in
compression and thus top and bottom flanges are of equal size. This type of buckling,
however, remains an important limit state that must be checked.
Failure of Dee Bridge,1847 in Chester (UK)
Cross-section, Dee
Bridge, 1847 (UK)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
12
1.2. Physical Vision of Lateral-Torsional Buckling and Phenomenological
Description
As explained above, the Lateral-Torsional Buckling is an instability phenomenon which
appears in elements subjected to bending, with insufficient lateral bracing of
compression flange, and rotating around its second moment of area major axis.
Considering the elastic, perfect member (no residual stress, or geometrical
imperfections) when the maximum bending reaches a certain value, called lateral-
torsional critical moment, the element that until then has been rotating around the acting
bending moment axis, starts to twist and roll out of the vertical plane, into the
perpendicular plane. This is the reason of its name, because of its lateral deformation.
Figure 1: sample of lateral-torsional buckling
The Eurocode 3 (European Standard) and AISC (American Steel Standard) provide
different methods to study this phenomenon and what it depends on.
Furthermore EC 3, which is very restrictive compared to other Standards [Trahair,
2010], take into account other aspects like imperfection of frames caused by lack of
verticality or straightness and any minor eccentricities present in joints of the unloaded
structure, and residual stress due to erection process e.g. welding and cooling. Those
factors reduce the buckling capacity of the beam [Martin and Purkiss, 2008]. However
AISC ignore geometrical imperfections and is more optimistic with respect to residual
stress [Trahair, 2010].
Both are developed on next chapters with its mathematical formulation and factors to
consider on different cross-sections.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
13
This critical moment affects differently to the structure, and it depends on 4 important
points, hereinafter:
End restraints
Type of applied loads
Load application point with respect the centroid of the cross section
Cross-section
The general expression of lateral-torsional critical moment is as follows:
[√
]
Where,
E modulus of elasticity
Iz minor moment of inertia
Iw warping constant
It torsional constant
L member length
C1 factor depending on loads and restraints [Acces Steel, 2010 or
Timoshenko and Gere 1961]
C2 factor depending on loads and restraints
C3 factor depending on loads and restraints
zg the distance between the application load point and the centroid of the
cross-section
zj depends on the symmetry of the cross-section
This expression can be simplified depending on the cross-section we are studying. For
example in rectangular sections of width b and height h, load applied on the centre of
the cross-section (zg=0) and for a double symmetric element (zj=0). Furthermore if a
uniform moment is applied (C1=1) [Acces steel, 2010] the equation for critical moment
(Mcr) reduces to:
[√ ]
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
14
The reason [Seaburg and Carter, 2003] is because torsion on a closed cross-section
(hollow or solid) is resisted by shear stresses in the cross-section that vary directly with
distance from the centroid. The cross-section remains plane as it twists (without
warping) and torsional loading develops pure torsional stresses only. So warping
constant (Iw) can be take it as zero.
The analysis and design of thin-walled closed cross-sections for torsion is therefore
simplified with the assumption that the torque is absorbed by shear forces that are
uniformly distributed over the thickness of the element [Siev, 1966].
1.3. Influencing Factors
1.3.1. End Restraints
The way how a structure is supported is an important aspect to consider since it affects
directly the lateral-torsional critical moment value. The main reason is because this
problem is more likely to be found when the supports allow rotation around the strong
axis (the rotation about the longitudinal axis is usually considered restrained), so the end
restraints will modify the range of the buckling.
As in any type of structure, the supports can be restrained as follows:
TYPE OF RESTRAINT ROTATION FIXED TRANSLATION FIXED
A
B
A
B
A
-
A
B
- A
B
A
-
A
-
Table 1: End restraints
The first supports, both A and B, do not allow any rotation in the plane. This plays
against lateral buckling and particularly with these boundary conditions, it will be more
unlikely that structure suffer the phenomenon.
The second case, support A does not allow rotations, but B allows rotation in the
horizontal plane and it has to be taken into account for the calculation.
The third one, both A and B allow the rotation in the horizontal plane.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
15
The fourth case, support A does not allow any movement, but B is completely free of
rotation and translation. That one is the worst situation in lateral buckling because the
restraint is lost in support B.
The first three cases [Acces Steel, 2010] are affected by two coefficients (C1 and C2)
depending on end restraints and loading (see section 1.2) that can modify (increase) the
elastic critical moment for lateral-torsional buckling.
On Standards (both Eurocode and AISC) we find typical cases, but some situations are
not currently covered. This is the case of cantilever, whose information is rather scarce
or incomplete. Some articles are focused on this issue such as [Clark and Hill, 1960]
who presented lower and upper bounds for coefficients factors applicable to cantilever
restraints. Moreover other articles published [Galéa, 1981 or Baláz and Koleková, 2004]
have looked into it more recently.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
16
1.3.2. Type of Loads
The lateral-torsional critical moment is function of the type of applied load on beams
and its boundary condition. Moment distribution takes part in the lateral buckling and it
has to be considered depending on the combination of both. The next table shows
different cases with loads and restraints and their moment distribution:
The values before must be taken to the coefficient C1 = 1/ ( kc2), and the Ψ=Mini/Mend
depends on the difference between both applied moments.
MOMENT DISTRIBUTION LOAD + RESTRAINT Value for kc
Uniform Bending Moment +
Simply supported
Linear Variation of Bending
Moment (same sign) + Simply
supported
Uniform Load + Simply
supported
Uniform Load + Clamped
Uniform Load + Clamped-Simply
supported
Point Load + Simply supported
Point Load + Clamped
Point Load + Clamped-Simply
supported
1,0
Ψ*
0,94
0,90
0,91
0,86
0,77
0,82
Table 2: Boundary conditions and values for kc [EN-1993-1-1]
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
17
1.3.3. Load Application Point
The load application point with respect the centroid of the cross section has influence on
lateral-torsional critical moment. The position of the point load application is directly
related with the beam shear centre and can introduce a secondary stabilizer or
destabilize moment.
Picture 1: Possible load application points
There are 3 types of section load application:
Above Shear Centre
If it is applied above the Shear Centre the load will have a destabilizer moment and that
will play negatively on the buckling (increases the torsional moment).
Just on Shear Centre
This situation will not cause any effect on lateral buckling (does not modify the
torsional moment). That is why we tend to apply loads on beams in this position.
Beneath Shear Centre
If it is applied beneath the Shear Centre the load will have a stabilizer moment and that
will play positively on the buckling (decreases the torsional moment).
Here it makes sense the parameter zg from the critical moment (see section 1.2). It is the
distance to the shear centre for above the shear it must be taken as > 0 and beneath the
shear centre it must be taken < 0. If it is just on the shear centre, has a value of zero.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
18
1.3.4. Cross-Section
The type of the cross-section is the most decisive factor on lateral buckling. There are
two main groups of cross-sections, closed cross-sections or opened cross-sections.
Closed Cross-Sections
As far as the studies are concerned, closed cross-sections normally tend not to show
lateral-torsional buckling. That is due mostly to their high torsional constant and their
warping constant equal to zero.
However slender sections, square, rectangular and round, may be subjected to lateral-
torsional buckling. According to American Standard [AISC & ANSI, 2010] this is
physically possible, but in standard design such slender sections are not used because
normally in these cases SLS deflection verification would govern the design.
Picture 2: Sample of closed-cross sections
Opened Cross-Sections
These sections are more susceptible to lateral-torsional buckling. That is due to their
low torsional constant and their warping constants different from zero.
If the bending moment is applied on major axis the beam is more susceptible to suffer
the phenomenon.
Picture 3: Sample of opened cross-sections
In addition to this other aspect related with the cross-section is the length of the beam,
as long slender beams will tend to suffer the problem.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
19
2. Overview of Lateral-Torsional Buckling in Standards and Guides
2.1. Eurocode 3: Design of Steel Structures
2.1.1. Background of the European Standards
The Commission of the European Community decided on an action programme in the
field of construction. The aim of this programme came up from the idea to eliminate
some possible technical obstacles to trade and the harmonization of technical
specifications.
Within this programme, the Commission took the initiative to establish a set of uniform
technical rules for the design of construction works which, in a first stage, would serve
as an alternative to the national regulations in force the Member States and, ultimately,
would replace them.
Few years later, The Commission and the Member States decided to transfer the
preparation and publication of the Eurocodes in order to provide them with a future
status of European Standard (EN).
These Eurocodes comprise the next standards generally consisting on a number of parts:
EN 1990 Eurocode 0: Basis of structural design
EN 1991 Eurocode 1: Actions on structures
EN 1992 Eurocode 2: Design of concrete structures
EN 1993 Eurocode 3: Design of steel structures
EN 1994 Eurocode 4: Design of composite steel and concrete structures
EN 1995 Eurocode 5: Design of timber structures
EN 1996 Eurocode 6: Design of masonry structures
EN 1997 Eurocode 7: Geotechnical design
EN 1998 Eurocode 8: Design of structures for earthquake resistance
EN 1999 Eurocode 9: Design of aluminium structures
Eurocode Standard recognize the responsibility of regulatory authorities in each
Member State and have safeguarded their right to determine values related to regulatory
safety matters at national level where these continue vary from State to State.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
20
2.1.2. EN 1993 - Eurocode 3: Design of Steel Structures
This part of the European Standard talks about the design of steel structures. It applies
to design of buildings and civil engineering works in steel. It complies with the
principles and requirements for the safety and serviceability of structures. It is intended
to be used with the rest of Eurocodes EN 1990, EN 1991 and EN 1992 to EN 1999.
Eurocode 3 is concerned only with requirements for resistance, serviceability, durability
and fire of resistance of steel structures. Other requirements such as concerning thermal
or sound insulation are not covered.
It is composed of twelve subparts EN 1993-1-1 to EN 1993-1-12 each addressing
specific steel components, limit states or materials.
The ones relevant to this work are EN 1993-1-1 that is about General rules and rules
for building.
2.1.3. Buckling Resistance of Members According to EN 1993-1-1
On this section it is explained how to study and the methodology to face the buckling
phenomenon according to EN-1993-1.
It is contained on EN-1993-1-1, inside section 6: Ultimate Limit State which considers
different forces on sections such as bending moment, shear, torsion, and also buckling
and so on.
2.1.3.1. Classification of Cross-Sections
The purpose of cross section classification is to identify the extent to which resistance
and rotation capacity of cross sections is limited by its local buckling resistance.
There are four different classes of cross-sections, as follows:
- 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.
- Class 3 cross-sections are those in which the stress in the extreme compression fibre 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.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
21
Picture 4: Classes of cross-section
A cross-section is classified according to the highest (least favourable) class of its
compression parts (see Table 3).
The classification depends on the width to thickness ratio of the parts subject to
compression, the height to thickness ratio of the parts subjected to bending and the steel
grade chosen.
On this work, Class 1, 2 and 3 are studied further on next chapters. They do not need
any reduction of their area, elastic modulus or inertia among others.
This Standard provides a table (see Table 3) to classify webs and flanges subjected to
bending or compression in closed sections. The fact is that lateral-torsional buckling
occurs when a bending moment is applied so webs are subjected to bending and one of
the flanges to compression. The other flange is subjected to tension so no classification
is applied.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
22
Table 3: Classification of cross-section. [EN-1993-1-1, 2005, Table 5.2]
Both compression parts include every part of a cross-section which is either totally or
partially in compression under the load combination considered.
The various compression parts in cross-section such as web or flange can be in different
classes.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
23
2.1.3.2.Uniform Members in Bending: Buckling Resistance
In order to analyse buckling, there are two main cases, according to section EN-1993-1-
1 where no lateral-torsional buckling checking is needed:
1. Beams with sufficient restraint to compression flange, which are not
susceptible to lateral-torsional buckling.
2. Beams with certain types of cross-sections, such as square or circular hollow
sections, fabricated circular tubes or square box sections are not susceptible
to lateral-torsional buckling.
This last statement is developed on this work because it is interesting to know the extent
of lateral-torsional buckling in cross-sections.
The verification, given by EN-1993-1-1 to study lateral-torsional buckling on the
second case, which is a laterally unrestrained member subject to major axis bending,
should be verified as the follow:
Where,
MEd is the design value of the moment
Mb,Rd is the design buckling resistance moment
This ratio must be 1,0 because Mb,Rd is the highest value that section can reach, so
MEd cannot exceed it. The design buckling resistance moment is calculated as:
Where, Wy is the appropriate section modulus as follows:
Wy is Wpl,y for Class 1 or 2 cross-sections
Wy is Wel,y for Class 3 cross-sections
Wy is Weff,y for Class 4 cross-sections (not studied)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
24
And,
is the partial factor for buckling resistance and a recommended value of
1,00
is the yield strength of the material
is the reduction factor for lateral-torsional buckling that ranges from
0 to 1 and it is non-dimensional. See below for other details of the factor.
2.1.3.3. Lateral-Torsional Buckling Parameter and Buckling Curves
For bending members of constant cross-section, the value of for the appropriate
non-dimensional slenderness , should be determined from the given formulation:
√
Where,
[ ( ) ] and ranges from 0 to 1. It is a non-
dimensional parameter.
is an imperfection factor and must be taken from Table 4. It is a non-
dimensional parameter.
is the non-dimensional slenderness, it is calculated as:
√
is the elastic critical moment for lateral-torsional buckling. It is based on
gross cross sectional properties and takes into account the loading conditions.
The Standard provides a table to determine what buckling curve we must chose and
other recommended values that belongs to each curve (see Table 4 and Table 5).
In addition, EN-1993-1-1 says that it is not necessary check the phenomenon for <
0.4. That will be analysed in a specific chapter of this work (see section 2.1.3).
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
25
Cross-section Limits Buckling curve
Rolled I sections h/b 2
h/b 2
a
b
Welded I sections h/b 2
h/b 2
c
d
Other cross-sections - d
Table 4: Recommended values for lateral torsional buckling curves for cross-sections using for
[EN-1993-1-1, Table 6.4
Buckling Curves
The graphical representation of these curves is the following one:
Figure 2: Buckling curves. [EN-1993-1-1, 2005]
- Curve a, represents quasi perfect shapes.
- Curve b, represents shapes with medium imperfections.
- Curve c, represents shapes with a lot of imperfections.
- Curve d, represents shapes with maximum imperfections.
These curves come from experimental tests of beams with different cross-sections and
with different values of slenderness. A probabilistic approach showed that it was
possible to draw some curves describing column strength as a function of the reference
slenderness [Batista and Rodrigues, 1994].
Buckling curve a b c d
Imperfection factor 0,21 0,34 0,49 0,76
Table 5: Recommended values for imperfection factors for lateral-torsional buckling curves
[EN-1993-1-1, Table 6.3]
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
26
2.2. Specification for Structural Steel Buildings – American Institute of Steel
Construction (AISC) & American National Standard Institute (ANSI)
The American Institute of Steel Construction (AISC) and American National Standard
Institute (ANSI), are a not-for-profit technical institutes and trade associations
established in 1921 and 1968 respectively, to serve the structural steel design
community and construction industry in the United States. The objective is to provide
the civil engineering specifications and code developments, research, standardization
and so on.
The Specification has been developed as a consensus document to provide a uniform
practice in the design of steel-framed buildings and other structures. The intention is to
provide design criteria for routine use and not to provide specific criteria for
infrequently encountered problems.
2.2.1. Studying of Buckling according to Specification for Structural Steel
Buildings 2010 (AISC/ANSI)
On this section is explained how to study and the methodology to face the buckling
phenomenon.
Chapter B called Design requirements has a specifically part 4, Member Properties, that
contains specifications to classify sections subjected to axial compression or flexure.
Chapter F called Design members of flexure, provides information of design members
for flexure such as I-shaped, square and rectangular, round or tees among others.
2.2.1.1.Design Requirements– Member Properties
2.2.1.1.1. Compression
For compression, sections are classified as non-slender element or as slender element
sections. For a non-slender element section, the width-to-thickness ratios of its
compression elements shall not exceed λr from the table given by this Standard. If the
width-to-thickness ratio of any compression element exceed, the section is a slender-
element section. This case is not important to this work because it is not about analysis
of compression elements.
2.2.1.1.2. Flexure
For flexure, which is our case of study because of lateral-torsional buckling, sections are
classified as compact, non-compact or slender-element sections. For a section to qualify
as compact, its flanges must be continuously connected to the web and the width-to-
thickness ratios of its compression elements shall not exceed the limiting of λp from the
Table 3. If the width-to-thickness ratio of one or more compression elements exceed,
the section is non-compact. Furthermore if the width to-thickness ratio of any
compression element exceeds λr, the section is a slender-element section.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
27
There is an important difference with Eurocode 3 because cross-sections are not
classified on different classes, but we can compare them as:
This is to make the comparison with the model of classification given by the EN-1993-1
which divides the cross-section in 4 classes. But AISC does not classify in different
classes so if value of width-to-thickness ratio (see Table 6) is minor than λp the section
can be considered as being in class 1 or 2. If the value is between λp and λr is similar to
class 3. Finally, if the value is higher than λr it can be understood as class 4.
2.2.1.1.3. Stiffened or unstiffened elements
Stiffened elements are those that become rigid and are not susceptible to buckle whereas
unstiffened elements may suffer local buckle on webs or flanges.
The stiffeness plays part into the section classification as it improves the capacity of the
cross-section. All sections can be stiffened or unstiffened as I-shaped, legs or angles or
tees among others.
Class 1 and 2 λp < Class 3 < < λr < Class 4
Eurocode 3 AISC Eurocode 3 AISC Eurocode 3
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
28
Table 6: Classification of cross-sections subjected to flexure. [AISC, 2010]
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
29
2.2.1.2. Design of Members for Flexure
Chapter F of the AISC specification applies to members subject to simple bending about
one principal axis. For simple bending, the member is loaded in a parallel to a principal
axis that passes through the shear centre or is restrained against twisting at load points
supports.
The chapter is organized in 13 sections, each one for a different member type. The one
dealing with hollow (square and rectangular) sections is the F7:
F7 Square and rectangular HSS box-shaped members
This Standard is divided into each particular case as in opposition to the European
Standard which provides a general method to study any type of cross-section.
In addition to this, it is included a summary table (see Table 7) about problems that can
suffer different cross-sections.The problem of lateral-torsional buckling does not appear
on the table below on limit states column. This is an important difference from the EN-
1993-1-1.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
30
Table 7: Limits states for different cross-sections. [AISC, 2010]
Since this work is focused on closed sections (square and rectangular), hereafter only
these specific cross-sections are discussed.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
31
2.2.1.3.Square and Rectangular HSS and Box-Shaped Members
F7 section of AISC applies to square and rectangular hollow steel sections (HSS), and
doubly symmetric box-shaped members bent about either axis, having compact or non-
compact webs and compact or non-compact slender flanges.
The AISC Standard mentions that in case of very long rectangular HSS bent about their
major axis are subject to lateral-torsional buckling, whereas the Specification provides
no strength equation for this limit state since beam deflection will control for all
reasonable cases as it is shown on table before, that it does not appear the phenomenon
in this type of cross-section. Thus, this Standard refuses the possibility that this type of
section may fail by this phenomenon.
So there are 3 possible cases that make failure the structure
Yielding
Flange Local Buckling
Web Local Buckling
Each case is developed on the Standard with its mathematical formulation but it is not
explained here since neither of them are not relevant for lateral-torsional buckling cases
of failure. An important fact is that on the study of this type of section, it is not given
any reduced factor for slender cross-sections as it is given on European Standard.
2.3. British Standard (BS 5950-1:2000)
The British Standard [British Standard, 2001] (now is withdrawn to EN-1993) was used
in the United Kingdom to the design and analysis of steel structures. This Standard
treated the problem of lateral-torsional buckling different from the Eurocode 3 and
AISC, it provides a table (see Table 11 in section 3.2.3) with limiting values below of
which is not necessary to check the problem. The values provided on this standard are
compared to the ones obtained with EN-1993-1-1 in the parametric study developed in
section 3.2.3.
2.4. Committee International for the Development and the Study of Tubular
Construction (CIDECT)
The Design Guide 2 from the CIDECT [CIDECT, 1996], as the name define itself, it is
not a Standard that must be follow compulsory, but provides guides to the design of
steel constructions. It is widely used as a guideline for the design of structures with steel
hollow sections. We have analysed it and compared Eurocode 3 in section 3.2.2. As in
the case of British standard, they provide a table (see Table 10 in section 3.2.2) with
limiting values of slenderness to account or not for lateral-torsional buckling the case of
lateral-torsional buckling.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
32
3. Parametric Study of Lateral-Torsional Buckling with Visual Basic
for Applications
In this chapter we carry out a parametric study in order to assess the extent of lateral-
torsional buckling in closed sections.
The parametric study has been developed on the basis of Eurocode’s formulation,
particularly the one provided in EN-1993-1-1.
The study has been carried out with Visual Basic for Applications (VBA), a
programming language that is determined by events and provides comprehensive
facilities to computer programmers, which allows the users to implement calculation
subroutines in junction with standard Microsoft office programs such as excel
There are 2 main steps on the VBA calculation program developed for this work:
1. Creating a calculation program that need to be calibrated justifying the obtained
results, and also allows to develop a first study of the influence of different
parameters on lateral-torsional buckling.
2. Creating a model on which is possible to obtain a large number of interesting
results from a parametric study that are useful to obtain conclusions on the need
to consider or not lateral-torsional buckling on closed sections.
3.1. Calibrating the VBA Calculation Program
The objective of this chapter is to asses in a first approximation the variation of the
different parameters affect to the lateral-torsional buckling of the beam, both taking into
account cross-sectional parameters and also the length of the beam, see figure below
Figure 3: Sample of studied beam
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
33
The way how the parameters influence has been studied is as follows:
1. Variable Length and remaining ones fixed
2. Variable Height and remaining ones fixed
3. Variable Base and remaining ones fixed
4. Variable Web Thickness (tw) and remaining ones fixed
5. Variable Flange Thickness (tf) and remaining ones fixed
6. Variable Web and Flange Thickness (tw = tf) remaining ones fixed
On the other hand the main that has been studied is the reduction factor for lateral-
torsional buckling ( . For this purpose, graphics have been produced with the
variation of with the parameters listed above. In a secondary way, and in order to
analyse the graphics of , the variation of the other properties such as the ones listed
below, has been assessed too.
1. Area
2. Inertia around weak axis (Iz)
3. Inertia around strong axis (Iy)
4. Elastic Modulus (Wy)
5. Torsional Inertia (It)
6. Cross-section class (c1, c2, c3 or c4)
7. Critical Moment (Mcr)
8. Non-dimensional slenderness ( )
9. , value to determine the reduction factor for lateral-torsional buckling( )
As a first idea for a good calibration of the graphics of variation of , all of them must
have the critical point (where reduction starts) at the same value. For this purpose we
have set a first geometry of a rectangular cross-section and get the critical length that
makes it buckle it. Then, once fixed the length try to get the critical height that makes it
buckle with the parameters fixed before, and the same for the rest of the geometric
parameters.
The initial parameters chosen are the follows:
PRAMETERS VALUE UNIT
Length 15 m
Height 0,745 m
Base 0,565 m
Web & Flange thickness 0,015 m
E 21x107
KN/m2
G 81x106
KN/m2
fy 355000 KN/m2
α 0,76 -
π 3,1415 -
ε 0,81 -
Table 8: Initial parameters
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
34
The table above shows the values of critical length, critical height and so on. In this
way, they are the inflexion point for reduction factor .
After a brief analysis of each graphic, there is a summary table that reflects the
behaviour of the main parameters.
In the following subsections the graphics for the variation of with each parameter
are shown and analyzed.
3.1.1. Variable Length
The lateral-torsional buckling is studied varying the length of the beam. The other
cross-section’s parameters are fixed. The behaviour is the following:
Graphic 1: Variable length
For lengths < 15m remains constant ( =1) and for lengths larger than 15m the
graphic follows a linear decreasing variation while the length is growing. This is logical
because the critical moment expression is inversely proportional to the length. This
makes the slenderness increase and decrease.
In addition, there are no changes on the cross-section class because the length does not
affect to the section classification.
Note that critical point occurs for 15m and this is why all the next graphics are done for
L=15m, as detailed in Table 8.
0,98
0,985
0,99
0,995
1
12 13 14 15 16 17 18 19 20
χ vs Length
χ
Limit
χ
Length (m)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
35
3.1.2. Variable Height
The lateral-torsional buckling is studied varying the height of the beam. The other
cross-section’s parameters are fixed and L=15m. The behaviour is the following:
Graphic 2: Variable height
After the critical point, the graphic follows a linear decreasing variation while the height
is growing. This is logical because the height makes the slenderness increase
which reduces the critical moment. This causes the decrease for increasing heights.
In addition, there are changes on the cross-section class because the height affects to the
section classification as can be seen in the figure.
Note that the critical point is obtained for a value of height = 0,745m which corresponds
to the expected value (Table 8)
0,995
0,996
0,997
0,998
0,999
1
0,7 0,75 0,8 0,85 0,9 0,95 1
χ vs Height
χ
Limit
Class 1 χ
Height (m)
Class 2
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
36
3.1.3. Variable Base
The lateral-torsional buckling is studied varying the base of the beam. The other cross-
section’s parameters are fixed. The behaviour is the following:
Graphic 3: Variable base
The graphic follows a non-linear increasing variation while the base is growing until
reaching a constant value = 1 for base > 0,565m. This is logical because a larger
base makes the Iz increase which results on a larger critical moment. So this makes the
slenderness go down and the goes up for increasing values of b.
In addition, there are changes on the cross-section class because the base affects to the
classification as it is shown on the graphic.
Again, the critical point is reached for the expected value
0,965
0,97
0,975
0,98
0,985
0,99
0,995
1
0,4 0,45 0,5 0,55 0,6
χ vs Base
χ
Limit
χ
Base (m)
Class 1 Class 2
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
37
3.1.4. Variable Web Thickness
The lateral-torsional buckling is studied varying the web thickness of the beam. The
other cross-section’s parameters are fixed. The behaviour obtained is the following:
Graphic 4: Variable web thickness
The graphic follows a non-linear increasing variation while the web thickness increases,
until reaching the critical point. This is due to the fact that larger tw means larger inertia
and increases the critical moment. So these effects make the slenderness go down
and the increase with tw.
In addition, despite of the fact that the web thickness affects directly, there are no
changes on the class classification for the range of tw considered here
Again, the critical value for tw matches the expected value on Table 8.
0,96
0,965
0,97
0,975
0,98
0,985
0,99
0,995
1
0,005 0,007 0,009 0,011 0,013 0,015 0,017 0,019 0,021
χ vs tw
χ
Limit
χ
tw (m)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
38
3.1.5. Variable Flange Thickness
The lateral-torsional buckling is studied varying the flange thickness of the beam. The
other cross-section’s parameters are fixed. The behaviour is the following:
Graphic 5: Variable flange thickness
The graphic follows a linear decreasing variation while the flange thickness is growing
after the critical point. This may seem an illogical behaviour because an increase of
thickness tf makes increase both inertia and critical moment. However, the growth of
flange thickness causes also that the slenderness increase and thus decreases,
which at the end is the governing behaviour and it goes against the expected results.
In addition, despite of the fact that the flange thickness affects directly, there are no
changes on the section classification, for the range of tf considered.
0,9986
0,9988
0,999
0,9992
0,9994
0,9996
0,9998
1
1,0002
0,0146 0,0148 0,015 0,0152 0,0154 0,0156 0,0158 0,016
χ vs tf
χ
Limit
χ
tf (m)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
39
3.1.6. Variable Web and Flange Thickness
The lateral-torsional buckling is studied varying the flange thickness and web thickness
of the beam at the same time. The other cross-section’s parameters are fixed. The
behaviour is the following:
Graphic 6: Variable thickness
The graphic follows a linear decreasing variation after the critical point while the flange
thickness is growing. This is the same variation as in the last case. We notice that the
variation of flange thickness is dominant over variation of web thickness. Once again, it
is not an expected result.
In addition, despite of the thickness affects directly to it, there are no changes on the
class of the same section. This is because there is no change in the range of interest.
Note that up to 18mm, the reduction factor decreases only to 0,9997.
Again we obtain the critical thickness at the expected value.
0,9997
0,99975
0,9998
0,99985
0,9999
0,99995
1
1,00005
0,014 0,0145 0,015 0,0155 0,016 0,0165 0,017 0,0175 0,018
χ vs t
χ
Limit
χ
t (m)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
40
3.1.7. Summary Table
Hereafter it is shown a summary table which reflects on a fast way the behaviour of
each variation in relation with the main parameters studied.
PARAMETER Iy It My Mcr
Length
Height f s
Base s f
Web Thickness s f
Flange Thickness f s
Thickness f s
Table 9: Summary table
Increasing all parameters’ cross-section (h, b, tw, tf), both mechanics features and
resistance improve. However slenderness ( and reduction factor for lateral-torsional
buckling ( ) do not always follow a logical behaviour as it happens in flange
thickness. In addition we have noticed that it is dominating when we vary both
thicknesses at the same time.
The subscripts “f” and “s” mean, fast and slow, e.g. in the case of the variable height My
grows faster (f) than the Mcr (s). However in the case of the variable base My grows
slower (s) than Mcr (f).
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
41
3.2. Parametrical Study of Closed Cross-Sections (Rectangular and
Square)
Once we have validated the VBA calculation program, we can apply it to perform an
exhaustive parametric study to obtain interesting results that we can find in real cases.
The objective is to assess the extent of lateral-torsional buckling on this type of usual
closed cross-sections and also try to compare the results with some of the values
provided in the literature (British Standard / CIDECT).
Analyzing the graphics above we have seen that the influence of the thickness is low
( ≈ 0,9996) , so we do not take into account the thickness for the parametric study,
and we consider a simple value of thickness as 0,015m. We are going to develop 3 types
of variation graphics:
1. - χ vs L/b (see 3.2.1)
2. - χ vs L/h or CIDECT case (see 3.2.2)
3. - χ vs L/h or British Standard (BS 5950-1:2000) case (see 3.2.3)
3.2.1. Parametrical Study: Case L/b
On this section, the behaviour of closed cross-sections on the lateral-torsional buckling
related with the ratio L/b is assessed.
On the graphics obtained we see, for a fixed value b, different curves of h/b (legend).
All versus reduction factor for lateral-torsional buckling (ordinate axis) and rate L/b
(abscissa axis). It has been carried out for bases between 0,1 to 0,5m (commercial ones)
and for the ratio h/b from 1 to 2,5 (commercial ones) and up to 3 (welded sections).
The graphics obtained are the following:
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
42
Figure 4: χ vs L/b for ratios of h/b 1 to 3 and b=0,1m
Figure 5: χ vs L/b for ratios of h/b 1 to 3 and b=0,15m
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,1m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,15m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
43
Figure 6: χ vs L/b for ratios of h/b 1 to 3 and b=0,2m
Figure 7: χ vs L/b for ratios of h/b 1 to 3 and b=0,25m
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,2m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,25m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
44
Figure 8: χ vs L/b for ratios of h/b 1 to 3 and b=0,3m
Figure 9: χ vs L/b for ratios of h/b 1 to 3 and b=0,35m
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,3m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,35m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
45
Figure 10: χ vs L/b for ratios of h/b 1 to 3 and b=0,4m
Figure 11: χ vs L/b for ratios of h/b 1 to 3 and b=0,45m
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,4m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,45m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
46
Figure 12: χ vs L/b for ratios of h/b 1 to 3 and b=0,5m
As a general view of graphics with b fixed, we see that on the first graphic (Figure 4),
all the curves of h/b are so separated, but while the b increases, the range starts to be
narrower. So if we have a relatively narrow base the difference between higher or lower
h/b ratios is important, but as the base grows up, the size of the base starts to be less
important and the reduction values tend to get closer to each other.
Also, we see how cross-sections with low h/b (square has h/b=1) (Figure 4) need a high
L/b ratio to buckle than slender cross-sections that start to buckle before. But this
difference becomes smaller when b increases.
For this kind of representation the curves are separated and thus, we cannot obtain any
global critical point. It seems more reasonable to group the sections using its h/b ratios
regardless its base or height actual dimensions. These new arrangements can be seen in
the following figures:
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
b=0,5m
1,00
1,22
1,44
1,67
1,89
2,11
2,33
2,56
2,78
3,00
h/b
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
47
Figure 13: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1
Figure 14: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,22
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=1
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=1,22
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
48
Figure 15: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,44
Figure 16: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,67
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=1,44
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
0,91
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=1,67
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
49
Figure 17: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=1,89
Figure 18: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,11
0,91
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=1,89
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
0,9
0,91
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=2,11
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
50
Figure 19: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,33
Figure 20: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,56
0,9
0,91
0,92
0,93
0,94
0,95
0,96
0,97
0,98
0,99
1
0 10 20 30 40 50 60
χ
L/b
h/b=2,33
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
h/b=2,56
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
51
Figure 21: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=2,78
Figure 22: χ vs L/b for b from 0,1 to 0,5 and ratio of h/b=3
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
h/b=2,78
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
0,88
0,9
0,92
0,94
0,96
0,98
1
0 10 20 30 40 50 60
χ
L/b
h/b=3
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
52
Taking a look at this series of figures we can draw the following comments:
First of all, as a general view of graphics with h/b fixed, we see that on the first graphic
(h/b=1, Figure 13), all the curves of h/b are so superposed, that means that for a fixed
rate h/b=1 having a wide or narrow base (except extreme curves b=1 that correspond to
very slender elements) does not have a significant influence. But we can state a critical
point, in this case for L/b=28. And the same for the critical point for h/b=3, whose ratio
L/b is around 20. So we can conclude that for ratios h/b 1 to 3 the critical zone is 20 <
L/b < 30. This means that the lengths around which the theoretical reduction for lateral-
torsional buckling starts can be very usual, between 2 or 3 metres for small sections of
0,1m of base, up to 15m for heavy sections of 0,5m base.
Secondly, we see how cross-sections with low h/b (square has h/b=1, Figure 13) need a
high rate L/b to buckle than slender cross-sections that start to buckle before. But this
difference becomes narrow when b increases.
An important appreciation of the graphics is that there are more differences between
graphics for a fixed b and different families of h/b than fixing h/b and different families
of b. It means that the graphic with title b=0,1m (Figure 4) if we see different families
of h/b (in general) they are separated, whereas if we see the graphic with title h/b=1
(Figure 13) the families of b (in general) are very narrow. So for this reason in the
graphics of a fixed b are not appropriate to draw conclusions as h/b fixed, and we
cannot conclude an exact critical zone as before. So we will not do this analysis on next
parametric graphics below.
3.2.2. Comparison EN-1993-1-1 with CIDECT
Eurocode 3 states that members with cross-sections such as square, are not susceptible
to this phenomenon and for other cross-sections such as I beams there is no need to
check them if ≤ 0,4 (see section 2.1.3.3). On the other hand CIDECT provides with
reference values on a table to study the lateral-torsional buckling, in square and
rectangular cross-sections (see below).
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
53
Table 10: Limit values; below them it is not necessary to check lateral-torsional buckling [CIDECT, 1996]
The CIDECT advices with this table that for ratios below of those shown on the table, it
is not necessary to check the lateral-torsional buckling.
Therefore, in order to compare our results, we now plot our graphics with L/h in the
abscissa instead of L/b. The point is to obtain graphics for different ratios h/b (inverse
of γ on the table before) and see the difference with the Eurocode 3 and CIDECT
according to the table before.
On the graphics obtained we see, for a fixed ratio h/b, different curves of b (legend). All
versus reduction factor for lateral-torsional buckling (ordinate axis) and rate L/h
(abscissa axis). It has been carried out for the ratio h/b from 1 to 2 as it is shown on the
table above.
The straights bars represent:
EC-3 bar: from the bar to the left side the ≤ 0,4 zone is found, where there is
no need to check the lateral-torsional buckling.
CIDECT bar: from the bar to the left side the zone where no checking is needed
is found.
The graphics obtained are the following:
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
54
Figure 23: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1
Figure 24: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,11
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=1
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=1,11
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
55
Figure 25: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,25
Figure 26: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,42
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=1,25
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=1,42
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
56
Figure 27: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=1,66
Figure 28: χ vs L/h for b from 0,1 to 0,5 and ratio of h/b=2
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=1,66
0,1
0,15
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
0,75
0,8
0,85
0,9
0,95
1
20 40 60 80 100 120
χ
L/h
h/b=2
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
CIDECT
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
57
We appreciate some important aspects related with the behaviour of the graphics while
the rate h/b increases from 1 to 2. Note that for the presented range of sections h/b ratios
(from 1 to 2), the most common commercial sections can be found
First of all, as a general view, we see that on the first graphic (h/b=1) (Figure 23), the
checking point is ratio L/h=100 for Eurocode 3 and 110 for CIDECT. If we do the same
for the last case (Figure 28), the checking point is a ratio L/h=40 for Eurocode and 50
for CIDECT. So we can state a checking point for each case of h/b. In addition the ratio
L/h for which χ starts to be < 1 is between 25 ≤ L/h ≤ 35.
As an example of the explained before, low ratios of h/b, is more difficult to buckle. If
we see Figure 23, the first that shows h/b=1 (it is a square cross-section), needs a rate
L/h almost 100 to have a reduction for lateral-torsional buckling of 0,85, this means for
example, a beam with length=10m, the height will be 0,1m. But if we see the last that
shows a rate h/b=2 (it is a slender cross-section)Figure 28, needs a rate L/h about 50 to
have the same reduction for lateral-torsional buckling. It means that as the same
example before, the height will be 0,2m. That is logical, slenderness is an important
characteristic that influences this problem, so the behaviour obtained is the expected.
In spite of this we must keep in mind that ratios of L/h=100 or L/h=50 are unusual
values on normal structures, so the limits of the Eurocode are applicable for special
structures. Note if we treat the member as an I beam and a lateral-torsional buckling
check is performed we would find reduction factors of 0,85 for the limit slenderness 0,4.
This means a 15% of reduction of the member bending strength which is rather high
value.
Secondly, the recommendations from Eurocode and CIDECT look a little bit different.
We see how Eurocode 3 is more conservative than CIDECT. The Eurocode fix the
checking point for cases between 0,85 ≤ χ ≤ 0,87 (or ≤ 0,4), as it says that it is not
necessary to check the lateral-torsional buckling of square and rectangular cross-
sections. On the other hand, CIDECT fix this value in between 0,88 ≤ χ ≤ 0,83. In that
way, we can define that the Eurocode is a little more conservative than CIDECT that is
to say that CIDECT allows higher lengths with no lateral-torsional buckling checking.
But indeed, we can make a whole lecture of the graphics obtained. Really, the
difference, appreciated in both standards that range on 0,03 points, is not extremely big.
So roughly, this difference does not have an important impact on the study of lateral-
torsional buckling between designing a beam with one standard or the other.
3.2.3. Comparison EN-1993-1-1 with British Standard
This section contains a comparison between Eurocode 3 (which is currently in force)
and British Standard (now withdrawn) that was the standard used in the United
Kingdom and was superseded by Eurocode. Both of them provide technical rules to
treat lateral-torsional buckling.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
58
As explained in the comparison of Eurocode 3 and CIDECT for RHS, no lateral-
torsional buckling checking is needed. According to Eurocode 3 for other cross-sections
this lateral-torsional buckling checking can be omitted for ≤ 0,4. However British
Standard provides guidance to study the lateral-torsional buckling in square and
rectangular cross-sections, as can see in the following table:
Table 11: Limit values of L/iz; below them it is no necessary to check the lateral-torsional buckling.
[British Standard, 2001]
The table before given by British Standard shows that for values below the indicated on
the table checking the lateral-torsional buckling is not necessary.
The point is to obtain graphics for different h/b ratios and see the difference of limits
between Eurocode 3 and ratios of British Standard.
On the graphics obtained we see, for a fixed rate h/b, different curves of b (legend). All
versus reduction factor for lateral-torsional buckling (ordinate axis) and rate L/iz
(abscissa axis). Where iz stands for the radius of gyration.
The straights bars represent:
EC-3 bar: from the bar to the left side means ≤ 0,4, it would not be needed
to check it.
British Standard bar: from the bar to the left side means that it is not needed to
check it.
The graphics obtained are:
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
59
Figure 29: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,25
Figure 30: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,33
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,25
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,33
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
60
Figure 31: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,4
Figure 32: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,44
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,4
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,44
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
61
Figure 33: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,5
Figure 34: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,67
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,5
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,67
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
62
Figure 35: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,75
Figure 36: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=1,8
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,75
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=1,8
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
63
Figure 37: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=2
Figure 38: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=3 (EC-3 and BS limits matching)
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=2
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=3
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
64
Figure 39: χ vs L/iz for b from 0,1 to 0,5 and ratio of h/b=4
We appreciate some important aspects related with the behaviour of the graphics while
the rate h/b increases from 1,25 to 4. Note that for the presented ranges section h/b,
from 1 to 2 are the commercial ones, and after this (h/b=3 and 4) would be for welded
sections.
First of all, as a general view, we see that on the first graphic, all the families of b’s are
completely superposed. So in that way we can obtain a critical zone of L/iz for each h/b.
The critical zone is between 40 ≤ L/iz ≤ 80. In addition, for the first ratio h/b=1,25, the
Eurocode has the checking point in L/iz =240 and the British Standard in 400. For the
last commercial section, h/b=2 the Eurocode has the checking point in L/iz =220 and the
British Standard in 260, so the difference is narrower than for square sections. Until that
ratio of the Eurocode is more conservative since lower lengths are allowed, but for
welded sections (h/b=3 and 4) the British standard becomes more conservative and
allows lower ratios of L/iz (see Figure 38 and Figure 39).
As an example of what is explained above, for low ratios of h/b is more difficult to
buckle. If we see Figure 29 that shows h/b=1,25 (it is almost a square cross-section),
needs a L/iz ratio of almost 800 to have a reduction for lateral-torsional buckling of
0,65. This means, for example, a section with iz =0,025, the buckle length will be 20m.
But if we see the last (Figure 39) that shows a rate h/b=4 (it is a very slender cross-
section) needs a rate L/iz about 500 to have the same reduction for lateral-torsional
0,65
0,7
0,75
0,8
0,85
0,9
0,95
1
0 200 400 600 800
χ
L/iz
h/b=4
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,45
0,5
EC-3
BS
base
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
65
buckling. It means that as the same example before, the length will be 12,5m, slightly
higher than half the previous one.
Secondly, the recommendations from Eurocode and British Standard look a little bit
different. We see how Eurocode 3 in general is more conservative than British Standard.
Higher lengths are allowed for h/b ratios up to 2 in the British Standard, whereas for
higher h/b ratios the trend is reversed and the lengths allowed by Eurocode 3 are higher
than the British Standard. The Eurocode fixes the checking point for cases between
(approximate) 0,85≤ χ ≤ 0,87 (corresponding to ≤ 0,4), as he says that it is not
necessary to check the lateral-torsional buckling of square and rectangular cross-
sections.
On the other hand, British Standard fix this value in between 0,77 < χ < 0,82 (it depends
on the rate given by this standard as it is shown on graphics). In that way, we can define
that the Eurocode is more conservative than the British Standard. However for high
ratios like h/b=3 this is not absolutely true, both of them share the checking point and
then (ratios higher to h/b=3) British Standards become more conservative than the
Eurocode as happened before this rate.
But indeed, we can make a whole lecture of the graphics obtained. Really, the
difference appreciated in both standards is not extremely big for high ratios of h/b as 3
or 4. It is true that for low ratios of h/b (commercial ones) we are taking even 0,1 point
(0,85 to 0,75) of reduction factor for lateral-torsional buckling less in British Standard
than Eurocode. But if we see high ratios of h/b the difference of both standards is 0,03
points only (0,85 to 0,88). So roughly speaking, for low h/b ratios the difference
between the two standards is quite important, doubling the length of the beam for which
no checking is needed (from L/iz ≈ 300 for Eurocode 3 to 600 for British Standard). The
difference is gradually reduced while increasing the ratio h/b, becoming negligible for
high values of h/b.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
66
4. Parametric Study of Lateral-Torsional Buckling with a Finite
Element Method Program
Hereafter a comparison of the results for critical moment from mathematical formulas
calculated with VBA (see chapter 3) and obtained results with a finite element method
program as carried out for some selected cases. The point is to compare the results of
different parameters such as critical moment. The program used is a commercial
program of structural analysis called SAP2000 v14, which is available on computers at
the School of Civil Engineering (UPC).
4.1. Linear Buckling Analysis with Finite Element Method
Buckling theory presumes the existence of bifurcation point. Consider the figure 1:
Figure 40: Bifurcation Graphic
Where there are two equilibrium configurations possible:
1. The beam could remain straight (primary path)
2. The beam could buckle (secondary path)
A bifurcation point exists if the beam is perfectly straight, perfectly uniform, perfectly
free of end moments and lateral loads, and forces are perfectly centred and perfectly
axial. In addition, the Mcr shown on the figure 1, represents the ideal critical moment
explained in chapter 1.
However, in reality, there are always imperfections, whose magnitude we denote by e.
If e ≠ 0, the beam displays no bifurcation point and structures in generally display limit
points. Curves corresponding to linear buckling analysis with finite elements allow
finding the critical ideal load and bifurcation points. In order to get the influence of
imperfections, residual stress and so on, we should perform a non-linear analysis, but
we are focusing in linear buckling analysis only.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
67
Critical buckling loads and associate buckling modes can be obtained by performing a
linear buckling analysis. Then, second order terms of the Green-Lagrange strain are not
negligible and the geometric stiffness matrix Kσ is obtained in similar way in which
stiffness matrix K is obtained from first order terms.
So the linear stiffness matrix [Oñate, 1992] is the following one:
∫ [ ] [ ][ ]
Where:
- K only depends on the geometry of the element, only take into account first
order terms
- B is the matrix that represents state of stress
- D represents the mechanical properties (EA)
And the second order terms stiffness matrix [Cook , Malkus, Plesha, 1989] is the
following one:
∫ [ ] [ ][ ]
Where:
- Kσ is defined by an element’s geometry, displacement field, and state of stress
distribution.
- G is obtained from a shape functions by appropriate differentiation taking into
account second order terms.
- S is the matrix that represents state of stress.
So in that way, we have the next expression:
[ ]
Let’s consider a reference state previous to buckling Fref, with its corresponding Kσ,ref
and define the buckling state as:
F=αcr· Fref and Kσ = αcr· Kσ,ref
When buckling occurs, there is an increment of displacements without a change of
applied load, so combining the equilibrium equation with and equations for
buckling state, we obtain the generalized eigenvalue problem:
[ ]
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
68
When solving it, dV gives us the shape of the buckling mode (not the amplitude), and
the critical buckling load is given by:
Fcr= αcr · Fref
All the previous expressions are particularized for shells considering that the analysis
with SAP2000 v14 has been done with shells elements.
4.2. Analysis of Different Cases with SAP2000 v14
In this section it is developed the modelling of different cases studied that have been
compared with VBA analytical mathematical formulation. At the end, there is a resume
table with the results obtained. The commercial program SAP2000 v14 [Hernández,
2009] is an integrated software for structural analysis and design.
We want to assess that the values of critical moment obtained are similar to the critical
moment given by Eurocode 3, and also to check that the VBA program is according due
to the fact that we obtain similar values.
4.2.1. Boundary Conditions
All the cases have been studied with the following boundary conditions:
a) Simply supported as a fork. This kind of restraint allows movement into X
direction but not into Y and Z, being X the longitudinal axis and Z the vertical as
usual. For numerical stability one node in the middle of the cross-section on the
top. But one in the middle the cross section on the top flange is completely
fixed. All supports allow the rotation.
b) Uniform Bending Moment, materialized as compression forces on top flange
and tension in the low flange.
c) Shells elements (type shell-thick), due to the cross-sections are a slender
elements and shells’ behaviour is the one that give approximate results.
d) Steel S355 from the EN-1993-1-1
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
69
Picture 5: Sample of boundary conditions
Picture 6: Sample of one whole beam as introduced in the program completely fixed. All supports allow the rotation.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
70
4.2.2. Particular Cases to Study
The cases we have carried out are a square cross-section (h/b=1) and two rectangular
cross-sections (h/b=2 and h/b=3) because they can be a commercial sections (h/b=1 to
2) or welded sections (h/b=3) and is interesting compare the behaviour with VBA and
SAP2000. All of them have the boundary conditions explained above with a uniform
bending moment applied on the extremes of the beam. In this way we could state that
the two ways drive to the same final results. The cross-sections used have the next
configuration:
Height x Base x Thickness (Length)
The different cases analyzed are the followings:
1st case: Square cross-section (h/b=1)
We want to know the results obtained with the SAP2000 v14 and compare them with
the ones obtained with VBA and get the relative error.
Cross section 0,2x0,2x0,015 (L=20m) Values Units
Med 500 KNm
α 3,13 -
Mcr,SAP 1565 KNm
Mcr,VBA 1532,94 KNm
Relative error 2,09 % Table 12: Obtained results (square cross-section, h/b=1)
The obtained results shown on the table confirm how the difference between the
formulation given by EN-1993-1-1 and calculated with VBA in section 3 is very close
to the value calculated with SAP2000. The factor α is the one explained in section 4.1
on the problem to the eigenvalues, so Mcr = Med· α. In addition, the relative error can be
taken as a lack of more discretized elements (2% of relative error), it means that if we
make a finer mesh, the relative error will decrease due to the fact that we will get more
accurate results.
Figure 41: square cross-section (0,2x0,2x0,15 with L=20)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
71
The buckling mode obtained with the boundary conditions explained above can be seen
below:
Figure 42: First buckling mode for RHS 0,2x0,2x0,015 with L = 20 m. αcr = 3,13
Figure 43: Longitudinal view of the first buckling mode
Figure 44: Front view of the first buckling mode
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
72
2nd
case: Rectangular cross-section (h/b=2)
We want to know the results obtained with the SAP2000 v14 and compare them with
the ones obtained with VBA and get the relative error.
Cross section 0,2x0,1x0,015 (L=20m) Values Units
Med 500 KNm
α 0,724 -
Mcr,SAP 370 KNm
Mcr,VBA 350,05 KNm
Relative Error 3,41 % Table 13: Obtained results (rectangular cross-section, h/b=2)
As in the first case, the results obtained with SAP2000 are quite close to the ones
obtained with VBA.
Figure 45: rectangular cross-section (0,2x0,1x0,15 with L=20)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
73
The buckling mode obtained with the boundary conditions explained above can be seen
below:
Figure 46: First buckling mode for RHS 0,2x0,1x0,015 with L = 20 m. αcr = 0,724
Figure 47: Longitudinal view of the first buckling mode
Figure 48: Front view of the first buckling mode
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
74
3rd
case: Rectangular cross-section (h/b=3)
We want to know the results obtained with the SAP2000 v14 and compare them with
the ones obtained with VBA and get the relative error.
Cross section 0,3x0,1x0,015 (L=20m) Values Units
Med 770,27 KNm
α 0,766 -
Mcr,SAP 590,54 KNm
Mcr,VBA 565,02 KNm
Err_rel 4,52 % Table 14: Obtained results (rectangular cross-section, h/b=3)
Once again, the obtained results are very similar to the ones calculated with the VBA.
Figure 49: rectangular cross-section (0,3x0,1x0,15 with L=20)
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
75
The buckling mode obtained with the boundary conditions explained above can be seen
below:
Figure 50: First buckling mode for RHS 0,2x0,1x0,015 with L = 20 m. αcr = 0,724
Figure 51: Longitudinal view of the first buckling mode
Figure 52: Front view of the first buckling mode
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
76
5. Conclusions
After all the chapters above, hereafter are explained and justified the conclusions of this
work about a parametrical study of lateral-torsional buckling in closed cross-sections
(square and rectangular).
First of all, we have noticed that each standard or guide design uses their own
guidelines to check the lateral-torsional buckling on closed cross-sections, but American
Standard (AISC) does not have any guidance to face the phenomenon on these type of
sections.
Secondly, if we take the Eurocode 3 to compare with British Standard and CIDECT in a
parametrical study (done in section 3.2.2 and 3.2.3) we noticed that the difference from
the limit checking point (before which it is not necessary to check the lateral-torsional
buckling) is relatively small and remains more or less the same if compared to the
CIDECT recommendations. However if the comparison is made considering the British
Standard the difference between the two standards is quite relevant, doubling the length
of the beam for which no checking is needed for low h/b ratios. This difference is
gradually reduced while increasing the h/b ratio, becoming negligible for high values of
h/b.
Also, the boundary conditions studied are unfavourable, because the coefficient applied
C1 which depends on the load application and restraints is taken as 1 (uniform bending
moment and simply supported), so if we modify any condition (clamped, uniform load
and so on), the coefficient can take values above 1, and then the increase on the critical
moment would improve the lateral-torsional buckling checking as the slenderness is
reduced.
5.1. Conclusions of the calibration of the VBA model
After obtaining the different graphics on how affect each parameter, we have noticed
that the thickness does not have much influence, note that on Variable Flange Thickness
section, the behaviour of the χ while thickness increases is insignificant.
This is the reason why we have not focused on its influence.
5.2. Conclusions of the parametric study
First conclusion of that section is that the graphics of a fixed h/b and curves with
different bases, give more information than the ones that b is fixed and have curved
with different h/b. The reason is because if you analyze the first case, all the curves are
superposed and you can get critical limits of L/b, L/h or L/iz with accuracy. In addition
the ratios where lateral-torsional buckling starts to be found are:
Case L/b
Almost L/b=30 for low ratios of h/b as 1 and almost L/b=20 for high ratios of h/b as 3.
But it is true that the reduction factor for lateral-torsional buckling is very small. Note
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
77
that the case of h/b=3, the checking point is for a ratio L/b=20, but if we take a ratio
L/b=40, the χ affected is 0,94. So it is not a relevant case of failure.
Case L/h
In the case of L/h the checking point is quite far from the normal ratios on structures.
Note that critical points are between 25<L/h<35 and the checking point for lateral-
torsional buckling of Eurocode is L/h=100 and L/h=115 for the CIDECT (low ratio
h/b), so are values really unusual on structures. But if we see the case of high ratios h/b
as 2, the checking point of Eurocode and CIDECT are around L/h=40 for a χ=0,85. So it
would be possible to find some structures affected by this analysis.
Case L/iz
In this case the critical point for lateral-torsional buckling between 50 < L/iz < 80 and
the checking points for low ratios h/b for the Eurocode is L/iz 300 and L/iz 600 for the
British Standard, so it is unlikely to have this ratios in normal structures. It happens the
same for high ratios h/b as 3 or 4 that checking points for the Eurocode is L/iz 150 and
L/iz 170 for the British Standard that continues being unusual. But in case it happens,
the χ is 0,85 and should be checked.
5.3. Conclusions on comparison with British Standard and CIDECT
Firstly, the critical values are between 50 < L/iz <80 and an important conclusion with
the British Standard is that for commercial sections (h/b from 1,25 to 2) the Eurocode is
more conservative than British Standard. In fact for this commercial section, British
Standard even allows twice the ratio of L/iz (specially for h/b=1,25), and this means that
reduction factor for lateral-torsional buckling for Eurocode is 0,85 and for the British
Standard is 0,75. But when we see welded profiles (h/b=3 or 4) the difference between
both standards is very insignificant, actually both fix the checking point at χ =0,85.
Although it is true that for ratios of L/iz=100 or 200 are not common in normal
structures.
Secondly, an important conclusion with the CIDECT is that, in contrast with the British
Standard, both checking points are quite close. In addition we can assess that in the
range of h/b studied (from 1 to 2, so all commercials), Eurocode is more conservative,
while CIDECT allows higher ratios L/h until the checking point (but not too much).
Furthermore, the deference between the checking point of Eurocode 3 and CIDECT is
only 0,03 points separated.
5.4. Conclusions of the SAP2000 v14 study
The obtained results with the comparison between SAP 2000v14 and VBA are quite
similar to the critical moment calculated with the VBA code. The relative error for the
studied commercial cross-sections ranges between 2 and 5% which is a very low value,
outweighed by far by the safety factors used in the codes.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
78
5.5. Future Works
As a future studies it would be interesting to know what happen in case of class 4
sections where the cross-section should be reduced to the effective area as it is
explained in EN-1993-1-5: Plated Structural Elements.
Other closed cross-sections remain to be studied, such as the cylindrical hollow sections
that are not really treated on Standards. Also the study could be extended to other
boundary conditions (e.g. clamped, continuous beams and so on), and loading
conditions (e.g. punctual load in the middle, uniform span load among others).
Also, an advanced study could be a non-linear analysis of the lateral-torsional buckling
taking into account initial imperfections, so in that way with the finite element analysis,
the load obtained would not be the ideal load, but the real buckling path.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
79
6. References
[Acces Steel, 2010] - Acces Steel, 2010.
[AISC, 2010] - AISC & ANSI, 2010 - Specification for Structural Steel
Buildings.
[Baláz and Koleková, 2004] - Baláz I, Koleková Y., 2004 - Factors C1, C2, C3
for computing the elastic critical moment Mcr. In: Zborník VI. sympózia Drevo
v stavebných konstrukciách so zahranicnou úcast’ou, Kocovce.
[Batista and Rodrigues, 1994] – E.M. Batistaa and F.C. Rodrigues, 1994 –
Buckling curves for cold-formed compressed members.
[British Standard, 2001] – British Standard, 2001 – BS5950-1:2000 : Structural
Use of Steelwork in Building
[CIDECT, 1996] - Committee International for the Development and the Study
of Tubular Construction (J.Ronald, K.G. Würker, D. Dutta, J. Wardenier, N.
Yeomans), 1996 – Design Guide 2: Structural Stability of Hollow Sections
[Cook , Malkus, Plesha, 1989] - Clark JW, Hill HN., 1960 - Lateral buckling of
beams. J Structural Division.
[Crew and Macmillan, 1914] - Henry Crew and Alfonso de Salvio Macmillan,
1914, Galileo Galilei “Two Dialogue Concerning Two New Sciences”. English
translation.
[Delatte, 2009] - Delatte Norbert J. Jr., 2009 – Beyond Failure: Case Studies for
Civil Engineers, PhD.
[EN-1993-1-1, 2005] – European Standard, 2005, Eurocode 3: EN-1993-1-1:
Design of steel structures, part 1-1: General Rules and Rules for buildings
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
80
[Galéa, 1981] - Galéa Y, 1981 - Abaques de d´eversement pour profilés laminés.
Construction Méallique.
[Gardner and Nethercot, 2005] – L.Gardner and D.A. Nethercot, 2005: Designers
Handbook to EN-19993-1-1.
[Hernández, 2009] - Eliud Hernández, 2009 - Manual de Aplicación del
Programa SAP2000 v14.
[Lagrange, 1781] - Joseph Louise de Lagrange, 1781– Oeuvres de Lagrange,
vol.2.
[Law and Gardner, 2012] – K.H. Law and L. Gardner, 2012 – Lateral instability
of elliptical hollow section beams.
[Martin and Purkiss, 2008] - Joseph Martin L. and Purkiss J., 2008 – Third
Edition of Structural Design of Steel Work to EN1993 to EN1994.
[Oñate, 1992] - Eugenio Oñate Ibáñez de Navarra, 1992, Cálculo de Estructuras
por el Método de Elementos Finitos, CIMNE (UPC).
[Petroski, 1994] - Henry Petroski, 1994 - Design Paradigms: Case Histories of
Error and Judgment in Engineering.
[Prandtl, 1899] - Ludwig Prandtl, 1899 - Kipperscheinungen, Dissertation.
Munich.
Parametric Study of Lateral-Torsional Buckling in Closed Cross-Sections
Javier Pascual Ramos Degree in Construction Engineering - UPC 2013
81
[Seaburg and Carter, 2003] - Seaburg P.A., Carter C. J., 2003 - Torsional
Analysis of Structural Steel Members – College of Engineering, Purdue
University.
[Siev, 1966] - Siev, A., 1966 - Torsion in Closed Sections, “Engineering
Journey, Vol3, No1, (January), pp 41-54, AISC, Chicago, IL.
[Trahair, 2010]- Trahair N.S., 2010, Research report R912, Sidney school of
civil engineering.
[Timoshenko, 1953] - Stephen P. Timoshenko, 1899 – History of Strength of
Materials. McGraw Hill 1953 republished.
[Timoshenko and Gere, 1961] - Stephen P. Timoshenko and James M. Gere,
1961 – Theory of Elastic Stability