Vous êtes sur la page 1sur 5

Term Paper for CVG5143-Advanced Structural Steel Design

Instructor M. Mohareb

The effect of partial end restraints on the lateral torsional


buckling capacity of doubly symmetric I-beams
Yassine Kandoussi
University of Ottawa, Ontario, Canada
ABSTRACT: The evaluation of the critical elastic moment for lateral torsion buckling is mainly established for beams
with full lateral and torsional restraints. However, many connections in practice exhibit partial ends restraints. Hence, the
evaluation of the critical moment based on the simplifying Canadian Code CSA S16 (2014) equation yields unconservative
results. This paper investigates the effect of partial end restraining on the lateral torsional capacity of doubly symmetric I
beams. The finite Element Method is used to conduct a parametric study that takes into account the real torsional stiffness
of the end connections. Then, the Finite Element results are compared to the Codes values. In the same manner, a mesh
sensitivity analysis is performed to choose the optimal discretization setting for the design examples. Also, the effect of
warping restraint is studied. Then, the trend of variation the critical moment when the warping fixity factor changes is pre-
sented.

factor which takes into account non-uniform moment


1 INTRODUCTION AND SCOPE distributions.

Partial end restraints are encountered in many structural The goal of this paper is to demonstrate the need to take
applications. Flush End Plates, Web cleats and Open into account the real stiffness of end connections. Also,
Web stiffeners illustrate the presence of partial end re- it provides a mean to derive the critical moment for dif-
straints in buildings. Lateral torsional buckling is mainly ferent stiffness values at the support. The critical mo-
a serious problem for long beams. This instability caus- ment variation is shown in relationship with the torsion-
es the beam to undergo a lateral displacement and a al stiffness factor. In the same manner, the variation is
twist. Also, this phenomenon is amplified for open sec- studied when the warping fixity factor is varied.
tion beams. I-Beams are massively used in steel fram-
ings.
2 LITERATURE REVIEW
The classical solution for the lateral torsional buckling
is established using the energy theorem, the classical 2.1 General
critical elastic moment M u is given in Eq1. However,
the beam is considered as simply supported In this literature review, the focus is placed on the effect
'' 0 and of the presence of partial end restraints on the lateral
torsional buckling capacity of doubly symmetric I-
0
. ''
beams.

E 2.2 The study of Amara et al. (2008)


Mu EI yy GJ ( )2 I y I ww (1)
L L
In this study, the influence of partial end restraints on
Where E is the modulus of elasticity, Iyy is the moment the lateral torsional buckling capacity is investigated. In
of inertia about the weak axis, G is the shear modulus fact, a numerical analysis based on the finite element
and Iww is the warping constant. method was performed. In addition, k z and kare de-
rived. They take into account the effect of the partial
lateral restraint and twist respectively. These coefficient
It is important to state that the previous formulation was are used to modify the Eurocode formula for the critical
established using the Vlasov theory. Hence, two as- moment.
sumptions are considered to derive the previous expres-
sion. Firstly, the beam distortion is neglected. Also, the The study assumes that the loads are applied at the shear
shear strains are considered to be zero on the middle center, so no load position effect is considered. Also, the
surface of the beam. interaction between the effect of k and k z is not inves-
tigated. Furthermore, the warping torsion variation is
Also, Eq1. is developed for a uniform moment distribu- not considered.
tion. For this reason, the Canadian code CSA S16
(2014) provides a tool to calculate the moment gradient The paper shows that the variation of k 1 is increas-
ing Linearly if plotted against the ratio of M crto M cr 0 .
M cr 0 is the reference Critical moment and M cr is the The authors suggested that a difference of 34% is ob-
actual critical moment. served between the free warping and the fully prevented
In the same manner, it is showed that the critical mo- warping case. The previous result is derived for a mid-
ment varies when changing the lateral bending stiffness. span loading scheme. Also, the study highlighted the
However, the change is not linear and significant as it is Crucial load position effect. The critical moment was
the case for the torsional stiffness parameter. reduced by 29% when the top flange was loaded.

2.3 The study of ivner (2012)


2.6 The study of Pi and Trahair (2000)
In this study, the author explores the influence of end
plates introduced in beam-column joints on the evalua- The authors studied the effect of end warping restraints
tion of the critical moment. As a matter of fact, the in- on the elastic lateral torsional buckling capacity of
fluence of the warping degree of freedom is particularly beams. It should be mentioned that the authors consider
studied. The main goal of the study is to derive the k w the web distortion along with the warping restraints.
factor introduced in the Eurocode formula. This coeffi- The authors used the Finite element model developed by
cient is given in the code only for a free and full warp- Pi and Trahair (1997). In addition, the local buckling is
ing fixation. The study was limited for the doubly sym- not incorporated in the model. Two beams with and
metric beams. Also, the load position effect is not without overhangs were used (h=640 mm, L=4m;
studied. For the columns, both IPE and HEB were stud- h=640 mm, L=4.8 m).
ied. Also, IPE beams were chosen.
The study found that there is a 54% difference in the
It was found that the increase of the torsional stiffness buckling capacity for a full warping end restrained beam
beyond a certain limit didnt change significantly the re- and a free one. In the same manner, the authors provid-
sults. There was a 15% increase of the critical moment ed effective values for the torsional and warping rigidi-
when the fork support conditions were taken into ac- ty. These values can be used in a suggested formula for
count. In addition, the torsional stiffness is maximum the determination of the elastic critical buckling mo-
when a combination of HEB columns and IPE beams ment.
are used.
2.4 The solution of CAN-CSA S16 (2014)
2.4 The study of Lopez et al (2006)
The Canadian code provides a formula for the determi-
In this study, the coupled effect between the bending nation of the elastic critical moment for doubly symmet-
moment and support conditions on the critical moment ric I-sections. However, the codes formula is on a
for buckling is investigated. In fact, the paper compares simply supported beam with regards to the twist and lat-
the values with traditional design code values focusing eral displacement. Also, the formula takes into account
on the Eurocode. The paper studies a doubly symmetric the non-uniform moment distribution. For this reason, a
beam. The problem is implemented numerical using Fi- moment gradient factor is calculated based on the real
nite Difference Method. A closed form solution is given moment distribution in a beam.
for the calculation of the moment gradient factor. The
paper deals mainly with free and prevented lateral bend- 2.7 Comparative Summary
ing and warping at each end. An enhanced formula for
the moment gradient factor is suggested. The full and Table illustrates the literature review assumptions and
free state of the lateral displacement and warping are in- results.
corporated into the formula.
Study Meth- Support type Load
2.5 The study of Piotrowski and Szychowski (2015) odolo- Posi-
gy tion
This study deals with the effect of an elastic warping re- Effect
straint on the lateral torsional buckling capacity of Amara et al. Finite Partial restraints for No
beams. The Energy method is used to derive the elastic (2008) Ele- lateral and torsional
critical moment. Then, power polynomials are used to ment degrees of freedom
estimate the twist angle function. It should be noted that Analy-
this study is limited for doubly-symmetric I-sections. sis
Also, the load position effect is taken into account. The ivner Finite End Plates, Partial No
authors presented results for the following loading (2012) Ele- restraints for lateral
schemes: 1) midspan concentrated load 2) uniformly ment and torsional de-
distributed load 3) triangular distributed load. Analy- grees of freedom
sis
Lopez et al Finite Free and Prevented No
1
(2006) Differ- lateral and torsional b ub 1*8
T
{[ K E ]8*8 [ KG ]} ub 8*1 (4)
ence degrees of freedom 2
Where ub 1*8 1 1 1 1 2 2 2 2
T ' ' ' '
Method
Piotrowski Finite Partial restraints for Yes
and Szy- Ele- lateral and torsional 1 , 1 are the displacement and twist angle for the
chowski ment degrees of freedom first node respectively.
(2015) Method [ K E ]8*8 is the elastic stiffness matrix for an element
[ KG ]8*8 is the geometric stiffness matrix for an ele-
Pi and Tra- Finite Partial restraints for No ment.
hair (2000) Ele- lateral and torsional
ment degrees of freedom From Eq. 4, evoking the stationarity condition of the to-
Method tal potential energy, i.e., 0 , one obtains the fol-
CAN-CSA Analyt- Simply supported No lowing condition :
S16 (2014) ical for lateral and twist
degrees of freedom
{[ K E ]8*8 [ KG ]} ub 8*1 {0}8*1 (5)
Table 1 : Comparative summary of the litterature review
articles Eq. 5 is representative of a typical eigenvalue problem.
3 OVERVIEW OF SOLUTIONS AND Before solving the eigenvalue problem, boundary condi-
METHODOLGIES tions are to be defined. This paper focuses on the effect
of the change of boundary conditions on the solution of
the eigenvalue problem.
The first step of the analysis is to write the total poten-
tial energy during buckling: According to CAN-CSA S16 (2009), the critical mo-
b U b Vb ment expression for a laterally unsupported beam
L L L
1 1 1
E
2 0
EI yy ( '' ) 2 dz GJ ( ' ) 2 dz EI ww ( '' ) 2 dz
20 20 Mu EI yy GJ ( )2 I y I ww (6)
L L L
M ( z ) '' dz (2) In which 2 is the moment gradient factor. This factor
0 takes into account the non-uniform moment distribution.
Its calculation is based on the assumption of a simply
Where U b is the strain energy during buckling, Vb is the supported beam with regards to lateral displacement and
load potential during buckling. is the lateral dis- twist.
placement while is the angle of twist for the section.
M(z) is the moment caused by the loads. In real cases, end connections may provide only partial
restraints. Hence, using the code formulas may yield un-
It should be stated that Eq2. Is established using the conservative results.
Vlasov theory. Hence, two important assumptions are
considered. The first is that section is considered to act To enhance the boundary conditions model at the end
as a rigid disk. So no distortional movement is allowed. supports, lateral and rotational springs are introduced.
Also, the shear strain along the middle surface is con- Fig illustrates the boundary conditions model
sidered to be negligible.
Then a linear variation of the moment M(z) is assumed.
So the destabilizing term becomes:
L
z z
M ( z) ( )dz [ M (1 L ) M ( )} ''dz
''
1 2
0
L
(3)
M 1 is the moment at the first node. M 2 is the moment
at the second node Eq3. Also, Four degrees per node are
considered for the analysis. , , , .
' '

Figure 1 : I-Beam with lateral and tor-


Then, the displacement and twist angle functions are re- sional springs
lated to the nodal displacements and rotations. This is
achieved by using shape functions. Hermite polynomi- K y denotes the lateral stiffness in the y direction. In the
als are used in this case. same manner, K is the rotational stiffness of the end
From Eq.2, substituting the displacement and twist restraint.
functions and rearranging terms, one obtains the follow-
ing expression for the total potential energy: The previous terms are added in the elastic stiffness ma-
trix. This is due to the addition of the load potential of
the springs induced forces to the total potential energy
expression. K should be added to the corresponding
lateral degree of freedom in the elastic stiffness matrix.
Likewise, is added to the corresponding twist degree of
Figure 2 : beam with full end restraints- Ideal
freedom.
Case
Using a FEM code to solve eq. 5, values for are ob- We calculate the buckling capacity using both the code
tained. stands for the load multiplier by which the equation and the Finite Element Method code in Matlab.
nominal 1KN is applied to obtain the elastic critical
moment. Its minimum value characterizes the elastic In the case illustrated by fig X., the FEM code gives the
critical buckling capacity of the beam. From this param- value of 17.2 . Thus, the corresponding critical mo-
eter, the critical moment can be calculated. ment is M cr1 55.8 KN m

To characterize the effect of partial end restraining on Regarding the code, using the gradient factor, the value
the lateral-torsional buckling, values for K are varied. of M cr 2 52.2KN m is obtained.
Then, results are compared to the traditional values pro-
vided by the code equations.
5.2 Example 2- Effect of End Restraint on the
Buckling Load Torsional Stiffness
4 MESH SENSITIVITY ANALYSIS
To determine an adequate mesh for the numerical study,
a mesh sensitivity analysis is conducted. To achieve this
goal, a W 250*45 beam is chosen with 13 m span. Also,
a concentrated load is applied at the middle. The beam
is simply supported with regards to lateral displacement
and twisting. A Finite Element program is used to ob-
tain the numerical solutions.
The results are illustrated in table X. Figure 3 : Effect of the Torsional Stiffness on

(1) (3)
Number of subdivi- Elastic Critical Then, the rotational stiffness of the left support is var-
sions Moment KN*m ied. The stiffness is changed from a zero value to fully
2 56.2 restrained represented by Fig3. In this first example, the
4 55.9 lateral displacement is restrained. The table below rep-
10 55.8 resents the key results for the previous beam:
20 55.8
40 55.8
Table 2 : Mesh sensitivity analysis (1) (2) (3) (4)
K 1 CSA S16 Finite Ele- Percentage
It can be inferred from the previous results that 10 sub-
(KN*m/r (2014) ment Pro- difference
divisions can be chosen for our analysis cases. (1)
ad) gram (2) ((3)-(2))/(3)
Free 52.2 5.94e-5 8.79e+7 %
Also, it is important to highlight that the lateral torsional
2 52.2 27.5 89%
capacity of beams can be overestimated if the mesh isnt
sufficient. 10 52.2 46.0 13%
50 52.2 53.4 2.24%
5 DESIGN EXAMPLES 90 52.2 54.5 4.22%
To illustrate the effect of partial end restraints on the 150 52.2 54.81 4.76%
buckling capacity of beams, two situations are studied. 200 52.2 55.2 5.34%
The first is a simply supported beam. The second one is 300 52.2 55.4 5.77%
partially restrained with regards to the twist degree of 400 52.2 55.5 6.45%
freedom. Also, the span of the beam is L=13m. Full 52.2 55.8 6%
Table 3 : The effect of torsional stiffness on Mcr
5.1 Example 1 Ideal Case -Full End Restraints (1)
Based on Eq. 6 * The moment gradient factor
In this reference example, the supports are considered (2)
Based on Eq. 5
fully restrained with regards to the lateral displacement
and twist degrees of freedom Fig2. Based on the previous results, the torsional stiffness of
the end support has a significant effect on the buckling
capacity. In fact, the critical moment decreases as the
torsional stiffness is reduced. The difference can reach cal moment of the beam. The critical moment is en-
13% compared to the result given by the code equation hanced when warping restraints are added.
if only 10% of the torsional stiffness is added by the
support. Also, it is observed that the buckling capacity
6 CONCLUSIONS
doesnt change significantly when increasing K by
100 %. This study highlighted the effect of partial end restraints
on the lateral torsional buckling capacity of beams. In
The torsional fixity factor can be introduced to char- fact, it has been demonstrated that the torsional stiffness
acterize the response of doubly symmetric I-beams with at supports plays a major role for the evaluation of the
regards to the lateral torsional buckling behavior. The critical moment. Also, the variation of the lateral stiff-
ratio of the actual critical moment to the reference criti- ness doesnt yield significant changes for the lateral tor-
cal moment is plotted against the fixity factor. is 0 sional buckling.
when no torsional stiffness is provided and 1 when it is
fully restrained in the twist degree of freedom. As a matter of fact, using a torsional stiffness of 10
KN*m/rad yields a value that differs 13% from the
codes. In addition, the codes formula is conservative
The influence of the fixity factor on the
for the simply supported case. However, it is uncon-
critical moment
servative when partial restraints are used. The fixity fac-
critical moment to the reference
Mcr/Mcr0 the ratio of actual

1,5 tor can be used to quantify the decrease in the critical


1 moment. For a fixity factor of 0.12 a 5% decrease in the
critical moment is observed from reference full re-
one

0,5 strained case.


0
0 0,5 1 1,5 Regarding the effect of the warping restraints, the criti-
Fixity Factor cal moment is enhanced when warping restraints are in-
troduced. In fact, a 0.26 fixity factor improves the criti-
Figure 4 : The influence of the torsional fixity factor cal moment by 7%. Also, the critical moment doesnt
change significantly when the fixation factor is beyond
5.3 Example 2- Effect of End Restraint on Buckling 0.83. Also, there is a 17 % difference between the code
Load Partial Warping restraint formula and the full warping restraint value.

To emphasize the effect of the warping stiffness on the


buckling capacity, the K w factor is introduced. In fact, 7 REFERENCES
it characterized the degree of warping fixation at the
ends of the beam. It is 0 when the beam is free to warp 1. Amara, S., Kerdal, D.E. and Jaspart J.P. (2008). Effect of end con-
and 1 when it is blocked from warping. nection restraint on the stability of steel beams in bending. Advanced
Steel Construction Journal, Vol. 4, No. 3, pp. 243-259
Using the Finite element program, the critical moment
is evaluated for an elastic warping restraint at the ends. 2. Lopez, A.,Yong D.J. and Serna M.A. (2006). Lateral-torsional
Hence, Table represents the variation of the ratio of the buckling of steel beams: A general expression for the moment gradi-
actual critical moment to the reference critical moment ent factor. International Colloquium on stability and ductility of steel
is plotted against the fixity factor K w structures, Lisbon, Portugal, September 6-8.

3. Pi, Y. and Trahair, N.S. (2000). Distortion and warping at beam


(1) (2) (3) (4) supports. Journal of Structural Engineering, 126 (11), pp 1279-1287
Kw CSA S16 Finite Ele- Percentage
(2014) ment Pro- difference 4. Piotrowski, R. and Szychowski, A. (2015). Lateral torsional
(1)
gram (2) ((3)-(2))/(3) buckling of beams elastically restrained against warping at supports.
0 52.2 5.94e-5 8.79e+7 % De Gruyter Open Journals, Archives of Civil Enginneering, Vol. LXI,
0.26 52.2 58.08 10% ISSUE 4.
0.33 52.2 60.4 13%
0.5 52.2 61.8 15% 5. Rusul, H. (2013). Distortional Lateral Torsional Buckling Analysis
for Beams of Wide Flange Cross-Section. (Doctoral Thesis, University
0.66 52.2 62.9 17% Of Ottawa, Ottawa, Canada).
0.83 52.2 63.1 17.2%
1 52.2 63.2 17.4 6. Zivner, T.J. (2012). The influence of constructional detail to lateral-
Table 4 : effect of Warping fixity on Mcr torsional buckling of beams. ELSEVIER JOURNAL, Procedia
Engineering 40, pp 504-509

The previous results demonstrate that the degree of


warping restraint has a significant influence on the criti-

Vous aimerez peut-être aussi