Int. J. Structural Engineering, Vol. 4, No.

4, 2013 273

A unified approach for assessment of second-order

effects and sway buckling strength in steel portal
frames

George E. Mageirou*, Charis J. Gantes,

Kaliroi P. Markou and Charis G. Bouras
Department of Structural Engineering,
Laboratory of Metal Structures,
School of Civil Engineering,
National Technical University of Athens,
3 Zefxippou Steet, Gr-16345 Helioupolis, Athens, Greece
Fax: +30-210-7723442
E-mail: geomageirou@gmail.com
E-mail: chgantes@central.ntua.gr
E-mail: imarkou@omete.gr
E-mail: harrisbouras@yahoo.gr
*Corresponding author

Abstract: The objective of this paper is to propose a criterion for the

importance of second-order effects, the choice of the appropriate analysis
method and the buckling strength of steel portal frames. To that effect, a
model of a column with rotational springs at the two end nodes as well as a
translational spring at the top node is proposed. Using the above model
analytical expressions for evaluation of the buckling strength of columns in
multi-story sway, non-sway as well as partially-sway steel frames are derived.
The importance of second-order effects is derived using the same model by
proposing a set of graphs providing the ratio of second order moment to
first order moment of the frame columns, depending on the boundary
conditions. These graphs are used for the choice of a suitable analysis method.
Comparisons to geometrically non-linear analysis results for several example
frames are used for verification.

Keywords: steel frames; second order effects; analysis methods; buckling;

design criteria; amplification factor.

Reference to this paper should be made as follows: Mageirou, G.E.,

Gantes, C.J., Markou, K.P. and Bouras, C.G. (2013) ‘A unified approach for
assessment of second-order effects and sway buckling strength in steel portal
frames’, Int. J. Structural Engineering, Vol. 4, No. 4, pp.273–294.

Biographical notes: George E. Mageirou studied at National Technical

University of Athens, Faculty of Civil Engineering. He received his MSc in
Structural Analysis and Design from National Technical University of Athens
and his PhD from Laboratory of Steel Structures, Division of Structural
Engineering, Faculty of Civil Engineering, National Technical University
of Athens. His research area focuses on the contribution to the design of
multi-story steel frames against flexural buckling. His scientific specialisation
is advanced analysis and design of structures using computer technology, steel
structures, stability of structures, method of finite element for structures with
non-linear behaviour, and computational mechanics. He is a Structural

Copyright © 2013 Inderscience Enterprises Ltd.

274 G.E. Mageirou et al.

Designer Engineer and Adjunct Professor in the Air Force Academy, Division
of Engineering, Department of Civil Engineering and Adjunct Professor in the
Technological Educational Institute.

Charis J. Gantes studied at National Technical University of Athens, Faculty of

Civil Engineering. He received his MSc in Civil Engineering and PhD in
Structural Engineering from Massachusetts Institute of Technology (ΜΙΤ). His
research area is design methodology for deployable structures. His scientific
specialisation is advanced analysis and design of structures using computer
technology, deployable structures, steel structures, steel connections, stability
of structures, method of finite element for structures with non-linear behaviour,
and computational mechanics. He is a Structural Designer Engineer and
Professor in National Technical University of Athens, Faculty of Civil
Engineering.

Kaliroi P. Markou studied at National Technical University of Athens, Faculty

of Civil Engineering. He received his MSc in Structural Analysis and Design
from National Technical University of Athens. His scientific specialisation is
the method of finite element for structures with non-linear behaviour, and
computational mechanics. He is a Structural Designer Engineer.

Charis G. Bouras studied at National Technical University of Athens, Faculty

of Civil Engineering. He received his MSc in Structural Analysis and Design
from National Technical University of Athens. His scientific specialisation is
the method of finite element for structures with non-linear behaviour, and
computational mechanics. He is a Structural Designer Engineer.

1 Introduction

The rapid development in computer hardware and engineering software in recent

years allows the use of more accurate numerical analysis algorithms to obtain the
structural response of steel frames, based on linear or also non-linear – in terms of
large displacements and/or material yielding – procedures (Chen and White, 1995;
White and Hajjar, 1997a, 1997b, 2000; Torkamani et al., 1997; Torkamani and Sonmez,
2001; Chan, 2001; Kim et al., 2004; Rubin and Vogel, 1982). Nevertheless, the large
majority of structural engineers still prefer to use linear analysis, at least at the
preliminary design stage (Chwalla, 1954; Petersen, 1982, 1988; Pflüger, 1978, 1980;
Rubin, 1997; Ghali et al., 2003). For most frame structures this is sufficiently accurate.
However, in many cases the non-linear behaviour of structures, particularly as they
approach their ultimate strength must be accounted for when evaluating their response
characteristics.
Eurocode 3 (1992) classifies steel frames with respect to their sensitivity to second
order effects based on the value of the ratio of their horizontal deflection to the story
height multiplied by the ratio of vertical to horizontal total forces at the base of the story.
If this value is smaller than 0.10, second order effects can be neglected, if it is between
0.10 and 0.25 the results of linear analysis should be multiplied with an approximate
magnification factor, and if it is larger than 0.25 geometrically non-linear analysis must
be carried out. LRFD (1999) proposes the multiplication of the results of linear analysis
by two different amplification factors, considering P-Δ and P-δ effects.
 A unified approach for assessment of second-order effects and sway buckling 275

Moreover, Eurocode 3 (1992) classifies steel portal frames as sway and non-sway in
order to evaluate their buckling strength. Frames are classified as non-sway if they have
have a bracing system capable of reducing the total horizontal translation of the frame by
at least 20% in comparison with the same frame without bracing system. Otherwise
frames are classified as sway. LRFD (1999) follows the same classification.
The objective of the present work is to propose alternative criteria for deciding about
the importance of second order effects, and for choosing a suitable analysis method, of
steel multistory portal frames compatible to the criteria used for evaluating their buckling
strength. For this reason, the same model which is employed for the evaluation of
buckling strength is also used for the choice of a suitable analysis method. To that effect,
columns in multistory frames are modelled as individual. The contribution of members
converging at the bottom and top ends of the column is taken into account by equivalent
springs. Namely, the restriction provided by the other members of the frame to the
rotations of the bottom and top nodes is modelled via rotational springs with constants cb
and ct, respectively, while the resistance provided by the bracing system to relative
transverse translation of the end nodes is modelled via a translational spring with constant
cbr. This is shown schematically in Figure 1. This model has been used extensively by
several investigators for the evaluation of the critical buckling load of the member (for
example, Wood, 1974; Aristizabal-Ochoa, 1994a, 1994b, 1997a, 1997b, 2002, 2003;
Cheong-Siat-Moy, 1991; Gantes and Mageirou, 2005; Mageirou and Gantes, 2004, 2006,
2007), and is adopted by most codes for that purpose. The proposed procedure aims at
establishing a unified criterion for treating second order effects as well as for evaluating
buckling strength.

Figure 1 (a) Steel frame (b) Proposed model of column under investigation

(a) (b)

The stiffness of the bottom (cb) and top (ct) rotational springs is obtained by summing up
the contributions of members converging at the bottom and top end node, respectively:

cb = ∑c
i
b ,i , ct = ∑c
j
t, j (1)

Rotational stiffness coefficients for all possible rotational and translational boundary
conditions at the far end of a converging member, as well as the eventual presence of
axial force, have been proposed in previous work of Gantes and Mageirou (2005).
276 G.E. Mageirou et al.

Distribution factors zb and zt are obtained by non-dimensionalisation of the end

rotational stiffnesses cb and ct with respect to the column’s flexural stiffness cc = 4EIc / h:
cc cc
zb = , zt = (2)
cc + cb cc + ct

The dimensionless stiffness of the translational spring is defined as:

cbr h3
cbr = (3)
EI c

A frame is characterised as non-sway if the stiffness cbr of the bracing system

is sufficiently large, and as sway if this stiffness is negligible. The design codes
(for example, Eurocode 3, 1992; LRFD, 1999) provide the effective length Kh of
columns in sway and non-sway frames via graphs or analytical relations as functions of
the rotational boundary conditions (Task Committee on Effective Length, 1997). The
critical buckling load is then defined as:
π 2 EI c
Pcr = (4)
( Kh) 2

where EIc is the flexural resistance.

This problem has been addressed by many investigators. The work of Wood (1974)
constituted the theoretical basis of EC3. Chen and Wang (1999) modified the LRFD
(1999) moment amplification factor. Kemp (2000) improved the amplification factor
of the Canadian code considering material inelasticity. Petrolito and Legge (1996)
developed a unified non-linear analysis technique for two-dimensional structural frames.
The procedure required minimum discretisation of the structure as a self-adaptive
procedure was used to solve the resulting equations. A new iteration method for problems
which involve interaction between two bodies having separation, bonding, re-boding and
slip at the interface was proposed by El-Sawy and Moore (1996). Boucard et al. (1997)
proposed a new method for estimation the pre- and post-buckling response of a
structure without any continuation technique and with less than ten iterations. A
large-displacement inelastic analysis technique was adopted by Liew et al. (1997) to
compute the maximum strength of spatial structures considering both member and
structure instability. A new stability design method for steel plane frames using
second-order elastic analysis was proposed by Oda and Usami (1997). To this effect, a
new formula for the equivalent initial deflection and a methodology using the curvature
of the buckling mode were introduced to allow for a systematic application of this
formula to arbitrary shaped irregular frames. A numerical procedure was presented by
Torkamani et al. (1997) for the solution of geometrically non-linear problems considering
second-order effects. A second-order analysis of frames with members subjected to
gravitational distributed axial loads was carried out by Zhou and Chan (1997). Rubin
(1997) proposed a general formula, valid for first- and second-order theory, which yields
correct results for any values of the member axial force. Imai and Frangopol (2000)
obtained the response of geometrically non-linear structures using Lagrangian
formulation for finite element discretisation. Chan (2001) reviewed the work conducted
on non-linear analysis and design of steel frames in the past few decades, from the point
of view of the current computer age. Kim et al. (2001) proposed a new design method for
 A unified approach for assessment of second-order effects and sway buckling 277

three-dimensional frames using advanced analysis. Aschheim and Montes (2003)

proposed a new effective height representing the influence of P-Δ effects of the structure.
yield point spectra were used to indicate the influence of P-Δ effects on the lateral
strength associated with constant ductility demands. Dobson (2003) summarised
alternative methods of accounting for P-Delta effects such as the amplified sway and
the extended effective length methods as well as the analytical methods such as the
pseudo loads and pseudo displacements approaches, the two-circle iterative method and
non-linear analysis. Cheong-Siat-Moy (1986) examined the k-factor paradox for leaning
columns and drew attention on the dependence of buckling strength not only on the
rotational boundary conditions of the member in question but also on the overall
structural system behaviour. Bridge and Fraser (1987) proposed an iterative procedure for
the evaluation of effective length accounting for the presence of axial forces in the
restraining members and thus considering also negative values of rotational stiffness. In
Cheong-Siat-Moy (1991) a fictitious lateral load method is proposed for the evaluation of
the strength of leaning columns. Aristizabal-Ochoa (1994a, 1994c) proposed analytical
relations for the evaluation of effective length for columns in sway, non-sway and
partially sway frames, which are applicable to leaning columns. He also examined the
influence of uniformly distributed axial load in the evaluation of effective length of
columns in sway and partially sway frames (Aristizabal-Ochoa, 1994c). Hellesland and
Bjorhovde (1996b) proposed a new restraint demand factor considering the vertical and
horizontal interaction in member stability terms. They also proposed a method for the
evaluation of effective length, especially in cases where sufficient changes of column or
beam stiffness occur, such as the top or bottom stories of braced frames (Hellesland and
Bjorhovde, 1996a). Kishi et al. (1997) proposed an analytical relation for the evaluation
of effective length of columns with semi-rigid joints in sway frames. Essa (1997)
proposed a design method for the evaluation of effective length for columns in unbraced
multistory frames considering different story drift angles. Cheong-Siat-Moy (1997b)
aimed at the conceptual understanding of the behaviour of leaning columns at the onset of
buckling. He then investigated the behaviour of columns of braced frames (Cheong-Siat-
Moy, 1997a). Aristizabal-Ochoa (1997c) considered load distribution and asymmetry in
member geometry. He derived a useful relation for the bracing’s stiffness so as to
consider the frame as non-sway. He then examined the behaviour of columns with
semi-rigid connections under loads such as those produced by tension cables that always
pass through fixed points or loads applied by rigid links (Aristizabal-Ochoa, 1997a).
Then he investigated the influence of semi-rigid connections for columns with uniform
axial load (Aristizabal-Ochoa, 1997b). In Cheong-Siat-Moy (1999), a general formula for
the evaluation of columns with lateral restraint which can vary from minus infinity to
plus infinity is proposed. Karamanos and Zissopoulou (2002) compared the pertinent
provisions of EC3 and LRFD for sway frames. Aristizabal-Ochoa (2002) examined the
behaviour of three-dimensional multicolumn systems under gravity loads and proposed
easy to use analytical relations for the evaluation of the effective length coefficient for
sway, non-sway and partially sway frames. He also looked at the effect of the column’s
orientation in 3D multi-column systems with sides-way, being twist uninhibited, partially
inhibited or totally inhibited, with semirigid connections (Aristizabal-Ochoa, 2003).
Gantes and Mageirou (2005) and Mageirou and Gantes (2004) proposed improved
stiffness distribution factors for the evaluation of effective buckling lengths in multi-story
frames with different levels of sway ability. Mageirou and Gantes (2006, 2007) also
investigated the buckling behaviour of multi-story sway, non-sway and partially-sway
278 G.E. Mageirou et al.

frames with semi-rigid connections. To that effect, they proposed analytical expressions
and graphs for the evaluation of effective buckling lengths in multi-story frames with
semi-rigid connections.
However, all these studies refer to buckling length evaluation and do not address the
issue of sensitivity to second order effects. The present work considers both of them
using the same model. The buckling strength of the frame as well as the choice of the
appropriate method are related to the values of the dimensionless stiffness of the
translational spring (cbr ) and the distribution factors (zb, zt). This work proposes a set of
graphs for calculating the factor of second order moment to first order moment of a frame
column, depending on boundary conditions. These graphs are used as a criterion for the
choice of an appropriate analysis method. Moreover, this factor can be used as an
amplification factor for the results of the linear analysis if the geometrical non-linearity is
not dominant. Furthermore, the same model is used for the evaluation of buckling
strength. Easy to use analytical relations and corresponding graphs are proposed for the
estimation of the buckling strength of columns (Gantes and Mageirou, 2005; Mageirou
and Gantes, 2004, 2006, 2007). The application of the proposed methodology is very
simple as well as sufficiently accurate for design purposes. Using the above method the
designer is aware of the real behaviour of the structure as well as its buckling strength.

2 Buckling strength of columns in multi-story frames

The analytical expressions for the evaluation of the effective buckling lengths in
multi-story frames with different levels of sway ability, proposed in past research work of
Gantes and Mageirou (2005) and Mageirou and Gantes (2004, 2006, 2007), are briefly
presented below for completeness.

2.1 Sway frames

Consider the simplified model of a column in a sway frame, shown in Figure 2(a)
resulting from the model of Figure 1(b) by omitting the translational spring. Denoting by
w the transverse displacement and by ′ differentiation with respect to the longitudinal
coordinate x, and using the sign convention of Figure 2(d), the equilibrium of this column
in its buckled condition is described by the well-known differential equation:

w′′′′( x) + k 2 w′′( x) = 0 (5)

where

P π
k= = (6)
EI c Kh

The general solution of this differential equation is given by:

w( x ) = A sin(kx) + B cos(kx) + Cx + D (7)

The integration constants A, B, C, and D can be obtained by the boundary conditions at

the two column ends. The four simultaneous equations of the boundary conditions have a
 A unified approach for assessment of second-order effects and sway buckling 279

non-trivial solution for the four unknowns A, Β, C, and D, if the determinant of the
coefficients is equal to zero. This criterion yields the buckling equation for the effective
length factor β:

⎛π⎞ ⎡ ⎤ ⎛π⎞
2
π ⎛π⎞
4 [ zt ( 2 zb − 1) − zb ] cos ⎜ ⎟ + ⎢ zt zb ⎜ ⎟ − 16 ( zt − 1)( zb − 1) ⎥ sin ⎜ ⎟ = 0 (8)
K ⎝K⎠ ⎣ ⎝K⎠ ⎦ ⎝K⎠

where zb and zt are distribution factors obtained from equation (3).

Equation (8) can be solved numerically for the effective length factor K, which is then
substituted into equation (4) to provide the critical buckling load.

2.2 Non-sway frames

The simplified model of a column in a non-sway frame, shown in Figure 2(b), is
considered, resulting from the model of Figure 1(b) by replacing the translational spring
with a roller support. The buckling equation for the effective length factor K is derived,
following the same procedure as above:

⎛π⎞
32 K 3 ( zt − 1)( zb − 1) − 4 K [8 K 2 ( zt − 1)( zb − 1) + ( zt + zb − 2 zt zb ) π 2 ] cos ⎜ ⎟
⎝K⎠
(9)
⎛ ⎞
π
+π ⎡⎣ −16 K 2 + 20 K 2 ( zt + zb ) + zt zb ( π 2 − 24 K 2 ) ⎤⎦ sin ⎜ ⎟ = 0
⎝K⎠

2.3 Partially-sway frames

Finally, considering the simplified model of a column in a partially-sway frame, shown in
Figure 2(c), the evaluation of the effective length factor K is derived from the following
equation:

−32 K 5 cbr ( zt − 1)( zb − 1)

+4 K ⎡⎣8 K 4 cbr ( zt − 1)( zb − 1) + K 2 cbr ( zt + zb − 2 zt zb ) π 2
⎛π⎞
+ ( − zt − zb + 2 zt zb ) π 4 ] cos ⎜ ⎟ (10)
⎝K⎠
+π ⎡⎣ 4 K 4 cbr ( 4 − 5 zt − 5 zb + 6 zt zb ) − 16 K 2 π 2 (1 − zt − zb + zt zb )
⎛π⎞
− K 2 cbr π 2 zt zb + π 4 zt zb ⎤⎦ sin ⎜ ⎟ = 0
⎝K⎠

where

cbr h3
cbr = (11)
EI

Easy to use graphs such as those presented in Figure 3 are obtained from the above
equations.
280 G.E. Mageirou et al.

Figure 2 Model of column in a (a) sway frame, (b) non-sway frame, (c) partially-sway frame and
(d) sign convention used

(a) (b) (c) (d)

Figure 3 Effective buckling length factor K for different levels of frame sway ability
 A unified approach for assessment of second-order effects and sway buckling 281

3 Alternative criterion for the importance of second-order effects

Next, the simplified model of a column in a partially-sway multi-story frame is

considered, shown in Figure 1(b). Two concentrated loads, a vertical PV and a horizontal
load PH, are applied at the top node. The amplification factor α is defined as the ratio of
the moment at the bottom node evaluated by non-linear analysis (MNL) to the first order
moment (ML) at the same node.

M NL
α= (12)
ML

A linear as well as non-linear analysis code is written based on the matrix method
(Dobson, 2003). The non-linear force-displacement equations are given in an incremental
form by:

F = KG ⋅ D (13)

where F and D are the incremental force and displacement vectors respectively, and KG
is the accurate geometrical stiffness matrix (Dinkler et al., 1995):

⎡ EA EA ⎤
⎢ h 0 0 − 0 0 ⎥
h
⎢ ⎥
⎢ N N
k1 − −k2 0 − k1 + − k2 ⎥
⎢ h h ⎥
⎢ ⎥
k3 0 k2 k4 ⎥
KG = ⎢ (14)
⎢ EA ⎥
⎢ 0 0 ⎥
⎢ h ⎥
⎢ N ⎥
⎢ sym k1 − k2 ⎥
⎢ h ⎥
⎢⎣ k3 ⎥⎦

where

2 EI c a ( s − ac)
k1 = F, B = , c = cos λh
h 3
−2(c − 1) − as
EI a(a − s)
k2 = 2c F , D = , s = sin λh
h −2(c − 1) − as
EI c a 2 (1 − c)
k3 = B, F = , α = λh
h −2(c − 1) − as
EI c N
k4 = D, λ = , N = axial force
h EI c

The axial force is evaluated in every step from:

Ni +1 = N i − Qi φi (15)
282 G.E. Mageirou et al.

where

Ni axial force at step i

Ni+1 axial force at step i + 1

Qi secant force at step i

φi rotation at step i.

Figure 4 Magnification factor (second order moment to first order moment) for sway
ability (cbr = 0), (a) PV* = 0.01, PH = PV / 2 (b) PV* = 0.01, PH = PV / 3 (c) PV* = 0.025,
PH = PV / 2 (d) PV* = 0.025, PH = PV / 3 (e) PV* = 0.05, PH = PV / 2 (f) PV* = 0.05,
PH = PV / 3 (g) PV* = 0.075, PH = PV / 2 (h) PV* = 0.075, PH = PV / 3 (i) PV* = 1.0,
PH = PV / 2 (j) PV* = 1.0, PH = PV / 3

(a) (b)

(c) (d)
 A unified approach for assessment of second-order effects and sway buckling 283

Figure 4 Magnification factor (second order moment to first order moment) for sway
ability (cbr = 0), (a) PV* = 0.01, PH = PV / 2 (b) PV* = 0.01, PH = PV / 3 (c) PV* = 0.025,
PH = PV / 2 (d) PV* = 0.025, PH = PV / 3 (e) PV* = 0.05, PH = PV / 2 (f) PV* = 0.05,
PH = PV / 3 (g) PV* = 0.075, PH = PV / 2 (h) PV* = 0.075, PH = PV / 3 (i) PV* = 1.0,
PH = PV / 2 (j) PV* = 1.0, PH = PV / 3 (continued)

(e) (f)

(g) (h)

(i) (j)
284 G.E. Mageirou et al.

Figure 5 Magnification factor (second order moment to first order moment) for different levels
of sway ability, (a) PV* = 0.01, PH = 1, 5PV, cbr = 0 (b) PV* = 0.05, PH = 1, 5PV, cbr = 0
(c) PV* = 0.01, PH = 1, 5PV, cbr = 2 (d) PV* = 0.05, PH = 1, 5PV, cbr = 2 (e) PV* = 0.01,
PH = PV, cbr = 5 (f) PV* = 0.05, PH = PV, cbr = 5 (g) PV* = 0.01, PH = PV, cbr = 10
(h) PV* = 0.05, PH = PV, cbr = 10

(a) (b)

(c) (d)

(e) (f)
 A unified approach for assessment of second-order effects and sway buckling 285

Figure 5 Magnification factor (second order moment to first order moment) for different levels
of sway ability, (a) PV* = 0.01, PH = 1, 5PV, cbr = 0 (b) PV* = 0.05, PH = 1, 5PV, cbr = 0
(c) PV* = 0.01, PH = 1, 5PV, cbr = 2 (d) PV* = 0.05, PH = 1, 5PV, cbr = 2 (e) PV* = 0.01,
PH = PV, cbr = 5 (f) PV* = 0.05, PH = PV, cbr = 5 (g) PV* = 0.01, PH = PV, cbr = 10
(h) PV* = 0.05, PH = PV, cbr = 10 (continued)

(g) (h)

The convergence tolerance is based on load as well as displacement norms. A parametric

non-linear analysis is conducted on the proposed model using this code. The parametric
characteristics are the stiffness coefficients cb, ct, cbr as well as the concentrated loads
PV* and PH* , which are defined as:

PV
PV* = (16)
PE*

PH
PH* = (17)
PE*
where
π * EI c
PE* = (18)
h2
Easy to use graphs for the evaluation of the amplification factor α are presented in
Figures 4 and 5, for different values of the concentrated loads PV* , PH* and PV* / PH* and
different values of sway ability.
The sway ability of the frame plays an essential role in the non-linearity of the
structure. A noteworthy observation from the proposed graphs is that very small values of
the dimensionless stiffness of the translational spring (cbr = 2) lead to a significant
reduction in non-linearity. For values of the dimensionless stiffness of the translational
spring (cbr = 10) the frame has almost linear behaviour. Moreover, the rotational ability
of the end nodes of the columns have an important impact on the non-linearity of the
frame. Small values of the distribution factors zb and zt, due to large end rotational
stiffnesses cb and ct, lead to small values of the magnification factor. On the other hand,
286 G.E. Mageirou et al.

large values of the distribution factors zb and zt (the upper right part of the graphs) lead to
large total translation and, thus, a larger magnification factor. Moreover, for constant
values of PH, the second order moment increases if the value of PV is increasing. On the
contrary, for constant values of PV, the second order moment decreases if the value of PH
is increasing. The second order moment is evaluated as M = Ph · h + Pv · Δ. Therefore, if
PH is constant, the second order moment increases if PV is increasing. In addition, if PV is
constant, the increase in PH leads to an increase in the first and second order moments.

4 Examples

In this section, examples of one-story, one-bay frames as well as of one-story, two-bay

frames are presented, for which the amplification factor α is evaluated from second order
finite element analysis as well as from the above nomograms using the proposed model
of Figure 2 and the approximate method of the amplification factor of Eurocode 3. The
small sections chosen for the beams are such that the proposed procedure is tested well
into the non-linear range.

4.1 Example 1
Consider the one-story, one-bay frame of Figure 6 with a single span L = 15 m and story
height h = 13 m, having columns with ΗΕΒ360 cross-section (Ιc = 43,190 cm4) and
beams with ΙΡΕ220 cross-section (Ιb = 2,770 cm4). The columns are considered pinned at
the base. Equal vertical concentrated loads P = 50 kN are imposed on both beam-column
joints and a horizontal concentrated load P = 50 kN is imposed on the left joint. The
frame is made of steel S 275 with Young’s modulus E = 210,000,000 kN/m2.

Figure 6 The one-story, two-bay frame of example 1

A linear as well as a geometrically non-linear analysis are conducted using the finite
element programme ADINA. The moments at the beam-column joints of the frame and
the amplification factor (aFEM) are evaluated, which are shown in Table 1. Due to the
symmetry of the frame the two columns exhibit identical behaviour. Therefore, in Table 1
the results refer to the left column.
 A unified approach for assessment of second-order effects and sway buckling 287

Table 1 Moments and amplification factor α of example 1

Moment (kNm)
aEC 3 − aFEM anom − aFEM
Node Linear Non-linear aFEM aEC3 anom % %
aFEM aFEM
analysis analysis
C 325.00 458.30 1.41 1.45 1.41 2.84 0

The distribution factors zb and zt are obtained for every column using equation (3). At
first the rotational stiffnesses cb and ct at the columns’ end nodes are evaluated using
equations which have been proposed in previous work of Gantes and Mageirou (2005)
and Mageirou and Gantes (2004).
6 EI b
ct = = 2,326.80 kNm (19)
L
and therefore:
zt = 0.92 (20)
zb = 1.0 (21)

the concentrated loads PV* and PH* , which are defined as:

PV
PV* = = 0.0095 ≅ 0.01 (22)
PE*

PH
PH* = = 0.0047 (23)
PE*
where
P
PV = P = 50 kN, PH = = 25 kN for each column, PE* = 5, 291.46 kN
2
Using the nomogram of Figure 4(a), αnom = 1.41 is evaluated. The results of the proposed
approach, which are found to be in excellent agreement with those of the finite element
analysis, are shown in Table 1.
Eurocode 3 proposes an approximate procedure of evaluating the second-order
1
moment with the amplification factor . The method is valid if the criterion
1 − Vsd / Vcr
0.10 < Vsd / Vcr < 0.25 is fulfilled. If the value of Vsd / Vcr is larger than 0.25 a non-linear
analysis must be conducted. The fraction of Vsd / Vcr is evaluated from:
Vsd ⎛ δ ⎞ ⎛ V ⎞
= ⎜ ⎟⎜ ⎟ (24)
Vcr ⎝ h ⎠ ⎝ H ⎠
where
δ the relative horizontal translation of the story
h story’s height
288 G.E. Mageirou et al.

V the total vertical reaction at the bottom of the story

H the total horizontal reaction at the bottom of the story.
The horizontal translation of the story is evaluated from the linear analysis of the frame
and for the structure of this example is δ = 2.018 m. Therefore:

Vsd ⎛ 2.018 ⎞ ⎛ 100 ⎞ Vsd

=⎜ ⎟⎜ ⎟⇒ = 0.31 (25)
Vcr ⎝ 13 ⎠ ⎝ 50 ⎠ Vcr

Thus, a non-linear analysis must be conducted. Nevertheless, the amplification factor is

evaluated.

1
= 1.45 (26)
1 − Vsd / Vcr

which is found to be in very good agreement with those of finite element analysis
(Table 1). Moreover, using the proposed model the buckling strength of frame is
estimated. The evaluation of the buckling length coefficient is conducted by means of
equation (8) and gives K = 5.63341. Thus, the Εuler buckling load is equal to:

π 2 EI c
Pcr , prop = = 166.91 kN (27)
( Kh)2

The above results are summarised in Table 2 and they are compared to the results of the
linearised buckling analysis, which are considered as ‘exact’. The proposed method is in
very good agreement with the numerical results.

Table 2 Critical loads according to different methodologies of column AB of example 1

Pcr − Pcr , FEM

Pcr(kN) %
Pcr , FEM
F.E.M. 160.73 0
Proposed 166.91 3.84

3.2 Example 2

Consider now the one-story, two-bay frame of Figure 7 with a single span L = 15 m and
story height h = 15 m, having columns with ΗΕΒ300 cross-section (Ιc = 25,170 cm4) and
beams with ΙΡΕ220 cross-section (Ιb = 2,770 cm4). The columns are considered pinned at
the base. Equal vertical concentrated loads P = 50 kN are imposed on all beam-column
joints and a horizontal concentrated load P = 50 kN is imposed on the left beam-column
joint. The frame is made of steel S 275 with Young’s modulus E = 210,000,000 kN/m2.
Following the previous procedure the distribution factors zb and zt are evaluated for
every column. The columns AD and CF exhibit identical behaviour due to the symmetry
of the frame and thus the results of the columns AD and BE are shown in Table 3.
 A unified approach for assessment of second-order effects and sway buckling 289

Figure 7 The one-story, two-bay frame of example 2

Table 3 Moments and amplification factor α of example 2

Moment (kNm)
aEC 3 − aFEM anom − aFEM
Node Linear Non-linear aFEM aEC3 anom % %
aFEM aFEM
analysis analysis
D 193.75 280.91 1.45 1.47 1.46 1.38 0.69
E 362.60 528.83 1.46 1.47 1.40 0.69 –4.11

Column AD:
zt = 0.86 (28)
zb = 1.0 (29)

6 EI b
ct = = 2,326.80 kNm (30)
L
The concentrated loads PV* and PH* are obtained as:

PV
PV* = = 0.0216 ≅ 0.02 (31)
PE*

PH
PH* = = 0.0072 (32)
PE*
where
P
PV = P = 50 kN, PH = = 16.67 kN for each column, PE* = 2,316.22 kN
3
290 G.E. Mageirou et al.

Using the nomograms of Figure 4, αnom is evaluated. A linear interpolation is needed

between the nomograms of Figures 4(b) and 4(d). The results are shown in Table 3, for
columns AD and BE, which have different values of rotational spring stiffness ct. A very
good agreement with the results of the finite element analysis is observed for column AD,
while a discrepancy of 8.20% for column BE is attributed to the fact that the horizontal
load P is in reality not equally divided between the columns of the frame.
Furthermore, the critical buckling load of column AD is evaluated according
to Eurocode 3, as above. The proposed method is in very good agreement with the
numerical results (Table 4).
Table 4 Critical loads according to different methodologies of column AD of example 1

Pcr − Pcr , FEM

Pcr(kN) %
Pcr , FEM
F.E.M. 154.15 0
Proposed 166.23 7.84

5 Summary and conclusions

An easy to use method for the assessment of the importance of second order effects in
steel moment frames and for the evaluation of the amplification factor α has been
presented. This factor is at the same time a criterion for choosing a suitable analysis
method. The amplification factor α is evaluated using nomograms for different values of
distribution factors zb and zt at the end nodes of the column and the dimensionless loads.
Moreover, using the proposed model the buckling strength of steel portal frames at all
levels of sway ability is evaluated. The advantage of the method is that the design
engineer can evaluate the buckling strength as well as the amplification factor α, using
the same proposed model, estimating the dimensionless stiffness of the translational
spring (cbr) and the distribution factors (zb, zt), On the other hand, Eurocode 3 proposes
two different criteria for the evaluation of the buckling load of the frame and the choice
of the appropriate method analysis. Examples of one-story, one-bay as well as one-story,
two-bay frames are presented, where the proposed approach is found to be in excellent
agreement with finite element results.

References
Aristizabal-Ochoa, J.D. (1994a) ‘K-factor for columns in any type of construction: nonparadoxical
approach’, Journal of Structural Engineering, Vol. 120, No. 4, pp.1272–1290.
Aristizabal-Ochoa, J.D. (1994b) ‘Slenderness K-factor for leaning columns’, Journal of Structural
Engineering, Vol. 120, No. 10, pp.2977–2991.
Aristizabal-Ochoa, J.D. (1994c) ‘Stability of columns under uniform axial load with semirigid
connections’, Journal of Structural Engineering, Vol. 120, No. 11, pp.3212–3222.
Aristizabal-Ochoa, J.D. (1997a) ‘Elastic stability of beam-columns with flexural connections
under various conservative end axial forces’, Journal of Structural Engineering, Vol. 123,
No. 9, pp.1194–1200.
 A unified approach for assessment of second-order effects and sway buckling 291

Aristizabal-Ochoa, J.D. (1997b) ‘Stability and second-order analyses of frames with semirigid
connections under distributed axial loads’, Journal of Structural Engineering, Vol. 127,
pp.1306–1315.
Aristizabal-Ochoa, J.D. (1997c) ‘Story stability of braced, partially braced, and unbraced frames:
classical approach’, Journal of Structural Engineering, Vol. 123, No. 6, pp.799–807.
Aristizabal-Ochoa, J.D. (2002) ‘Classic buckling of three-dimensional multicolumn systems under
gravity loads’, Journal of Engineering Mechanics, Vol. 128, No. 6, pp.613–624.
Aristizabal-Ochoa, J.D. (2003) ‘Elastic stability and second-order analysis of three dimensional
frames: effects of column orientation’, Journal of Engineering Mechanics, Vol. 129, No. 11,
pp.1254–1267.
Aschheim, M. and Montes, E. (2003) ‘The representation of P-Δ effects using yield point spectra’,
Engineering Structures, Vol. 25, No. 11, pp.1387–1396.
Boucard, P-A. Ladeveze, P. Poss, M. and Rougee, P. (1997) ‘A nonincremental approach for large
displacement problems’, Computers and Structures, Vol. 64, Nos. 1–4, pp.499–508.
Bridge, R.Q. and Fraser, D.J. (1987) ‘Improved G-factor method for evaluating effective lengths of
columns’, Journal of Structural Engineering, Vol. 113, No. 6, pp.1341–1356.
Chan, S-L. (2001) ‘Non-linear behavior and design of steel structures’, Journal of Constructional
Steel Research, Vol. 57, No. 12, pp.1217–1231.
Chen, Sh-J. and Wang, W-Ch. (1999) ‘Moment amplification factor for P-δ effect of steel
beam-column’, Journal of Structural Engineering, Vol. 125, No. 2, pp.219–223.
Chen, W.F. and White, D.W. (1995) ‘A selection of calbration frames in North America for
second-order inelastic analysis’, Engineering Structures, Vol. 17, No. 2, pp.104–112.
Cheong-Siat-Moy, F. (1986) ‘K-factor paradox’, Journal of Structural Engineering, Vol. 112,
No. 8, pp.1747–1760.
Cheong-Siat-Moy, F. (1991) ‘Column design in gravity-loaded frames’, Journal of Structural
Engineering, Vol. 117, No. 5, pp.1448–1461.
Cheong-Siat-Moy, F. (1997a) ‘K-factors for braced frames’, Engineering Structures, Vol. 19,
No. 9, pp.760–763.
Cheong-Siat-Moy, F. (1997b) ‘Multiple K-factors of a leaning column’, Engineering Structures,
Vol. 19, No. 1, pp.50–54.
Cheong-Siat-Moy, F. (1999) ‘An improved K-factor formula’, Journal of Structural Engineering,
Vol. 125, No. 2, pp.169–174.
Chwalla, E. (1954) Einführung in die Baustatik, Stahlbau-Verlag, Köln, Germany.
Dinkler, D. et al. (1995) Der Ingenieurbau, Baustatik, Baudynamik, Ernst & Sohn, Berlin.
Dobson, R. (2003) ‘P-Delta analysis’, Journal of New Steel Construction Engineering,
January/February, Vol. 6, No. 2, pp.30–31.
El-Sawy, K. and Moore, I.D. (1996) ‘A two-level iterative FEM technique for rigorous solution
of non-linear interaction problems under large deformations’, Journal of Computers and
Structures, Vol. 61, No. 1, pp.43–54.
Essa, H.S. (1997) ‘Stability of columns in unbraced frames’, Journal of Structural Engineering,
Vol. 123, No. 7, pp.952–957.
Eurocode 3 (1992) Design of Steel Structures Part 1.1: General Rules and Rules for Buildings,
CEN Brussels, CEN Document N419E 1994.
Gantes, C. and Mageirou, G. (2005) ‘Improved stiffness distribution factors for evaluation of
effective buckling lengths in multi-story sway frames’, Engineering Structures, Vol. 27, No. 7,
pp.1113–1124.
Ghali, A., Neville, A.M. and Brow, T.G. (2003) Structural Analysis, A Unified Classical and
Matrix Approach, Spon Press, London, Great Britain.
Hellesland, J. and Bjorhovde, R. (1996a) ‘Improved frame stability analysis with effective lengths’,
Journal of Structural Engineering, Vol. 122, No. 11, pp.1275–1283.
292 G.E. Mageirou et al.

Hellesland, J. and Bjorhovde, R. (1996b) ‘Restraint demand factors and effective lengths of braced
columns’, Journal of Structural Engineering, Vol. 122, No. 10, pp.1216–1224.
Imai, K. and Frangopol, D. (2000) ‘Response prediction of geometrically nonlinear structures’,
Journal of Structural Engineering, Vol. 126, No. 11, pp.1348–1355.
Karamanos, S.A. and Zissopoulou, E. (2002) ‘Sway-frame stability: comparison of AISC and EC3
provisions for some special cases’, 4th National Conference on Steel Structures, Patra, Greece,
pp.309–316.
Kemp, A.R. (2000) ‘Simplified amplification factors representing material and geometric
inelasticity in frame instability’, Engineering Structures, Vol. 22, No. 12, pp.1609–1619.
Kim, S., Lee, J., Choi, S. and Kim, C. (2004) ‘Practical second-order inelastic analysis for steel
frames subjected to distributed load’, Engineering Structures, Vol. 26, No. 1, pp.51–61.
Kim, S-E., Park, M-H. and Choi, S-H. (2001) ‘Direct design of three-dimensional frames using
practical advanced analysis’, Engineering Structures, Vol. 23, No. 11, pp.1491–1502.
Kishi, N., Chen, W.F. and Goto, Y. (1997) ‘Effective length factor of columns in semirigid and
unbraced frames’, Journal of Structural Engineering, Vol. 123, No. 3, pp.313–320.
Liew, J.Y.R., Punniyakotty, N.M. and Shanmugan, N.E. (1997) ‘Advanced analysis and design of
spatial structures’, Journal of Constructional Steel Research, Vol. 42, No. 1, pp.21–48.
LRFD (1999) Load and Resistance Factor Design Specification for Structural Steel Buildings,
American Institute of Steel Construction Inc., Chicago.
Mageirou, G. and Gantes, C. (2004) ‘Rotational stiffness coefficients for evaluation of critical
buckling loads in multi-story sway, non-sway and partially sway frames’, 5th International
PhD Symposium in Civil Engineering, 16–19 June 2004, Delft, The Netherlands, Proceedings,
Vol. 2, pp.1433–1440.
Mageirou, G. and Gantes, C. (2006) ‘Buckling strength of multi-story sway, nonsway and
partially-sway frames with semi-rigid connections’, Journal of Constructional Steel Research,
2005, Vol. 62, No. 9, pp.893–905.
Mageirou, G. and Gantes, C. (2007) ‘Evaluation of critical buckling load of semirigid steel
frames-an alternative approach’, Proceedings: 6th International Conference on Steel and
Aluminium Structures, 24–27 July 2007, Oxford, Great Britain, pp.907–914.
Oda, H. and Usami, T. (1997) ‘Stability design of steel plane frames by second-order elastic
analysis’, Engineering Structures, Vol. 19, No. 8, pp.617–627.
Petersen, C. (1982) Statik u. Stabilität der Baukonstruktionen, Vieweg-Verlag, Braunschweig,
Germany.
Petersen, C. (1988) Stahlbau Grundlagen der Berechnung und baulichen Ausbildung von
Stahlbauten, Vieweg, Wiesbaden, Germany.
Petrolito, J. and Legge, K.A. (1996) ‘Unified nonlinear elastic frame analysis’, Computers and
Structures, Vol. 60, No. 1, pp.21–30.
Pflüger, A. (1978) Statik der Stabtragwerke, Springer-Verlag, Berlin, Germany.
Pflüger, A. (1980) Elementare Schalenstatik, Springer-Verlag, Berlin, Germany.
Rubin, H. (1997) ‘Uniform formulae of first- and second-order theory for skeletal structures’,
Engineering Structures, Vol. 19, No. 11, pp.903–909.
Rubin, H. und Vogel, U. (1982) Baustatik ebener Stabwerke in: Stahlbau Handbuch, Vieweg,
Cologne, Germany.
Task Committee on Effective Length (1997) Effective length and Notional Load Approaches for
Assessing Frame Stability: Implications for American Steel Design, ASCE, USA.
Torkamani, M. and Sonmez, M. (2001) ‘Inelastic large deflection modeling of beamcolumns’,
Journal of Structural Engineering, Vol. 127, No. 8, pp.876–887.
Torkamani, M., Sonmez, M. and Cao, J. (1997) ‘Second-order elastic plane frame analysis using
finite element method’, Journal of Structural Engineering, Vol. 123, No. 9, pp.1225–1235.
 A unified approach for assessment of second-order effects and sway buckling 293

White, D. and Hajjar, J. (1997a) ‘Buckling models and stability design of steel frames: a unified
approach’, Journal of Structural Engineering, Vol. 42, No. 3, pp.171–207.
White, D. and Hajjar, J. (1997b) ‘Design of steel frames without consideration of effective length’,
Engineering Structures, Vol. 19, No. 10, pp.797–810.
White, D. and Hajjar, J. (2000) ‘Stability of steel frames: the case of simple elastic and rigorous
inelastic analysis/design procedures’, Engineering Structures, Vol. 22, No. 2, pp.155–167.
Wood, R.H. (1974) ‘Effective lengths of columns in multi-storey buildings’, Structural Engineer,
Vol. 52, No. 7, pp.341–346.
Zhou, Z.H. and Chan, S.L. (1997) ‘Second-order of slender steel frames under distributed axial and
member loads’, Journal of Structural Engineering, Vol. 123, No. 9, pp.1187–1193.

Notations

The following symbols are used in this paper:

A, Β, C, D = integration constants
A = section area
E = modulus of elasticity
H = the total horizontal reaction at the bottom of the story
I = moment of inertia
K = effective buckling length coefficient
L = span length of adjoining members
M = bending moment
N = axial force
P = concentrated load
Q = secant force
V = the total vertical reaction at the bottom of the story
Δ = the relative horizontal translation of the story
c = stiffness coefficient
c = dimensionless stiffness
h = column height
k = non-dimensional compressive load
x = longitudinal coordinate
z = dimensionless distribution factor at the end nodes of the column
w = transverse deflection
δ = the relative horizontal translation of the story
Subscripts
A = bottom end node of the column
B = top end node of the column
E = Euler
FEM = finite element method
H = horizontal
294 G.E. Mageirou et al.

V = vertical
b = bottom
br = bracing system
c = column
cr = critical
i = member I
sd = design
t = top
φ = rotation