STABILITY OF STRUCTURES

2. Elastic Buckling of Columns

The present chapter is devoted to the basic concept of interest in stability of beam-columns.
Analogies and differences with respect to the discrete systems utilized in the previous chapter for
pointing out some basic definitions of equilibrium and stability will be also underlined.

2.1 Euler critical Load for beam-columns

A simply supported beam column can be firstly considered for determining the equilibrium
configurations of more general continuous structures under axial load P, within the Euler hypotheses of
small displacements and (and deformations) and perfect systems.
The beam represented in Figure 2.1 is in equilibrium when resting in the straight configuration; in
this situation the axial force P only produces axial deformations ε = P EA , being EA the axial stiffness
of its cross section. Nevertheless, under the Euler hypotheses a further possible configuration is
possible in the neighbours of the mentioned “trivial” one. Let this solution described be described by
the law v ( z ) ; in the general cross section at the abscissa z, an external load M ext arises and can be
evaluated as follows:
M ext = P ⋅ v ( z ) . (2.1)
Moreover, the internal moment M int in the same cross section can be expressed in terms of the
local curvature which is proportional to the second derivative of the deflection according to the well
known equation reported below:
M int = − EI ⋅ v '' ( z ) . (2.2)

P
v(z)

Figure 2.1: Simply supported beam with axial force P.

Equality between the two bending moments expressed by equations (2.1) and (2.2) is need for
equilibrium and the following first order can be consequently obtained after few mathematical
simplifications:
v '' ( z ) + k 2v ( z ) = 0 , (2.3)
where k 2 = P EI .
The second-order linear equation reported above can be solved according to several methods;
since the discussion about the nature of solution depending upon the values of the coefficient k2 is of
main concern, the above equation can be approached through the Fourier Series, taking account of the
fact that any function defined within the range [0, L] can be expressed as series of sine and cosine
functions. In particular, being the deflection of the beam in Figure 2.1 represented by an odd function

Dr. Enzo MARTINELLI 16 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

(namely, a function v in which v ( z ) = −v ( −z ) like whatever sine wave), it can be expressed in terms of
sine series as follows:
∞
nπ z
v ( z ) = ∑ v n sin . (2.4)
n =1 L
Since a series is basically a sum of (infinite) term, its derivatives can be easily determined as a sum
of the derivatives of the various members. Consequently, the second derivative of v(z), which is
involved in equation (2.4), can be evaluated as follows:
∞ 2
⎛ nπ ⎞ nπ z
v '' ( z ) = −∑ ⎜ ⎟ ⋅ v n sin . (2.5)
n =1 ⎝ L ⎠ L
The last two expressions assumed for the deflection function v(z) and its second derivative can be
introduced in equation (2.3) obtaining the following relationships:
∞ ⎡ 2 ⎛ nπ ⎞ 2 ⎤ nπ z
∑ ⎢k − ⎜
n =1 ⎢
⎟ ⎥ ⋅ v n sin
⎝ L ⎠ ⎥⎦ L
=0 . (2.6)
⎣
Two possible kinds of solutions can be obtained by the above relationship:
- if vn=0 for every value of n, the trivial solution is obtained, since all the coefficients of the
series in equation (2.4) are null;
⎡ nπ ⎞ ⎤
2
- if ⎢k 2 − ⎜⎛ ⎟ ⎥ = 0 , then the corresponding coefficient vn could be non-zero and the
⎢⎣ ⎝ L ⎠ ⎥⎦
deflection relationship described in equation (2.4) is different by the trivial one.
The latter condition occurs as
2
⎛ nπ ⎞
k2 = ⎜ ⎟ , (2.7)
⎝ L ⎠
or, equivalently, as the axial load P achieves the following value:
n 2π 2 EI
Pcr ,n = 2
. (2.8)
L

P/PE

v
Figure 2.2: Critical loads and corresponding adjacent configurations of the beam.
Load values Pcr,n determined by equation (2.8) have the meaning of critical load in the Euler sense;
as load P achieves that value a bifurcation in equilibrium path can occur. An infinite number of critical

Dr. Enzo MARTINELLI 17 Draft Version 13/01/2008

STABILITY OF STRUCTURES

loads Pcr,n can be theoretically determined according to the mathematical approach carried out above.
Nevertheless, the smaller critical value, namely that obtained for n=1, for practical application is
currently named the Euler critical load:
π 2 EI
PE = Pcr ,1 = 2
. (2.9)
L
The mechanical meaning of the above critical load can be easily understood by keeping in mind
the theory developed in the previous chapter with reference to simple discrete systems. Since a finite
number of kinematical parameters are needed for describing the behaviour of discrete systems, a finite
number of critical loads can be obtained corresponding to different adjacent configurations in the
neighbours of the trivial one. In particular, only one critical load value corresponding to the only
possible adjacent configuration of the system has been found since the two discrete structures are
basically single-degree-of-freedom systems. On the contrary, an infinite number of adjacent
configurations can be found for general continuous systems, even those made out of a single elastic
beams. Consequently, as a matter of principle, an infinite number of critical loads described by equation
(2.8) can be found for the simply-supported beam considered above. Nevertheless, for the practical
applications only the first one is usually of interest since bifurcation firstly occur under that load.
Finally, equations (2.8) and (2.9) directly apply to simply-supported beams; in the next sections, the
extension to whatever restraint condition will be exposed.

2.2 Magnification factor

Since structural members are often loaded even in transverse direction by a distributed load
q ( z ) of general shape, its effect on stability needed to be investigated. Let v 0 ( z ) be the deflection due
to only the transverse load q (z ) without considering the axial load P, and v ( z ) the total deflection
taking account of both. First of all, the relationship between the bending moment M 0 ( z ) due to q (z )
and the corresponding deflection law v 0 ( z ) is reported below:
M 0 ( z ) = − EIv 0 '' ( z ) . (2.10)
The expression of the external bending moment introduced by equation (2.1) has to be slightly
modified for looking after the effect of bending moments induced by the transverse load:
M ext = P ⋅ v ( z ) + M 0 ( z ) . (2.11)
Consequently, the equilibrium condition between external and internal moments (the latter one
always defined by equation (2.2)) results in the following second-order linear differential equation:
v '' ( z ) + k 2v ( z ) = v 0 '' ( z ) . (2.12)
A solution procedure based on series development of both the initial (first-order, in the
following) and complete deflections v 0 ( z ) and v ( z ) can be utilized for solving the problem without
précising the particular shape of the distributed load q (z ) . In fact, for whatever distribution of
transverse load an equivalent Fourier series can be introduced; under the hypotheses of load
distribution symmetrical with respect to the mid-span point only sine-terms appears in that series:
∞
nπ z
v 0 ( z ) = ∑ v 0,n sin . (2.13)
n =1 L
and the general expression of coefficients can be expressed as follows:
L
2 nπ z
v 0,n = ∫ v 0 ( z ) sin ⋅ dz . (2.14)
L 0
L
Since the total deflection v ( z ) can be always developed through the Fourier series reported in
equation (2.4), the differential equation (2.12) can be easily simplified as follows:
∞ ⎡⎛ 2 ⎛ nπ ⎞ 2 ⎞ ⎛ nπ ⎞
2 ⎤ nπ z
∑ ⎢⎜ k − ⎜
n =1 ⎢
⎜ ⎝ L
⎟
⎠
⎟
⎟
⋅ v n + ⎜
⎝ L
⎟
⎠
⋅ v 0,n ⎥ ⋅ sin
⎥ L
=0 , (2.15)
⎣ ⎝ ⎠ ⎦

Dr. Enzo MARTINELLI 18 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

which is verified only if every value of the coefficient between square brackets is zero or, equivalently,
under the following condition :
2 2
⎛ nπ ⎞ ⎛ nπ ⎞
⎜ ⎟ ⎜ ⎟ 1
vn = ⎝ L ⎠ ⋅ v = ⎝ L ⎠ ⋅ v 0,n = ⋅ v 0,n .
0, n (2.16)
⎛ nπ ⎞
2
⎛ nπ ⎞
2 P
1 −
⎜ ⎟ −k ⎟ −k
2 2
⎜ Pcr ,n
⎝ L ⎠ ⎝ L ⎠
Making reference to the first term the following relationship can be pointed out:
v1 1
= .
v 0,1 1 − P (2.17)
PE
meaning that the ratio between the coefficient of the first term of the series (2.4) and the corresponding
one of the series in equation (2.13) are related to the critical load PE through the expression on the right
member of equation (2.17) which, consequently, assumes the meaning of magnification factor in the
sense that total deflection (v(z) in equation (2.4)) can be obtained by amplifying the first order
deflections (v0(z) in equation (2.13)) according to that magnification factor.
Magnification factor

18.0
16.0
14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
P/PE

Figure 2.3: Magnification factor as a function of the critical load.

The same relationship can be easily recognised in the hypothesis that the two sine series in
equation (2.4) and (2.13) are basically controlled by the corresponding first terms, resulting in the
following approximate relationship:
v (z ) v1
≈ . (2.18)
v 0 (z ) v 0,1
Consequently, the total-to-first-order ratio in terms of strain measures (namely, curvatures) can be
even expressed in terms of the magnification factor defined in equation (2.16). Indeed, due to linear
behaviour in terms of stress strain relationship, total bending moments M(z) can be even expressed in
terms of the initial ones M0(z):
M (z ) v 1 '' ( z ) v1
= ≈ . (2.19)
M 0 (z ) v 0,1 '' ( z ) v 0,1
Consequently, the following general expression can be assumed for evaluating the effect of axial
forces on bending moments and, generally, on stresses in structures:
1
M (z ) = ⋅ M 0 (z ) .
P (2.20)
1−
PE
The above equation and the definition of the magnification factor is often utilized for considering
whether axial force affects or not the total stress and strain state of structures.

Dr. Enzo MARTINELLI 19 Draft Version 13/01/2008

STABILITY OF STRUCTURES

2.3 Theoretical classification of beam-columns

Equation (2.9) points out the value of axial load under which a possible bifurcation in equilibrium
path; in other words, under the load PE=Pcr,1 a perfect (namely, perfectly straight and unaffected by
accidental eccentricities) beam-columns, made out of an elastic material, two alternative equilibrium
configurations are possible:
- The first one is the original straight (“trivial”) configuration which is admitted under
whichever value of axial force P;
- The second one is a buckled configuration which can be approximated by a sine half-wave.
A critical stress can be easily evaluated through equation (2.9) dividing both members by the cross
section area A:
π 2E
σ cr = 2 . (2.21)
λ
Since real beam-columns, far from not being crooked, are surely made out of a material behaving
elastically up to a limited value of stress fy (i.e., the yielding stress in steel) a critical value of slenderness
for which the critical stress defined by equation (2.21) is equal to the yielding stress fy can be derived:
E
λp = π . (2.22)
fy

Figure 2.4: Axial strength of beam-columns as a function of slenderness.

Consequently, the overall field of slenderness can be subdivided into two main ranges:
- Stocky columns, for λ < λ p in which yielding stress fy is always greater than the critical stress
resulting in buckling;
- Slender columns, for λ ≥ λ p which buckle within the elastic range since the critical stress is
always smaller than the yielding stress.

2.4 The role of imperfection

Since geometrical imperfections, either in terms of lack in straightness in basic configuration or
accidental eccentricity of loads, can be assumed as responsible for initial curvature of the beam-column,
the equation (2.12) can be even utilized for describing the behaviour of members affected by an initial
curvature (crookedness) due to imperfections rather that external transverse loads. The behaviour of
imperfect systems with respect to the perfect one investigated in the light of Euler hypotheses has been
already treated within the first chapter with reference to simple discrete systems: simple bifurcation
have been pointed out for perfect ones, while asymptotical behaviour toward that bifurcation has been
demonstrated for imperfect systems which experienced lateral deformations as large as the importance
of imperfection.

Dr. Enzo MARTINELLI 20 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Now, introducing the relative displacement w (z ) = v ( z ) − v 0 ( z ) the behaviour represented in Figure 2.5
can be intuitively understood, where the equilibrium path of columns is as far from the trivial
configuration assumed by the perfect column for P < PE as imperfection is relevant.

Figure 2.5: Behaviour of columns affected by initial curvature.

Since imperfection play a relevant role in affecting the equilibrium behaviour of beam-columns,
an easy and reliable way for estimating its amount is needed. The definition of magnification factor
reported in equation (2.17) can be proficiently utilised for evaluating the amplitude of imperfections, at
least under the hypothesis of initial imperfection well approximated by the first term of the sine series
in equation (2.14). In this case, equation (2.17) strictly applies and the following relationship in terms of
relative displacement can be derived:
w w v0
= + , (2.23)
P PE PE
relating the initial displacement v0 (namely the maximum deflection throughout the column axis due to
imperfection) and the relative displacement w, which can be monitored during an experimental test
through some displacement transducers placed along the column axis. Since the load P, can be applied
and measured through load cells, the couples (w, w/P) can be plotted on a in the bidimensional graph:
a straight line can be identified by means of a numerical regression and its slope can be related to the
critical load as reported in Figure 2.6.

Figure 2.6: The “Southwell plot” for evaluating critical loads in imperfect columns.

2.5 Differential equation of beam-columns

The effect of axial force on elastic beam-column can be generally considered within the classical
Bernoulli Theory for beams in bending which is basically founded upon the following two assumptions:
- Plane section remains plane after deformation;
- Deformed sections keep perpendicular to the (deformed) beams axis.
Under these hypotheses, the differential equations governing equilibrium can be easily derived
obtaining the following two relationships:

Dr. Enzo MARTINELLI 21 Draft Version 13/01/2008

STABILITY OF STRUCTURES

dV
dV + qdz = 0 ⇒ +q =0 , (2.24)
dz
dz dM
dM − Vdz + qdz ⋅ = 0 ⇒ =V , (2.25)
2 dz
where M and V are the local bending and shear forces (Figure 2.7).

Figure 2.7: Equilibrium of an infinitesimal element without axial load effect.

The effect of axial force can be taken into account if the axial force is thought on the deformed
configuration of the beam segment (Figure 2.8). Under this assumption equation (2.25) can be
completed by considering the increase in moment due to axial load eccentricity dv evaluated with
respect to the deformed shape of the beam-column:
dz dM dv
dM − Vdz + qdz ⋅ − P ⋅ dv = 0 ⇒ − P =V , (2.26)
2 dz dz
where P is the axial (compressive) force.

Figure 2.8: Equilibrium of an infinitesimal element with axial load effect.

Since the usual compatibility equations can be written between deflection v(z) and curvature, the
following (generalized) stress-strain relation applies at least under the hypothesis of materials within the
linear range:
M ( z ) = − EIv ''( z ) . (2.27)
Finally, deriving once equation (2.26) and introducing equation (2.24) and, finally, deriving once
more before of introducing equation (2.27) the following fourth-order differential equation with linear
coefficients can be obtained:
P q
v iv + v '' = . (2.28)
EI EI
where two further hypotheses have been assumed:
- constant axial load P throughout the column axis;
- uniform value of the flexural stiffness EI.
In the cases of strictly positive (namely, compressive) axial load P the following definition can be
introduced:
P
k2 = , (2.29)
EI
and the general solution of equation (2.28) can be placed in the following form:
v ( z ) = A sin kz + B cos kz + Cz + D + v p ( z ) , (2.30)
where v p ( z ) is a particular solution depending on the transverse load applied upon the beam.

Dr. Enzo MARTINELLI 22 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

The following equation can be utilized for determining the behaviour of beam-columns with axial
load up to possibly deriving the critical value defined in the first paragraph of the present chapter. For
instance, the behaviour of the simply supported beam-columns represented in Figure 2.1 can be studied
providing equation (2.30) (in which vp(z)=0 since no transverse load is applied) with the relevant
boundary conditions imposed by end restraints:
- for z=0:
v (0) = 0 (2.31)
M ( 0 ) = 0 ⇒ v '' ( 0 ) = 0
- for z=L:
v (L ) = 0 (2.32)
M ( L ) = 0 ⇒ v '' ( L ) = 0
The above boundary conditions can be written in the following matrix form by pointing out
some significant algebraic properties:
⎡ 0 1 0 1⎤ ⎡ A ⎤ ⎡0 ⎤
⎢ ⎥
⎢ 0 −k 2 0 0 ⎥ ⎢ B ⎥ ⎢0 ⎥
⋅⎢ ⎥ = ⎢ ⎥ . (2.33)
⎢ sin kL cos kL L 1⎥ ⎢ C ⎥ ⎢0 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢−k sin kL
2
−k 2 cos kL 0 0 ⎥⎦ ⎣ D ⎦ ⎣ 0 ⎦
The above set of simultaneous equations has the trivial solution (A=B=C=D=0) as unique
possible solution if the matrix is non singular. On the contrary, other possible solutions are admissible
if the matrix determinant vanishes:
−k 4 L sin kL = 0 , (2.34)
which is true if
kL = nπ with n ≥ 1 (2.35)
and, equivalently, if
P n 2π 2
= (2.36)
EI L2
Equation (2.36) confirms that an infinite number of alternative equilibrium positions, stemming
out by bifurcation at given values of the axial load P=Pcr,n, are possible. If the first one is of interest,
Euler load PE defined by equation (2.9) can be find out by equation (2.36) and a different ad more
general approach to evaluating critical loads in elastic beam-columns can be put in place by means of
the differential equation of beam-columns under axial loads whose general solution is reported in
equation (2.30).

2.6 Critical loads of perfect columns with various end-restraint

Evaluating critical loads of elastic beam-columns considering different end-restraint conditions
can be pursuit by introducing suitable boundary conditions instead of those written in eqs. (2.31) and
(2.32) for simply-supported beams. For instance, the structural schemes reproduced in Figure 2.9 could
be easily solved as described in the previous section. The first one, for example, which is fixed at one
end and clamped at the other one could be solved by introducing the following boundary conditions:
- for z=0:
v (0) = 0 (2.37)
v ' (0) = 0
- for z=L:
v (L ) = 0 (2.38)
M ( L ) = 0 ⇒ v '' ( L ) = 0
resulting in a set of simultaneous equations whose matrix of coefficient is singular if its determinant
vanishes as described by the following equation:
tan kL = kL . (2.39)

Dr. Enzo MARTINELLI 23 Draft Version 13/01/2008

STABILITY OF STRUCTURES

The above equation can only be solved through numerical or graphical methods providing the
following solution, among other grater ones:
kL ≈ 4.4934 ≈ 1.4303π , (2.40)
and the following expression can be found for the critical load value:
π 2 EI
PE = . (2.41)
( 0.699 ⋅ L )
2

Figure 2.9: Effective length L0 of columns with various end restraints.

It is possible to demonstrate that the reduced span length at the denominator of the previous
formula is the distance between the two zero-bending points, namely the one in which curvature is zero
in the deformed configuration represented in Figure 2.9 and the clamped end of the beam. Based on
this observation a general way for deriving the critical value PE = Pcr ,1 of beam-columns under
whichever restraint condition can be obtained by extending equation (2.9) by replacing the physical
length L with the effective length L0 of the beam-column which bends as an half sine-wave in the
adjacent equilibrium condition. In particular, a coefficient β can be introduced as a ratio between L and
L 0:
L0 = β L , (2.42)
and the following general expression of the Euler load can be introduced:
π 2 EI
PE = 2
. (2.43)
L0
Finally, in a more general sense, critical loads of beam-columns with flexible end restraints can be
evaluated depending upon flexibility of such restraints. Although analytical solution of the problem is
always possible, a more general and handy way for determining the values of L0 can be undertaken by
using the so-called alignment charts or nomograms. Figure 2.10 reports to different alignment-charts
related to non-say and sway systems, namely those in which relative transverse displacement of the two
ends are restrained or allowed (non-sway or sway structures).

Dr. Enzo MARTINELLI 24 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

a) non-sway members b) sway members

Figure 2.10: Alignment charts for beam-columns with flexible end restraints.
In both cases the input data are the non-dimensional beam end-flexibilities kA and kB and the
corresponding value of the coefficient β can be easily derived as the intersection between the aligning
straight line and the central axis. Considering non-sway schemes, coefficient β values ranges between
0.50 and 1.0 as the end restraints range from completely stiff ( k A = kB = 0 ) to completely deformable
( k A → ∞, kB → ∞ ). On the contrary, for the same flexibility values, coefficient β in sway columns
ranges from 1.0 to infinity.
The effective length or the coefficient β can be alternatively determined according to the
following analytical relationships:
- non-sway members:
⎛ kA ⎞ ⎛ kB ⎞
β = 0.5 ⋅ ⎜ 1 + ⎟ ⋅ ⎜1 + ⎟ , (2.44)
⎝ 0.45 + k A ⎠ ⎝ 0.45 + kB ⎠
- sway members:
⎧⎪ k A kB ⎛ k ⎞ ⎛ kB ⎞ ⎫⎪
β = max ⎨ 1 + 10 ;⎜1 + A ⎟ ⋅ ⎜1+ ⎟⎬ . (2.45)
⎩⎪ k A + kB ⎝ 1 + kA ⎠ ⎝ 1 + kB ⎠ ⎭⎪
Other possible approximate expressions for the β coefficient can be found within the scientific
and technical literature, in terms of the nodal flexibilities (or stiffnesses). Column end flexibility can be
easily determined in the case of single members (Figure 2.11); the conceptual definition of nodal
flexibility εi can be expressed as the ratio between the observed nodal rotation θi and the applied
moment Mi:
θ
εi = i . (2.46)
Mi
In other words the nodal flexibility is the nodal rotation obtained for a unit nodal moment. The
relative flexibility coefficient ki to be used for deriving the b coefficient through the alignment charts
in Figure 2.10 or the approximate relationships (2.44) and (2.45) is defined by dividing such nodal
flexibility by a term related to the column flexural deformability:
εi
ki = . (2.47)
Lcol / EI col

Dr. Enzo MARTINELLI 25 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Figure 2.11: Definition of nodal flexibility for single members.

Furthermore, both the alignment charts and the mentioned approximate relationships can be
used for the more general task of defining the effective length and the β coefficient for a beam-column
in a general frame. According to the symbols briefly introduced in Figure 2.15 the relative flexibility
terms kA and kB can be derived as follows:
ε A + L col ,1 / EI col ,1
kA = . (2.48)
L col ,1 / EI col ,1 + Lbeam ,11 / EI beam ,11 + Lbeam ,12 / EI beam ,12
ε B + L col ,2 / EI col ,2
kB = . (2.49)
L col ,2 / EI col ,2 + Lbeam ,21 / EI beam ,21 + Lbeam ,22 / EI beam ,22

L2/EIcol,2

L2/EIbeam,21 εB L2/EIbeam,22

εA
L1/EIbeam,11 L1/EIbeam,12

L1/EIcol,1

Figure 2.12: Definition of nodal flexibility for a member in frame.

2.7 Role of shear flexibility on stability of beam-columns

Shear flexibility is usually neglected especially when slender members are considered.
Nevertheless, at least under a theoretical point of view, shear flexibility affects stability of members and
structures as can be easily demonstrated in the following. Coming back to the assumptions of the first
chapter, equation (2.1) generally applies, while equation (2.2) has to be updated to take into account the
deformation contribution due to shear flexibility. Since it basically consists of a shear strain γ which is
as great a shear force V, the following relationship can be introduced:
χ
γ = V ⋅V , (2.50)
GA

Dr. Enzo MARTINELLI 26 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

where χV is the shear factor of the transverse section, G is the shear elastic modulus and A the cross
section area.
Consequently, shear strain results substantially results in an increase in beam curvature χ which
can be evaluated as the second derivative of γ:
dϕ d 2v d γ M
χ= =− 2 + = , (2.51)
dz dz dz EI
and the following definition of the bending moment as a function of flexural and shear strains can be
obtained to replace equation (2.2):
⎡ d 2v V'⎤
M int = − EI ⎢ 2 − χV ⋅ ⎥ . (2.52)
⎣ dz GA ⎦
Since shear force V is the first derivative of the bending moment M, the following relationship
can be stated:
d
V ' = M '' = ( Pv ) = Pv '' . (2.53)
dz
which can be substituted in equation (2.52) with aim of expressing the (internal) bending moment as a
function of the second derivative of deflection function v(z):
⎡ P ⎤ d 2v
M int = − EI ⎢1 − χV ⋅ . (2.54)
⎣ GA ⎥⎦ dz 2
If the second members of equation (2.1) and (2.54) are equal, as required by equilibrium, the
following second order differential equation can be derived instead of equation (2.3):
d 2v P
+ ⋅v = 0 .
dz 2
⎡ P ⎤ (2.55)
EI ⎢1 − χV ⋅ ⎥
⎣ GA ⎦
A new expression of the critical load Pcr can be derived under the same hypotheses mentioned in
the first paragraph where only flexural strains have been considered:
2
Pcr ,V ⎛ π ⎞
=⎜ ⎟ ,
⎡ Pcr ,V ⎤ ⎝ L 0 ⎠ (2.56)
EI ⎢1 − χV ⋅
⎣ GA ⎥⎦
and
π 2 EI 1 1
Pcr ,V = ⋅ = Pcr ⋅ .
L0 2
π 2
EI π 2
EI (2.57)
1+ 2
⋅ χV ⋅ 1+ 2
⋅ χV ⋅
L0 GA L0 GA
An alternative relationship for the critical load Pcr,V can be placed in the following form:
1 1 χV
= + . (2.58)
Pcr ,V Pcr GA
Finally, it is worth noticing that shear flexibility results in a reduction of the theoretical value of
the critical load determined in the first paragraph for the Bernoulli beam. Such reduction is usually
negligible for the most common simple profile, while can be relevant in built-up columns (namely,
columns made out of angle or channel profiles assembled one another through braces or other
members) as will be exposed in the following.

Dr. Enzo MARTINELLI 27 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Figure 2.13: Reduction in critical load due to shear flexibility.

2.8 Code specifications for beam columns

Code provisions on stability of beam-columns can be regarded under the light of the general
theoretical bases outlined within the previous sections whose contents are a fundamental background
for whatever design-oriented discussion on stability of members and structures.
In the present paragraph the basic provisions of two codes of standards of interest for designers
mainly working in Italy and Europe will be described and commented with reference to the key aspects
related to stability checks of steel members. Indeed, the basic provisions of both the Italian Code (D.M.
96 [12]) and the European one (Eurocode 3, [13] and [14]) will be analyzed.
Global stability will be basically discussed even if some insights above local buckling possibly
affecting steel members will be even addressed with particular reference to the main rules provided by
Eurocode 3.

2.8.1 Stability check of beam-columns according to the Italian Code

The Code of Standard usually adopted in Italy is the D.M. 96 which allows the designers to adopt
either the Permissible Stress Method or the Limit States Methods. The latter choice will be only
considered in the following, although the use of the former one is still very popular among the Italian
designers. The usual equation of safety checks according to Limit States Method can be consequently
introduced as the design value of stresses Sd should not be greater than the corresponding design value
of Resistance Rd:
Sd ≤ R d . (2.59)
Stability check is one of the possible Ultimate Limit Verifications to be carried out on structures.
Design values of stresses have to be derived by analyzing the structure under the design forces obtained
by the well-known combination of permanent (self weights Gk and permanent loads Gk’) and live
actions Qk:
⎡ n
⎤
Fd = γ g ⋅ ( Gk + Gk ' ) + γ q ⋅ ⎢ Qk1 +
⎣
∑ψ
i =2
0i Qki ⎥ .
⎦
(2.60)

Resistance Rd can be generally evaluated with reference to the design values of material strengths;
in particular, yielding stress fyk is the mechanical properties of key importance for steel structures and
the design value can be obtained as a function of the characteristic one through the definition of the
partial safety factors γm:
f yk
f ad = . (2.61)
γm

Dr. Enzo MARTINELLI 28 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Two possible ways can be followed for evaluating the design strength Rd of members as a
function of material properties fad:
- plastic analysis of cross section could be carried out for deriving the ultimate (plastic) bending
moment of members as follows:
M Rd = W pl f ad ; (2.62)

- conventional elastic analysis of cross section could be even carried out for evaluating the
ultimate (plastic) bending moment of members:
M Rd = Wel f ad ; (2.63)
Since the plastic modulus Wpl is always greater than elastic one Wel, the two definitions of
(flexural) strength stemming by equations (2.62) and (2.63) will not be self-consistent. Consequently, the
Italian code states that different values of the partial safety factor have to be considered when plastic or
elastic analysis of section is carried out. In particular, γ m = 1.12 must be considered for plastic
hypothesis (equation (2.62)), while a unit value of γ m can be taken in equation (2.63). In this way, being
1.12 the approximate value of plastic-to-elastic modulus ratio for I-shaped profiles the two equations
provide the same value at least for those sections which are the most widely used steel ones.
Stability check is one of the key steps for steel members according to the present code and it is
carried out according to the well-known Omega-Method, theoretically described in section 2.3.
Although the value of ω is basically related to the column slenderness, it is also affected by the amount
of imperfections deriving by the production and mounting process. Four different relationships (curves,
in the following) are defined between the ω factor and the corresponding slenderness λ, depending on
the kind of profile of interest (and, consequently, the expected amount of imperfections):
- curve a, for single members of hollow square, rectangular or circular shape, welded or hot-
rolled, whose maximum thickness is not greater than 40 mm;
- curve b: for simple members made out of I-shaped wide flange profiles whose height-to-
width ratio is not smaller than 1.2 and thickness not greater than 40 mm (basically all IPE
profiles, HEA starting from HEA400, HEB starting from HEB360 and HEM from
HEM340). Furthermore, I-shaped profiles strengthened by welded planes and closed-section
welded profiles, both thinner than 40 mm, follows this curve ;
- curve c: for simple or built-up members was thickness is not greater than 40 mm;
- curve d: all the sections thicker than 40 (i.e. the so-called jumbo profiles).
Finally, the ω factor can be derived for each kind of profile depending on slenderness and
according to the above classification; curves are different for the three possible steel grades (S235,
S275, S355). The values in terms of non-dimensional slenderness and for the four curves mentioned
above are reported in Table 2.1.

2.8.1.1 Stability check under axial load

Stability check under axial load N can be carried out by verifying that the given member complies
with the following relationship given in terms of stresses:
ωN
σa = ≤ f ad , (2.64)
A
where N is the axial force and A is the cross section area. Since two values of the second moment of
inertia Ix and Iy (and, correspondingly, two values of the radius of gyration ρx and ρy) can be defined
around the two principal axes of the cross section, the maximum value of slenderness has to be
considered:
⎛ L 0, x L 0, y ⎞
λ = max ⎜ ; ⎟ , (2.65)
⎜ ρx ρ y ⎟
⎝ ⎠
with the purpose of determining the relevant value of the ω factor:
ω = ω ( λ ; Stability curve , Steel grade ) . (2.66)

Dr. Enzo MARTINELLI 29 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Table 2.1: Values of the w factor as a function of the non-dimensional slenderness λ λ p .

2.8.1.2 Stability check under eccentric axial load

Eccentricity hugely affects both strength and stability checks in steel members and the validity of
the following relationship has to be verified for the member not to fail in buckling:
ωN M eq , x M eq , y
σa = + + ≤ f ad ,
A ⎛ N ⎞ ⎛ N ⎞ (2.67)
ψ x Wel , x ⎜⎜ 1 − ⎟⎟ ψ yWel , y ⎜ 1 − ⎟
N cr , x ⎠ ⎜ N cr , y ⎟⎠
⎝ ⎝
where bending moments Mx and My around the two principal axes of inertia of the section are involved
through an equivalent value whose meaning will be better explained in the following; magnification
factors also appear for amplifying the stress contributions of bending moments and the two terms
denoted as ψ x and ψ y are named “plastic adaptation coefficient” and can be conservatively placed to
one for the sake of simplicity.

Dr. Enzo MARTINELLI 30 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

The two values of the critical loads in x and y directions can be easily evaluated as a function of
the geometrical properties of the column and its cross section:
π 2 EI x π 2 EI y
N cr , x = 2
, N cr , y = 2
. (2.68)
L 0, x L 0, y
Finally, the meaning of the equivalent bending moments Meq needs to be clarified as a function of
the shape of the corresponding diagram. First of all, equivalent bending moment Meq=M0 as an uniform
bending moment diagram is considered. Secondly the following relationship can be considered among
the equivalent value and the two nodal ones Ma and Mb, being M a ≥ M b :
M eq = 0.6 M a − 0.4 M b ≥ 0.4 M a . (2.69)
Finally, in a more general case of non linear (i.e. parabolic) shape of the bending moment diagram
the equivalent bending moment can be determined as follows:
M eq = 1.3 M m . (2.70)
M m being the average value of bending moment throughout the column axis and always considering
the following limitation:
0.75 M max ≤ M eq ≤ M max . (2.71)

2.8.2 Stability check of beam-columns according to Eurocode 3

A more general approach to stability of beam-columns is pursuit in Eurocode 3 in which a clear
difference is firstly stated among steel members depending on the geometric and mechanical properties
of their cross section. Consequently, steel profiles are divided into four different classes with respect to
their ultimate behaviour in terms of both (flexural) strength and ductility. The first of the following
subsections will explain with enough details this issue, while the other two will deal with the stability
check of members under axial load with or without eccentricity.
Finally, it is worth noticing that safety checks in EC3 are formulated within the general
framework of the Limit States Methods. Consequently, stability check deals with the Ultimate Limit
State of the member and the action can be combined according to a rule like the one reported by
equation (2.60). Similar formula can be even assumed for defining the design strength of material, but a
more complicated way of defining the value of partial safety factor is considered in EC3, in which a
different value has to be taken into account depending on the kind of failure mechanism of concern.
Precisely, a value γ M 1 = 1.05 can be assumed for Stability while all the other possible values of the safety
factors are reported in Table 2.2 for the sake of completeness.
Table 2.2: Values of the safety factors according to EC3.
SAFETY
LIMIT STATE SUBJECT
FACTORS
Elastic limit state of the section γM=1.00
Plastic collapse of the structure γM=1.12
Transverse section in Class 1, 2, 3 γM0=1.05
MATERIALE
Transverse section in Class 4 γM1=1.05
Stability of members γM1=1.05
Strength of net section γM2=1.20
Bolts γMb=1.35
Rivets γMr=1.35
Pivots γMp=1.35
JOINTS
Angle welds γMw =1.35
Weldings in first class γMw =1.05
Weldings in secondo class γMw =1.20
Ultimate Limit State γMs,ult=1.25
FRICTION JOINTS Serviceability Limit State γMs,ser=1.25
Ultimate Limit State (lerger holes) γMs,ult=1.50
FATIGUE Fatice strength γMf=1.00
Non welded γC1=1.00
FRAGILITY
Welded γC2=1.50

Dr. Enzo MARTINELLI 31 Draft Version 13/01/2008

STABILITY OF STRUCTURES

2.8.2.1 Classification of steel cross sections

Behaviour of structural steel is ductile in behaviour since huge values of ultimate strain can be
developed, namely in tension, after yielding. Nevertheless, slenderness of both section and member as a
whole can undermine the natural ductility of steel as compression arises. Since, the above discussion
mainly focused upon the global buckling issue, the present section will address the aspects related to
local buckling of steel members which hugely affects their flexural behaviour. In particular, four
different classes of profiles can be defined looking after their behaviour in terms of sectional strength
and ductility, qualitatively depicted in Figure 2.14 in terms of moment-curvature relationships.
M

Mpl
Class 2
Class 1
My
Class 3
Class 4

Figure 2.14: Possible moment-curvature behaviour for steel profiles.

In particular, the four classes corresponding to the various responses represented in the
mentioned figure will be defined as follows:
- Class 1 cross-sections are those can form a plastic hinge with the rotation capacity required
for plastic analysis;
- Class 2 cross-sections are those which can develop their plastic moment resistance, but have
limited rotation capacity;
- Class 3 cross-section are those in which the calculated stress in the extreme compression
fibre of the steel member can reach its yield stress, 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-sections.
Beam-column sections can be classified into one of the above classes depending upon the three
aspects listed below:
- effective length-to-thickness d/t of web and free length-to-thickness b/t of flanges;
- state of stress (pure bending, pure compression or bending and compression);
- grade of steel, being stronger steels more sensitive to local buckling phenomenon than
weaker ones.
Table 2.3 summarizes the above concepts with reference to the most common hot-rolled I-
shaped sections; similar tables can be found in EC3 for other joist sections.

Dr. Enzo MARTINELLI 32 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Table 2.3: Classification of sections for local buckling according to EC3: webs.

Dr. Enzo MARTINELLI 33 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Table 2.4: Classification of sections for local buckling according to EC3: flanges.

The web and the flanges of the section can be classified according to their dimensions, the steel
grade and the state of stress according to the rules briefly reported in Table 2.3 and Table 2.4. The
section as a whole has to be classified in class of the most slender of its members. If such a section falls
in Class 4, local stability occurs before of yielding moment and, consequently, the flexural strength of
the member cannot be completely developed. For this reason an effective section have to be
determined by reducing the compressed area of web and flange in order to obtain a reduced virtual
section to be considered in both strength and stability check. The way in which such section can be
determined are not completely explained within this notes for the sake of brevity; nevertheless, the
reader could directly refer to Eurocode 3 (section 6.2.2.5) for this topic.

2.8.2.2 Stability check under axial load

Stability check under axial load can be carried out through the following inequality:
f ay
N Sd ≤ N Rd = χ min β A A , (2.72)
γ M1
being χ min = min ( χ ( λx ); χ ( λ y )) a reduction factor related to the relative slenderness λ defined as
follows:
β A Af y
λ= . (2.73)
N cr
The factor β A is defined as a ratio between the effective area and the gross section area; for
section belonging to the first three classes defined above β A = 1 , while values smaller than the unity
characterize profiles in class 4. Further details about the definition of the effective area for Class 4
profiles will be given in the following. If β A = 1 , the following relationship can be stated between the
relative slenderness λ , the absolute one λ and the critical one λ p :

Dr. Enzo MARTINELLI 34 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Af y fy L02 λ
λ= ⋅ L02 = ⋅ = . (2.74)
π EI
2
π E ρ
2 2
λp
Under a conceptual standpoint the parameter χ is basically the inverse of the ω factor reducing
plastic strength of the section for looking after the global slenderness λ on the column. The
relationship ω ( λ ) depends once more by the kind of imperfections and, consequently, by the type of
profile. Four curves denoted as a, b, c and d can be utilized for that relationship:
- Curve a represents quasi perfect shapes: hot-rolled I-sections (h/b>1,2) with thin flanges
(tf≤40mm) if buckling is perpendicular to the major axis; it also represents hot-rolled hollow
sections;
- Curve b represents shapes with medium imperfections: it defines the behaviour of most
welded box-sections; of hot-rolled I-sections buckling about the minor axis; of welded I-
sections with thin flanges (tf > 40mm) and of the rolled I-sections with medium flanges
(40<tf≤100mm) if buckling is about the major axis; it also concerns cold-formed hollow
sections where the average strength of the member after forming is used;
- Curve c represents shapes with a lot of imperfections: U, L, and T shaped sections are in this
category as are thick welded box-sections; cold-formed hollow sections designed to the yield
strength of the original sheet; hot-rolled H-sections (h/b ≤ 1,2 and tf ≤ 100mm) buckling
about the minor axis; and some welded I-sections (tf ≤ 40mm buckling about the minor axis
and tf > 40mm buckling about the major axis);
- Curve d represents shapes with maximum imperfections: it is to be used for hot-rolled I-
sections with very thick flanges (tf > 100 mm) and thick welded I-sections (tf > 40 mm), if
buckling occurs in the minor axis.
Figure 2.15 shows how to choose the right curve for each shape and bending direction according
to the properties mentioned above.

Figure 2.15: Stability curves for the various kinds of profiles.

Dr. Enzo MARTINELLI 35 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Once the right curve has been chosen the value of χ can be determined; different curve have to
be used for determining the corresponding χ values with reference to the two principal axis of the
section.

Figure 2.16: Stability curves according to EC3.

The curves in Figure 2.16 can be are defined by the following general relationship in terms of the
non-dimensional slenderness defined by equation (2.73):
1
χ= ≤ 1.0 , (2.75)
Φ + Φ2 − λ 2
with
⎣ ( )
Φ = 0.5 ⋅ ⎡1 + α ⋅ λ − 0.2 + λ 2 ⎤ .
⎦ (2.76)
The parameter α is related to the level of imperfection affecting the structural member; since
four curves have been introduced for describing the various kinds of imperfections, four values of α
have to be considered, each one for the corresponding stability curve (Table 2.1).
Table 2.5: Imperfection factors for buckling curves.

2.8.2.3 Stability check under eccentric axial load

Since there slight differences exist between the method for stability check of members under
eccentric loads according to the ENV version [7] and the final EN one [14], both procedures will be
proposed in the following. Indeed, while the first one is not yet valid, in the authors’ opinion, is fitter
for grasping the mechanical meaning of the various terms whose final EN formalization seems only
formally more complicated. Moreover, since in the symbols adopted within the Eurocodes, y and z are
the two principal axes of inertia, this choice will be assumed in the following sections.

2.8.2.3.1 ENV 1-1-1993 [13] procedure.

Different formulae are provided depending on the class of the cross section. If it belongs to
Classes 1 or 2 the following relationship, conceptually close to the one in equation (2.67) has to be
considered:
N Sd k y M y ,Sd kz M z ,Sd
+ + ≤1 ,
χ min Af ay W pl , y f ay W pl ,z f ay (2.77)
γ M1 γ M1 γ M1
where the coefficient χ min = min ( χ ( λ y ); χ ( λz )) can be evaluated according to the above remarks. Plastic
moduli W pl , y and W pl ,z are considered since plastic bending moment can be completely developed in

Dr. Enzo MARTINELLI 36 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

class 1 and 2 profiles. Second-order effects and the shape of diagram are considered through the factors
ky and kz, the first of which is defined as follows:
μ y N Sd
ky = 1− , (2.78)
χ y Af ay
and
W pl , y − Wel , y
μ y = λ y ⋅ ( 2 β My − 4 ) + , (2.79)
Wel , y
and, finally, the value of β M accounts for the shape of bending moments and can be deduced by
Figure 2.17.

Figure 2.17: βΜ factors depending on the shape of bending moment diagram.

Dr. Enzo MARTINELLI 37 Draft Version 13/01/2008

STABILITY OF STRUCTURES

Stability check of beam columns in class 3 can be carried out by simply substituting the plastic moduli
with the elastic ones in equations from (2.77) to (2.79). Finally, for slender sections (Class 4) the relevant
properties (area and strength moduli) of the effective section have to be evaluated and the bending
moments need to be updated for taking into account the eccentricities eN,x and eN,y between the original
centroid and the one of the effective section:
M y ,Sd ,eff = M y ,Sd + N y ,Sd ⋅ e Ny . (2.80)

2.8.2.3.2 EN 1-1-1993 [14] procedure.

Few formal variations has been introduced in the final version of the Eurocode accepted as EN.
In particular, a general expression for the stability check of beam-columns is proposed in the following
form:
N Sd M y ,Sd + N Ed e Ny M z ,Sd + N Ed e Nz
+ k yy + k yz ≤1
χ y N Rk χ y M y ,Rk χ y M z ,Rk
γ M1 γ M1 γ M1
, (2.81)
N Sd M y ,Sd + N Ed e Ny M z ,Sd + N Ed e Nz
+ kzy + kzz ≤1
χz N Rk χ y M y ,Rk χ y M z ,Rk
γ M1 γ M1 γ M1
provided that no lateral-torsional buckling phenomena (which will be addressed in the 4th chapter)
exist.
The values of NRk, My,Rk and Mz,Rk are defined in the following general form:
N Rk = Ai f ay , (2.82)
M y ,Rk = W y f ay M z ,Rk = Wz f ay . (2.83)
The geometrical properties reported in equations (2.82) and (2.83) can be assumed depending on the
class of the transverse section as briefly summarized in Table 2.6.
Table 2.6: Values of geometrical properties depending of class section.

No substantial differences exist among the aspects described above. Nevertheless, as one
compare equation (2.77) (and the corresponding ones for sections in class 3 or 4) with equation (2.81)
can easily observe the key difference between the two approaches. In fact, four interaction factors kij
(rather than two) are involved in equation (2.81) meaning that the bending contribution is different in
the cases of buckling occurring either in y or z direction.

Dr. Enzo MARTINELLI 38 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Table 2.7: Interaction factors according to Method 2 – Annex B [14].

Two alternative approaches are reported in EC3 [14] for determining the values of such factors;
they are reported in two different annexes at the same document. Only the so-called “Method 2”
reported in Annex B is explicitly reported herein for the sake of brevity; its formulation for the case in
which members are not susceptible of lateral-torsional buckling is summarized in Table 2.7. Finally,
Table 2.8 summarized how the equivalent uniform moment factors have to be evaluated according to
the mentioned Method 2.
Table 2.8: Equivalent uniform moment factors according to Method 2 – Annex B [14].

2.9 Applications
Application of the above theory is proposed in the following. Some worked examples deal with
the main topics covered by this section, while few unworked one are left to the reader.

2.9.1 Worked examples

Three worked examples dealing wit the key topics discussed in the above sections are proposed
in the following.

Dr. Enzo MARTINELLI 39 Draft Version 13/01/2008

STABILITY OF STRUCTURES

2.9.1.1 Euler load for a generally restrained beam-column

The first example deals with the evaluation of the Euler critical load for a generally restrained
beam-column. In particular, flexible restraints are present at both ends and their flexibility is defined as
follows:
L col
εA = , (2.84)
10 EI col
L
ε B = col . (2.85)
5 EI col
being Lcol=5.0 m the member span length and Icol the moment of inertia of the transverse section with
reference to an axis normal to the plane of possible buckling occurrence. The beam-column is
represented in Figure 2.18; since it is a non-sway member equation (2.44) or the alignment chart in
Figure 2.10a has to be used for evaluating the effective length L0 or, equivalently, the β coefficient to be
adopted in equation (2.43).

Figure 2.18: Structural scheme of the beam-column.

First of all, the values of the non-dimensional flexibilities kA and kB have to be determined according to
their definition:
εA 1
kA = = = 0.10 ,
L col EI col 10
(2.86)
εB 1
kB = = = 0.20 .
L col EI col 5
As a matter of principle, the corresponding value of the β coefficient is within the range [0.5, 1.0],
being the column a non-sway member. The equation (2.44) can be applied for determining its value:
⎛ 0.10 ⎞ ⎛ 0.20 ⎞
β = 0.5 ⋅ ⎜ 1 + ⎟ ⋅ ⎜1 + ⎟ = 0.622 . (2.87)
⎝ 0.45 + 0.10 ⎠ ⎝ 0.45 + 0.20 ⎠
The above value is rather conservative with respect to the one which could be derived by the alignment
chart in figure Figure 2.10a, as desirable for an approximate formula.

2.9.1.2 Stability check of an axially loaded beam-column

Let us consider a member in compression whose transverse section is realized by a profile HE
200 B made out of steel S235. The overall span length of the member is 7.5 m and displacement in the
direction perpendicular to the web plane constraints the displacements in that direction (Figure 2.19).

Dr. Enzo MARTINELLI 40 Draft Version 13/01/2008

 STABILITY OF STRUCTURES

Figure 2.19: Beam-column under axial force.

The transverse section of the beam is characterized by the following geometrical parameters:
- depth h 200 mm;
- width b 200 mm;
- flange thickness tf 15 mm;
- web thickness tw 9 mm;
- radius r 18 mm;
- area A 7810 mm2;
- Moment of inertia with respect to the strong axis Iy 5696 104 mm4;
- Moment of inertia with respect to the weak axis Iz 2003 104 mm4.
For the sake of brevity the exercise will be only solved with reference to the EC3 provisions for
stability check.
Step #1: classification of the transverse section:
Since the adopted steel grade is fy=235 MPa the value ε=1 can be assumed for the parameter
mentioned in Table 2.3and Table 2.4. The following values of the length-to-thickness ratios can be
evaluated for flange and web:
- flange c/tf=(200/2)/15=6.7≤10 Class 1;
- web d/tw=(200-2·15-2·18)/9=14.9≤33 Class 1
Finally, the profile HE200B made out of steel S235 is in class 1 if loaded in compression.

Step #2.1: evaluating the (elastic) Euler load for buckling along the strong axis:
Since hinged restraints can be recognized for both ends and no further constraints control
displacements in the mentioned direction, the effective length L0,y is equal to the nominal one (L=7500
mm) and the Euler load can be easily derived:
π 2 EI y π 2 210000 ⋅ 56960000
N cr , y = 2
= 2
= 2098.78 kN . (2.88)
L 0, y 7500
Step #2.2: evaluating relative (non-dimensional) slenderness along the strong axis:
β A Af ay 1 ⋅ 7810 ⋅ 235
λy = = = 0.935 . (2.89)
N cr , y 3
2098.78 ⋅ 10
Step #2.3: determination of the reduction factor χy:
According to Figure 2.15 the profile follows the curve b and, consequently, the following value of
the reduction factor χy can be evaluated:

⎣ ( )
Φ y = 0.5 ⋅ ⎡1 + α ⋅ λ y − 0.2 + λ y 2 ⎤ = 0.5 ⋅ ⎡⎣1 + 0.34 ⋅ ( 0.935 − 0.2 ) + 0.9352 ⎤⎦ = 1.062 ,
⎦ (2.90)
and
1 1
χy = = = 0.6387 . (2.91)
Φ y + Φ y −λy 2 2
1.062 + 1.062 2 − 0.9352
Step #3.1: evaluating the (elastic) Euler load for buckling along the weak axis:
Dr. Enzo MARTINELLI 41 Draft Version 13/01/2008
STABILITY OF STRUCTURES

Since transverse displacements are constrained at mid-span, the effective length L0,z is one half of
the nominal one (L=7500 mm) and the Euler load can be easily derived:
π 2 EI z π 2 210000 ⋅ 20030000
N cr ,z = 2
= 2
= 2952.10 kN . (2.92)
L 0,z 3250
Step #3.2: evaluating relative (non-dimensional) slenderness along the weak axis:
β A Af ay 1 ⋅ 7810 ⋅ 235
λz = = = 0.788 . (2.93)
N cr ,z 3
2952.1 ⋅ 10
Step #3.3: determination of the reduction factor χy:
According to Figure 2.15 the profile follows the curve c and, consequently, the following value of
the reduction factor χz can be evaluated:
⎣ ( )
Φz = 0.5 ⋅ ⎡1 + α ⋅ λz − 0.2 + λz 2 ⎤ = 0.5 ⋅ ⎡⎣1 + 0.49 ⋅ ( 0.788 − 0.2 ) + 0.788 2 ⎤⎦ = 0.955 ,
⎦ (2.94)
and
1 1
χz = = = 0.6695 . (2.95)
Φz + Φz 2 − λz 2 0.955 + 0.9552 − 0.788 2
Step #4: evaluating the ultimate axial load capacity:
The minimum value of the reduction factor evaluated along the two directions has to be
considered for determining the ultimate bearing capacity of the member.
f ay 235
N b ,Rd = χ min β A A = 0.6387 ⋅ 7810 ⋅ = 1116.4 kN . (2.96)
γ M1 1.05

2.9.1.3 Stability check of an eccentrically loaded beam-column

The beam-column represented in Figure 2.20 has the same transverse section described in the
previous example. Transverse displacements are constrained at the top of the column since the two
following values of the β coefficient can be assumed:
- buckling in the strong direction (perpendicular to y axis) βy=2.0;
- buckling in the weak direction (perpendicular to z axis) βz=1.0;

Nd=100 kN
Fd=20 kN

H=3000 mm

e=400mm
z

Figure 2.20: Beam-column under eccentric axial force.

Step #1: classification of the transverse section:
According to the findings of the previous exercise, the profile HE200B made out of steel S235 is
in class 1 if loaded in compression.

Dr. Enzo MARTINELLI 42 Draft Version 13/01/2008

STABILITY OF STRUCTURES

and
1 1
χz = = = 0.7664 . (2.106)
Φz + Φz 2 − λz 2 0.805 + 0.8052 − 0.6312
Step #5: evaluation of the relevant interaction factors:
Since no bending moment is applied around the z-axis the only interaction factor to be
determined for applying the second one of the two equations (2.69) is the term kzy. An easy relationship
is stated for determining this factor as a function of kyy as follows:
kzy = 0.6k yy = 0.625 . (2.107)
Step #6: final stability check:
The two equations (2.69) can be finally applied for checking the given structure against global
buckling:
100000 100000000
+ 1.042 ⋅ = 0.076 + 0.011 = 0.087 ≤ 1
0.7559 ⋅ 7810 ⋅ 235 0.7559 ⋅ 56960000 ⋅ 235
1.05 1.05
. (2.108)
100000 100000000
+ 0.625 ⋅ = 0.075 + 0.018 = 0.093 ≤ 1
0.7664 ⋅ 7810 ⋅ 235 0.7664 ⋅ 20030000 ⋅ 235
1.05 1.05

2.9.2 Unworked examples

The following exercises are left to the readers:
1) for the same beam-column reported in paragraph 2.9.1.1, evaluate the β coefficient in the case
of sway member. Compare the results obtained by the simplified formula and the alignment
chart;
2) for the same beam-column described in paragraph 2.9.1.2, evaluate the ultimate load bearing
capacity according to the Italian Code;
3) for the same beam-column described in paragraph 2.9.1.2, evaluate the ultimate load bearing
capacity considering a steel grade s355;
4) for the same beam-column described in paragraph 2.9.1.3, evaluate the maximum lateral load
FSd,max with reference to the global stability check of the structure;
5) for the same beam-column described in paragraph 2.9.1.3, perform the stability check according
to the Italian code.

Dr. Enzo MARTINELLI 44 Draft Version 13/01/2008