You are on page 1of 8

Creating and Renewing Urban Structures

Second Order Effects on Building Structures an Approximate Evaluation


Lszl P. KOLLR
Professor
Budapest University of
Technology and Economics,
Department of Mechanics,
Materials and Structures,
Budapest, Hungary

Lszl P Kollr, born 1958,


received his civil engineering
degree (1982), PhD (1986) and Dr.
habil (1996) from the Technical
University of Budapest.

lkollar@eik.bme.hu

Summary
Slender building structures must be designed by taking into account the second order effects.
Whether this effect is significant or not, can be decided by comparing the vertical load to the
buckling load of the building. It was suggested by several researchers that second order effects can
be neglected, when this ratio is below ten percent. This statement is, for example, part of Eurocode2 (Design of Concrete Structures), which also contains approximate procedures, how to evaluate the
buckling load of buildings. It is also well known that the second order effect can be taken into
account by a fictitious horizontal load, which is a function of the vertical load/buckling load ratio.
The key of the evaluation is the proper approximation of the buckling load, which must contain the
effect of the stiffnesses of the frames and/or the shear walls, the deformability of the foundations,
the effect of the number of stories, etc. Expressions, which takes into account these effects are also
given in Eurocode, however, for some cases, they may lead to unacceptable, unconservative results.
In this paper some of the formulas are corrected, and for a few cases better approximations are
suggested. The expressions are based on the solutions of the replacement continua of buildings.
Keywords: second order effect; Eurocode; approximation; summation theorem; high-rise buildings.

1. Introduction
Loads mostly at the floors
Continuos load
Second order effects in buildings
may be ignored, if they are
significantly
less
than
the
corresponding first order effects (e.g.
less than 10%, [7]). This criterion is
H
part of the building codes, for
uniform
example Eurocode 2 (Design of
stiffnesses
Concrete Structures) and Eurocode 8
(Design of structures for earthquake
Fig. 1: The structure and the replacement continuum
resistance). The former one states
that second order effects can be neglected, if the following relation holds:

FV, Ed 0.1 N cr ,

(1)

17TH CONGRESS OF IABSE, CHICAGO, 2008

where FV, Ed is the total vertical load,


while N cr is the global buckling load.
To obtain the buckling load for
(regular)
building
structures
approximate
expressions
are
suggested, which are based on the
buckling load of replacement columns
with uniform stiffnesses (Fig. 1). The
columns may have bending stiffness
(D=EI) and/or shear stiffness ( S ).
Buckling loads of columns with
concentrated loads on the top and
continuous load along the height
having either bending deformation or
shear deformation are summarized in
Tables 1 and 2 [6, 8].

Table 1: Critical load, Ncr of a cantilever (bending or


shear deformation only)
Loading

Bending deformation only

N cr =

2 EI
4H 2

N cr = 7.837

EI
H2

Shear deformation only

N cr = S
N cr = S
Using these results the buckling loads
are approximated with the summation
theorems [5], which are summarized Table 2: Critical load, Ncr of a rigid cantilever supported
in Table 3 and discussed briefly below. by a spring
Loading
When a structure is subjected to the
sum of different loads (e.g. N1 and N2,
Table 3, left), according to
Dunkerleys theorem its buckling load
(critical load parameter) can be
k
k
approximated with the buckling loads
(critical load parameters) of the
individual loads, as it is shown in
k
k
Table 3, left.
N cr, c =
N cr, c = 2
H
H
Table 3: Illustration of summation
theorems: Approximate expressions for the critical load parameter

Dunkerley summation

1
1
+
cr
cr,1 cr,2

Fppl summation

1
1
N cr
+
N cr,1 N cr,2

Southwell summation

Ncr Ncr,1 + Ncr,2

For the illustration of the Fpple theorem, consider a structure, which is characterized by two

Creating and Renewing Urban Structures

stiffnesses in such a way that if any of the stiffnesses is set equal to zero the structure becomes a
mechanism. According to Fppls theorem, the buckling load can be approximated with the
buckling loads of two structures, which are obtained from the original one by setting one of the
stiffnesses equal to infinity. This is illustrated in Table 3, middle.
Finally, consider a structure, which is characterized by two stiffnesses in such a way that if any of
the stiffnesses is set equal to infinity the structure becomes rigid. According to Southwells theorem,
the buckling load can be approximated with the buckling loads of two structures, which are
obtained from the original one by setting one of the stiffnesses equal to zero. This is illustrated in
Table 3, right.

2. Buckling load of shear walls


When the structure is braced by solid shear walls, the buckling load can be approximated by
assuming a distributed load along the height of the structure, which results in (Table 1, right)
N cr, B =

D
H2

= 7.837 7.8 ,

(2)

where H is the height of the building and D is the sum of the bending stiffnesses (EI) of the shear
walls. This expression may be used only, if the foundation is preventing the rotation of the base of
the shear wall, the shear deformations of the walls are negligible, and the number of stories is high
enough to assume uniform load along the height. If these approximations are unrealistic, expression
(2) must be modified. This will be discussed in the following
subsections.
2.1

Modification due to the rotation of the base

We may define the rotational stiffness of the base as (Fig. 2)


k = M / . Using Fppls expression (Table 3, middle), we obtain
that the buckling load can be calculated as:
1
1
N cr =
+
N cr,1 N cr,2

1
1

=
+
N cr, B 2k / H

= N cr, B

1
= N cr, B
. (3)
D
D / H 2
1 + 3.9
1+
k H
2k / H

We note that this formula (which was derived by Zalka [6, 8]) is given in Eurocode 2 with the
multiplier 0.7, instead of 3.9. This must be a misprint with serious consequences, as it may
overpredict the buckling load by up to 400%.
2.2

Modification due to the shear deformation

When the shear walls have openings, the shear deformation can be significant and may not be
neglected. The inverse of the shear stiffness ( S ) can be defined as the interstorey drift over storey
height from unit horizontal load. Columns having both flexural (bending) and shear deformations
are called Timoshenko beams. Their buckling load can be approximated by Fppls expression,
which gives:

17TH CONGRESS OF IABSE, CHICAGO, 2008

1
1
+
N cr =
N cr,1 N cr,2

1
1
=
+
N cr, B S

1
.
N cr, B
1+
S

= N cr, B

(4)

This expression, given also in Eurocode 2, is conservative, however may have an error up to 40%. A
much better approximation was derived by Hegedus and Kollar [1], see Table 5. With this
expression we obtain
N cr = N cr, B

1
,
N cr, B
1 + 0.6
S

with the upper limit N cr S .

(5)

When there are several bracing elements, Eurocode 2 suggests that the shear stiffness, S is
calculated as the sum of their shear stiffnesses. We must emphasize that simple summation may lead
to unconservative results if the shear stiffness / bending stiffness ratios of the bracing elements are
different. (For example, let us consider two bracing elements: one with high bending, but low shear
stiffness, while the other with low bending and high shear stiffness. If the stiffnesses are determined
by summation, both the calculated bending and the shear stiffness are high, which is not realistic.) A
possible solution is that the buckling loads of the bracing elements are calculated individually, and
then, according Southwells summation, the buckling loads (not the stiffnesses) are added together.

2.3

Modification due to the number of storeys

PF
h

PF=ph

PF

Fig. 3: Unifromly distributed load


replaced by concentrated forces

PT

PF/2
PF

....

PF/2

PF

PF

PF

PF

PT

PF/2

PF

(c)
(b)
(a)
Fig. 4: Separation of concentrated forces (a) into two
groups. a: uniform distribution and b: force on the top

Most of the loads are acting at the levels of the floors (Fig. 1), however in the analysis above we
assumed a uniformly distributed load along the height. When a distributed load (p) is replaced by
concentrated loads (Fig. 3) it results in a force ph at every storey (where h is the height of the
storey), except at the very top, where only ph/2 is acting. As a consequence, the concentrated storey
loads must be separated into two groups such that the first group can be replaced by a uniform load
(Fig. 4b), while the rest is a concentrated load on the top (Fig. 4c).
This additional concentrated force can be taken into account by using Dunkerleys expression.
From Table 1, the critical load parameter for uniformly distributed load is

cr,1 pH = cr,1

PF
EI
,
H = 7.837
h
H2

cr,1 =

7.837 EI
,
nPF H 2

while for (PT PF / 2 ) concentrated force on the top:


2 EI
2 EI
,

,
cr,2 =
cr,2 (PT PF / 2) =
4H 2
4 H 2 (PT PF / 2 )

(6)

(7)

Creating and Renewing Urban Structures

where PT and PF are the loads on the top of the building and on an arbitrary storey, respectively.
From Dunkerleys expression (Table 3) we obtain:

1
1
+
cr =
cr,1 cr,2

n
.
(8)

PT
n + 1.59 2
1

PF
This expression (with PT = PF ) is given in Zalka [6, 8], and in Eurocode 2. Note that multiplier n
decreases with the number of storeys, n. (The application of the above expression is recommended
only if PT PF / 2 ).
,

N cr = N cr, B n ,

n =

When the deformation of the wall is dominated by shear (flexural deformation is negligible) the
buckling load is not affected by the position of the load along the height, hence the number of
storeys has no effect either.
When the deformation is governed by the rotation of the base (rigid wall), the reduction due to the
storey number is less than n (defined in Eq.8).
When all the three effects discussed in the previous subsections are taken into account, we may use:

N cr, n

1
D
1 + 0.6 N cr, n / S
N cr = N cr,S
, N cr, n = N cr, B n , N cr, B = 7.8
.
, N cr,S = min
{
2
N cr,S
H

storey
1+

2 k
number
S

n
144424443
H
shear deformation
14243
base rotation

(9)
where n is defined by Eq.(8). Note that coupled shear walls may behave like frames [2, 6, 8],
which are discussed in the next section.

3. Buckling load of frames


The simplest model of a frame structure (Fig. 5) is a replacement column, which has shear
deformation only. In this case the shear stiffness can be determined from the deformations of one
storey (Fig. 6). It can be approximated by the following expression [2, 8]:
S =

1
1 / S b + 1 / S c

where

n 112 EI
bi ,
S b =
d
h
i =1
i

n 2 EI
ci ,
S c =
2
i =1 h

(10)

where n and n1 are the number of columns and beams, EIci and EIbi are their bending stiffnesses, h
is the storey height and di is the bay length. The buckling loads are given in Table 1.
This model was improved in the fifties by Sigalov, Beck and Csonka [2, 8], and a coupled bar
model was introduced, which has both shear and bending stiffness. This model is referred to as a
Csonka beam. Its shear stiffness is calculated by Eq.10, while the bending stiffness is [2]:
Dl =

n
1
EI ci .
1 + S / S i =1

(11)

17TH CONGRESS OF IABSE, CHICAGO, 2008

Its buckling load under concentrated load is given in Table 4. When the column is subjected to a
distributed load the critical load was determined by Zalka [6, 8], an approximate expression was
fitted on his results, which is presented in Table 5. (Note that Suthwells expression, which is also
presented in Table 5, may underestimate the buckling load by almost 50%.)

Table 4: Critical load of columns with shear and bending deformations used as replacement
continua of buildings (concentrated load)
Timoshenko beam
N

D0
S

Stiffnesses

Critical load
(Maximum error)

Dl

bending: D0
shear: S
1
1

+
N 0 S

bending: Dl
shear: S
1

(0 %)
N0 =

Csonka beam
N

2 D0
4H 2

S + N l
(0 %)

Nl =

Sandwich beam
N

Dl

D0
S

Global bending: D0
Local bending: Dl
shear: S
1
1

+
N 0 S

+ Nl

(0 %)

2 Dl
4H 2

Both of these models neglect the compressibility of the columns, and hence the presented formulas
may overestimate the buckling load. For slender frames, a new bending stiffness is introduced,
which can be calculated from the parallel axes theorem, as follows:
n

D0 = EAci si2 ,
i =1

(12)

where EAci is the tensile stiffness of the ith column, and si is its distance from the center of the
building. These three stiffnesses ( D0 , Dl , S ) are the rigidities of a sandwich beam. (This model
was introduced for the analysis of wind loads by Rosman and for earthquake analysis by Rutenberg,

Creating and Renewing Urban Structures

and Skattum [8, 3, 4]). Its buckling load under concentrated load is given in Table 4. When the
column is subjected to a distributed load the critical load is given in [2]. Two approximate
expressions are presented in Table 5.
Eurocede 2 does not give a criterion on the calculation of the buckling load in case of (unbraced)
frames, the expressions given in Table 4 and 5 are recommended.
Table 5: Critical load of columns with shear and bending deformations used as replacement
continua of buildings (uniformly distributed load) (All the expressions are conservative, i.e. they
underpredict the buckling load.)
Timoshenko beam

Csonka beam
Dl

D0
S

Critical load
(Summation
theorem)

Improved critical
load

(Maximum error)
N 0 = 7.837

D0

Global bending: D0
Local bending: Dl
shear: S

bending: Dl
shear: S

1
1

+
N 0 S

(Maximum error)

Dl

D0
S

bending: D0
shear: S

Stiffnesses

Sandwich beam

(29 %)
0.6
Tim 1

N cr
=
+
S
N0
with N Tim S

1
1

+
N 0 S

S + N l
(47 %)
1

cr

1
S + Nl
1 f

(5 %)

(5 %)

D
, N l = 7.837 l ,
2
H
H2

f = 8.2

+ Nl

(47 %)

Csonka
N cr
=

Tim
N cr + N l

max
1

1
1

Csonka
N cr
N 0

(25 %)
8.2

1 + 0.12 e 0.28 e 0.6

, =

SH 2
Dl

Eurocode 8 (Design of structures for earthquake resistance) presents the following rule: second
order effects need not be taken into account, if the following condition is satisfied for all storeys:

Ptot d r
0.10 ,
Vtot h

(13)

where Ptot is the total load above the storey, dr is the interstorey drift, Vtot is the shear force. We
may observe that d r / h = (Fig.6), and hence Eq.(13) is approximately the total load over the
shear stiffness of one storey. If the compressibility of the columns is negligible Eq.(13) is identical

17TH CONGRESS OF IABSE, CHICAGO, 2008

to the condition of Eurocode 2, given also by Eq.(1). However, for slender frames Eq.(13) should be
modified, and the compressibility of the columns must be taken into account. For example, by using
the third row of Table 5 (Timoshenko beam), we arrive at the following expression:
1
H 2 Ptot d r
+
=
S 7.8 D0 Vtot h

= Ptot

Vtot h H 2
1 +
0.10 ,

d
7
.
8
D
r
0

(14)

or, to obtain a better approximation, the fourth row of Table 5 (Timoshenko beam) can be used,
which results in:

0.6
H 2
Ptot
+
=
7.8 D0

= max S
Ptot

Ptot d r
Vtot h

0.6 + Vtot h H

d r 7.8 D0

0.10 .
Ptot d r

Vtot h

(15)

4. Conclusion
In this paper simple expressions are presented for the calculation of the buckling load of building
structures. Based on these expressions, modifications are suggested in the expressions of Eurocode
to avoid unconservative (or too conservative) results.

Acknowledgements.
This work was supported by the Hungarian Academy of Sciences. Professor Zalkas useful
comments are highly appreciated.

References
[1]

HEGEDS, I. and KOLLR, L. P., Buckling of sandwich columns with thin faces under
distributed normal loads, Acta Technica Hungarica. Vol. 97. 1984, pp. 111-122.

[2]

HEGEDS, I. and KOLLR, L. P., Application of the sandwich theory in the stability
analysis of engineering structures, (Chapter of the book ``Structural stability in engineering
practice Ed. Lajos Kollr) E & FN Spon, London, 1999

[3]

RUTENBERG, A., Approximate natural Frequency of Coupled Shear Walls. Earthquake


Engineering and Structural Dynamics. Vol. 4. 1975. pp. 95-100

[4]

SKATTUM, S.K., Dynamic Analysis of Coupled Shear Walls and Sandwich Beams, PhD
Thesis, California Institute of Technology, 1971.

[5]

TARNAI, T., Summation theorems concerning critical loads of bifurcations, (Chapter of the
book ``Structural stability in engineering practice Ed. Lajos Kollr) E & FN Spon, London,
1999

[6]

ZALKA, K.A. and ARMER, G..S.T., Stability of Large Structures, Butterworth Heinemann,
Oxford, etc.1992

[7]

ZALKA, K.A. and MACLEOD, I A, The equivalent column concept in stability analysis of
buildings, The Structural Engineer, Vol. 74, No 23-24, 1996, pp. 405-411.

[8]

ZALKA, K.A., Global Structural Analysis of Buildings, E& FN Spon. 2000, London.

You might also like