Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

Contents lists available at SciVerse ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Improved earthquake resistant design of eccentric steel buildings

Miltiadis T. Kyrkos, Stavros A. Anagnostopoulos n
Department of Civil Engineering University of Patras, Greece

a r t i c l e i n f o a b s t r a c t

Article history: Recently, a design modification has been proposed for eccentric, torsionally stiff, braced steel buildings,
Received 9 October 2011 designed according to the current Eurocodes 3 and 8, that improves noticeably their inelastic response
Received in revised form under the action of design level earthquakes. The improvement consists in a more uniform distribution
4 March 2012
of ductility demands throughout the building. In the present paper, a similar, though differently
Accepted 10 July 2012
Available online 1 August 2012
derived, modification is applied to torsionally flexible eccentric buildings and their response is again
evaluated under pairs of design earthquake motions. A substantial improvement of their inelastic
Keywords: response is also observed, similar to what had been obtained for torsionally stiff buildings. The new
Asymmetry approach is also tested with torsionally stiff buildings and leads to similarly satisfactory results. Thus it
Torsion
may be recommended for general application.
Multi-story steel buildings
& 2012 Elsevier Ltd. All rights reserved.
Braces
Non-linear seismic response
Design eccentricity
Plastic hinge model
Torsionally flexible buildings

1. Introduction models of the shear beam type [1,11,12,13]. Only recently,

One of the open research areas in earthquake resistant design
of buildings is associated with torsion caused by strong earth-
quake excitations when the building is irregular or simply non-
symmetric. Torsional motion during earthquakes can be caused
due to a number of factors, some of which can be accounted for in
design, while some others may be unknown at the time the
structure is designed or may be difficult to quantify and be
accounted for properly. Examples of such factors are accidental
eccentricities due to mass and/or stiffness uncertainties and non
coherent ground motion at the supports. Given also that buildings
are designed to respond in the inelastic range under strong
earthquakes, stiffness changes due to non symmetric yielding
that current design and analysis methods cannot reliably predict,
induce additional eccentricities and hence extra torsional motion.
This problem has led to many studies, in a number of which code
provisions for torsion have been questioned [1–9]. Thus it comes
as no surprise that torsion has frequently been identified as a
main cause of irregular building failures during catastrophic
earthquakes [10]. Most of the past studies of earthquake induced
torsion in irregular buildings have been based on elastic analyses
in multistory buildings or on simplified one-story, inelastic
n
Corresponding author.
E-mail addresses: mkyrkos@upatras.gr (M.T. Kyrkos),
saa@upatras.gr (S.A. Anagnostopoulos).
detailed inelastic building models of the plastic hinge type have
been used to study the inelastic response of realistic, code
designed asymmetric buildings under the action of two compo-
nent earthquake ground motions [5,14,15,16,17,18]. These more
recent studies have generated some interesting results that cast
doubt about the adequacy of modern code provisions for frame
type, multistory, asymmetric buildings. More specifically, using
the response of the associated symmetric building as the basis for
comparison, it has been found that when such code designed,
actual eccentric buildings are subject to strong earthquake
motions, the ductility demands in the so called ‘‘flexible’’ edge
are substantially greater than the ductility demands in the so
called ‘‘stiff’’ edge of the building. Obviously, such uneven
distribution is undesirable as it can lead to premature member
failures. This was initially found for concrete buildings where
both rotational ductility factors and damage indices were used as
measures of inelastic deformations [3,5], but the same behavior
has also been confirmed for torsionally stiff, braced, steel frame
buildings, designed in accordance with Eurocodes 3 and 8
[15,16,17,18]. The observed results are the opposite of what has
been reported in past publications based on simplified one-story
models, in which the load bearing, shear beam type, elements
were designed only for lateral load resistance.
In the present paper, the distribution of earthquake induced
ductility demands is examined for multistory, torsionally flexible,
eccentric steel buildings, designed in accordance with Eurocodes
3 (for steel structures) and 8 (for earthquake resistance). In these

0267-7261/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soildyn.2012.07.011
 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156
buildings the fundamental mode is torsional and the respective
period longer than the periods of the two lowest translational
modes. Three sets of buildings are designed: The first with one
story, the second with three stories and the third with five stories.
Fig. 1. Layout of 1, 3 and 5-story steel buildings.
Fig. 2. Elevations in X and Y directions of the 3-story buildings.
145
Each set includes a symmetric building—with a uniform mass and
stiffness distribution- and two eccentric buildings with normal-
ized biaxial mass eccentricities e ¼0.10 and e ¼0.20. These build-
ings are subjected to sets of ten, two component semi-synthetic
motions, generated to closely match the design response spec-
trum. Rotational ductility demands of beams and axial ductility
demands of braces, as well as interstory drifts and total displace-
ment in every story are used to evaluate the overall performance
of the buildings. The objective is to investigate whether the
results obtained for torsionally stiff eccentric steel buildings
[17], are also observed in torsionally flexible buildings and in
such case propose some design modification to alleviate the
problem, as done for the torsionally stiff buildings.
2. Methodology
The present investigation was carried out using three sets of
typical buildings, with one, three and five stories each set, all steel
braced frames. The layout, same for all floors of all three building
sets, can be seen in Fig. 1. In Fig. 2, the elevations of the two sides
of the 3-story building are also shown. Each building is formed by
4 frames along the x axis, (FR-X01 to FR-X04) and 4 frames along
the y axis (FR-Y01 to FR-Y04). In order to have torsionally flexible
buildings for the purpose of our work, braces were used to stiffen
the central bays as shown in Fig. 1. All buildings have a typical
story height of 3.00 m and ground story height 4.00 m.
146 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

Using appropriate distributions of the floor loads, e.g. through
non-symmetric live load distribution, non-symmetric balconies
(common causes of mass eccentricity in typical Greek buildings,
not shown in the given layout), non-symmetric joint masses were
assigned at each floor and thus biaxial mass eccentricities were
introduced in all floors. In this manner, in addition to the
symmetric layouts for each of the 1, 3 and 5-story buildings,
eccentric variants were generated and designed with the follow-
ing mass eccentricities: em ¼0.10 L and em ¼0.20 L, where L is the
building length along each direction.
The models used for both design and analyses are 3-D models
with masses lumped at the joints and the floors acting as
diaphragms. All buildings were designed as spatial frames for
gravity and earthquake loads using the dynamic, response spec-
trum method, according to Eurocodes EC3- steel structures- and
EC8-earthquake resistant design. Earthquake actions were
0.80
MEAN
Spectral Acceleration (g)
described by the design spectrum specified by the Greek Code
for ground acceleration PGA¼ 0.24 g and soil category II. As input
for the nonlinear dynamic analyses, ten sets of two component
semi-artificial motion pairs were used. They were generated from
a group of five, two-component, real earthquake records, to
closely match the code design spectrum (with a descending
branch 1/T2/3 ), using a method based on trial and error and
Fourier transform techniques [19]. Results were excellent, as
Fig. 3 indicates, where the mean response spectrum of the ten
semi-artificial motions is compared with the target design spec-
trum. Each synthetic motion pair, derived from the two horizontal
components of each historical record, was applied twice by
mutually changing the components along the x and y system
axes. Thus, each design case was analyzed for ten sets of
2-component motions and mean values of peak response indices
were computed. In this manner, the effects of individual motions
are smoothed and the conclusions become less dependent on
specific motion characteristics. The design spectrum can be seen
in Fig. 3, along with the mean spectra of the motions used for

Ag,max=0.24g subsequent analyses. It should be noted that the dimensioning of

0.60 the frame members took into account the uneven distribution of
member forces due to the mass eccentricities and hence stiffness
eccentricities were also generated, as it happens in actual practice.
0.40
The complete set of the lowest three periods of all building
variants in each set is listed in Table 1, along with the initial mass
0.20 eccentricities emx ¼ emy ¼0.10 and 0.20 and the resulting physical
eccentricities ex and ey. The latter are the mean distances (for all
0.00 stories) between the mass center (CM) and the approximate
0.00 1.00 2.00 3.00 4.00 5.00 stiffness (or rigidity) center (CR) in each floor, normalized by the
Period (sec) length of the corresponding building side. It is noted that in
multistory buildings, the CR cannot be really defined, except under
Fig. 3. Design spectrum and mean spectrum of the ten semi-artificial motions. very restrictive conditions. Thus, an approximate CR was computed
herein for reference purposes, on a floor by floor basis as follows:

Table 1 P
m P
n
K f iy xi K f ix yi
Eccentricities and fundamental periods of the buildings. i¼1 i¼1
esx ¼ esy ¼ ð1Þ
Pm Pn
Story number Mass eccentricity Mean natural Fundamental periods of K f iy K f ix
eccentricity buildings (sec) 1 1

emx ¼ emy ex ey Ty Tx Ty
 1 X
24E 2 1 1 AE
K f i ¼ 2
P þP þP þ cos2 f ð2Þ
1 0.00 0.00 0.00 0.48 0.36 0.32 h Kc K ba K bb L
0.10 0.085 0.075 0.49 0.34 0.30
0.20 0.17 0.15 0.53 0.32 0.27
where: esx, esy are the x and y coordinates of the approximate
3 0.00 0.00 0.00 0.82 0.56 0.55 stiffness center CR, Kf-i designates the approximate story stiffness
0.10 0.085 0.075 0.84 0.54 0.52 of frame i, x and y the directions of the frame axes, m and n the
0.20 0.165 0.145 0.88 0.53 0.46 number of frames along the y and x axes, respectively, E¼modulus
5 0.00 0.00 0.00 1.11 0.84 0.78
of elasticity, K c ¼ Ic =h,K b ¼ Ib =‘, Ic, Ib ¼ section moment of inertia of
0.10 0.09 0.085 1.18 0.81 0.75
0.20 0.17 0.155 1.27 0.77 0.67 columns and beams, respectively, h¼story height and ‘ ¼beam
length, A¼area of brace section, L¼brace length and j ¼ angle of

Fig. 4. (a) Nonlinear moment-rotation relations for beam-columns, (b) Column M–N interaction diagram and (c) nonlinear force deformation diagram for braces.
 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156 147

brace member and the horizontal plane. The second indices, a and b,
in Kba kai Kbb designate the upper (above) and lower (below) floor
beams of the frame in the considered story. From Table 1 we can see
that these physical eccentricities vary from 0.075 to 0.17. It is also
noted that for all models, the first torsional period is longer than the
two translational periods, so all buildings are torsionally flexible.
3. Non-linear dynamic analyses
The non linear analyses were carried out using the program
RUAUMOKO [20]. Frame beams and columns were modeled with
the well-known plastic hinge model, in which yielding at member
ends is idealized with plastic hinges of finite length having
bilinear moment-curvature relationship and strain hardening
ratio equal to 0.05. A moment-axial force interaction diagram
was also employed for columns, giving the yield moment as a
function of the applicable axial force on the column section.
Bracing members, yielding in tension and buckling in compres-
sion, were modeled with a non-symmetric bilinear force-axial
deformation relationship (Fig. 4).
The basic measure used to assess the severity of inelastic
response is the ductility factor of the various members. For
bracing members the ductility factor is defined as:
 
u
mu ¼ 1þ p ð3Þ
uy

Table 2 where up is the maximum plastic member elongation and uy the

Edge displacements for the one-story frames (in m). elongation at first yield.
For beams and beam-columns the rotational ductility factor
Eccentricity Direction Y Direction X
has traditionally been defined as:
 
‘‘Flex’’–edge ‘‘Stiff’’–Edge ‘‘Flex’’–edge ‘‘Stiff’’–edge y
(FR-Y04) (FR-Y01) (FR-X01) (FR-X04) my ¼ 1 þ p ð4Þ
yy
e ¼0.00 0.0201 0.0201 0.0325 0.0325 where yp is the maximum plastic hinge rotation at either end of a
e ¼0.10 0.0266 0.0244 0.0450 0.0263
e ¼0.20 0.0320 0.0294 0.0511 0.0269
member (beam or column) and yy is a normalizing ‘‘yield’’
rotation, typically set equal to yy ¼ M y ‘=6EI. For columns, the

Table 3
Member ductility demands for one-story frames.

Eccentricity Braces Beams

Direction Y

‘‘Flex’’–edge (FR-Y03) ‘‘Stiff’’–edge (FR-Y02) ‘‘Flex’’–edge (FR-Y04) ‘‘Stiff’’–edge (FR-Y01)

e ¼ 0.00 3.55 3.55 1.51 1.51

e ¼ 0.10 3.62 2.73 2.16 1.08
e ¼ 0.20 3.59 2.36 1.49 1.26

Direction X
‘‘Flex’’–edge (FR-X02) ‘‘Stiff’’–edge (FR-X03) ‘‘Flex’’–edge (FR-X01) ‘‘Stiff’’–edge (FR-X04)

e ¼ 0.00 2.13 2.13 1.16 1.16

e ¼ 0.10 2.12 1.85 1.76 1.26
e ¼ 0.20 2.10 1.76 1.31 1.00

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

TOTAL DISPLACEMENTS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
3 3
Story

Story

2 2

1 1
0.00 0.05 0.10 0.00 0.05 0.10
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
symm FR-X01(flex) FR-X04(stiff) symm FR-Y04(flex) FR-Y01(stiff)
3 3

2 2
Story

Story

1 1

0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04
Relative displacement (m) Relative displacement (m)

Fig. 5. Total displacements and interstory drifts of 3-story building with em ¼0.10 and comparison with the symmetric building (FR-X01 & FR-Y04: ‘‘flexible’’ edges,
FR-X04 & FR-Y01: ‘‘stiff’’ edges).
148 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

yield moment My is usually taken to correspond to the yield In addition to the above measures, peak floor displacements
moment under the action of gravity loads. In the present study, an and interstory drifts are used to assess the inelastic behavior of
alternative definition of the rotational ductility factor, based on the buildings.
the post yield plastic moment, has been used [21],
 
DM 4. Results from non-linear analyses of ‘‘As designed’’
m ¼ 1þ ð5Þ
buildings
pM y

where: DM ¼Mmax  My, My ¼yield bending moment and p¼ 0.05,
the strain hardening ratio. It has been shown that this definition is
essentially a rotational ductility factor, ratio of the maximum end
rotation (including the elastic joint rotation plus the plastic hinge
rotation of the member end), divided by a ‘‘yield rotation’’
reflecting the instantaneous boundary conditions of the consid-
ered beam [21].
Results from time history analyses for the various buildings are
presented in terms of mean values of the peak response parameters
over the ten pairs of applied motions. In the case of the beam ductility
factors, the response parameter averaged over the ten pairs of motion
is the maximum rotational ductility demand in any of the beams in
the considered frame and floor. Following standard terminology
based on static application of the lateral load in torsionally stiff

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

TOTAL DISPLACEMENTS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
3 3
Story

Story
2 2

1 1
0.00 0.05 0.10 0.00 0.05 0.10
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
symm FR-X01(flex) FR-X04(stiff) symm FR-Y04(flex) FR-Y01(stiff)
3 3

2 2
Story

Story

1 1

0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04 0.05
Relative displacement (m) Relative displacement (m)

Fig. 6. Total displacements and interstory drifts of 3-story building with em ¼0.20 and comparison with the symmetric building (FR-X01 & FR-Y04: ‘‘flexible’’ edges,
FR-X04 & FR-Y01: ‘‘stiff’’ edges).

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

AXIAL STRAIN DUCTILITY IN BRACES
symm FR-X02(flex) symm FR-Y03(flex)
FR-X03(stiff) FR-Y02(stiff)
3 3
Story

Story

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
3 3
Story

Story

2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 7. Member ductility demands of 3-story building with em ¼ 0.10 and comparison with the symmetric building (FR-X01,X02, & FR-Y03,Y04: ‘‘flexible’’ sides,
FR-X03,X04 & FR-Y01,Y02: ‘‘stiff’’ sides).
 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156 149

buildings, the edge where the displacement from rotation is added to
the pure floor translation is called ‘‘flexible’’ edge, while the opposite
edge, where the displacement due to rotation is subtracted from the
pure translation is called ‘‘stiff’’ edge. Since the examined buildings
have biaxial eccentricity, the edge distinction just mentioned applies
to both the x and y horizontal directions of the buildings. Thus, results
are presented for each edge frame and each direction. In torsionally
flexible buildings, however, it is not necessarily the ‘‘flexible’’ edge
that experiences the largest translation but it could well be the stiff
edge, depending on the relative values of the torsional and transla-
tional periods and on the input characteristics.
4.1. One-story frames
Results for the one-story frames are summarized in
Tables 2 and 3. Ductility demands are listed only for beams and
brace members because the columns remained essentially elastic.
Looking into Table 2, we can see that displacements at the
‘‘flexible’’ edges of the two eccentric buildings are, as expected,
substantially greater than those at the ‘‘stiff’’ edges due to the
induced earthquake rotations. The same occurs for the ductility
demands of the braces and the beams as the demands are again
greater in the ‘‘flexible’’ edges.

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

AXIAL STRAIN DUCTILITY IN BRACES
symm FR-X02(flex) symm FR-Y03(flex)
FR-X03(stiff) FR-Y02(stiff)
3 3
Story

Story
2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
3 3
Story

Story

2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 8. Member ductility demands of 3-story building with em ¼0.20 and comparison with the symmetric building (FR-X01,X02, & FR-Y03,Y04: ‘‘flexible’’ sides,
FR-X03,X04 & FR-Y01,Y02: ‘‘stiff’’ sides).

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

TOTAL DISPLACEMENTS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
5 5

4 4
Story

Story

3 3

2 2

1 1
0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
symm FR-X01(flex) FR-X04(stiff) symm FR-Y04(flex) FR-Y01(stiff)
5 5
4 4
3 3
Story
Story

2 2

1 1

0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04 0.05

Relative displacement (m) Relative displacement (m)

Fig. 9. Total displacements and interstory drifts of 5-story building with em ¼0.10 and comparison with the symmetric building (FR-X01 & FR-Y04: ‘‘flexible’’ edges,
FR-X04 & FR-Y01: ‘‘stiff’’ edges).
150 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

4.2. Three-story frames
Displacement results for the three-story frames are presented
in Figs. 5 and 6 for biaxial mass eccentricities emx ¼ emy ¼0.10 and
0.20, respectively, where in the same figures values for the
symmetric case are also shown for comparison. Similarly, ducti-
lity demands in the braces and beams of the same buildings are
presented in Figs. 7 and 8 for emx ¼ emy ¼0.10 and 0.20
respectively.
We observe that the displacements of the asymmetric build-
ings, of both ‘‘stiff’’ and ‘‘flexible’’ edges, are greater than those of
the symmetric case. This is due to the dominant contribution of
the torsional motion to the total building response, as a result
of which the ‘‘stiff’’ edge may even experience larger total
displacements than the ‘‘flexible’’ edge, as can be seen in Fig. 6
along direction x for both values of the eccentricity. This does not
happen with torsionally stiff buildings. As far as ductility
demands, the differences in the braces that are placed near the
core are very small. They become large in the beams at the two
edges of the buildings with the ‘‘flexible’’ edge experiencing
substantially greater demands than the ‘‘stiff’’ edge, same as in
torsionally ‘‘stiff’’ buildings [ 16].
4.3. Five-story frames
Displacement results for the five-story frames are presented in
Figs. 9 and 10 for biaxial mass eccentricities emx ¼ emy ¼0.10 and
0.20, respectively, where in the same figures values for the

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

TOTAL DISPLACEMENTS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
5 5

4 4
Story

Story
3 3

2 2

1 1
0.00 0.05 0.10 0.15 0.00 0.10 0.20
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
symm FR-X01(flex) FR-X04(stiff) symm FR-Y04(flex) FR-Y01(stiff)
5
5
4
4
3
Story

3
Story

2 2
1 1
0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04 0.05 0.06
Relative displacement (m) Relative displacement (m)

Fig. 10. Total displacements and interstory drifts of 5-story building with em ¼ 0.20 and comparison with the symmetric building (FR-X01 & FR-Y04: ‘‘flexible’’ edges,
FR-X04 & FR-Y01: ‘‘stiff’’ edges).

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

AXIAL STRAIN DUCTILITY IN BRACES
symm FR-X02(flex) symm FR-Y03(flex)
FR-X03(stiff) FR-Y02(stiff)
5 5

4 4
Story

Story

3 3

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
5 5

4 4
Story

Story

3 3

2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 11. Member ductility demands of 5-story building with em ¼0.10 and comparison with the symmetric building (FR-X01,X02, & FR-Y03,Y04: ‘‘flexible’’ sides,
FR-X03,X04 & FR-Y01,Y02: ‘‘stiff’’ sides).
 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156 151

symmetric case are also shown for comparison. Similarly, ducti- each horizontal direction:
lity demands in the braces and beams of the same buildings are ui,f lex ui,stif f
presented in Figs. 11 and 12 for emx ¼ emy ¼0.10 and 0.20 f i,f lex ¼ 2 f i,stif f ¼ 2 ð6Þ
ðui,f lex þ ui,stif f Þ ðui,f lex þ ui,stif f Þ
respectively.
Again, the location of the braces near the core, results in small where ui,flex is the top story displacement of the ‘‘flexible’’ edge
differences in their ductility demands at the various sides in both in the i-direction and ui,stiff the top story displacement of the
x and y directions. On the other hand, such differences become ‘‘stiff’’ edge also in the i - direction. These displacements are
substantial in the beams, where again the demands at the obtained by the equivalent static method for the seismic
‘‘flexible’’ edges exceed substantially the demands at the ‘‘stiff’’ combinations considered. The factors are ratios of the top story
edges, as it happens with the one and three story buildings and displacements at the ‘‘flexible’’ and ‘‘stiff’’ edges in a given
also with the torsionally stiff ones [16]. direction (x or y), to their mean values. The design modification
that was subsequently applied was to multiply the axial areas
of the bracing members in both the ‘‘stiff’’ and ‘‘flexible’’ edges
by the corresponding factors in each direction and to do the
same for the beam and column sections but only in the
5. Modification procedure ‘‘flexible’’ edges to increase both stiffness and strength of the
corresponding frames. The cross sections of columns and beams
A structural design can be characterized as satisfactory when of the ‘‘stiff’’ edges are not reduced, as their strength is
the limiting values of the controlling response parameters do not controlled mainly by gravity loads (loading combinations with-
have wide variations within the groups of structural members to out earthquake). These factors vary from 1.25–1.50 for the
which they apply. In the opposite case, suboptimal use of material ‘‘flexible’’ edges and from 0.85 to 1.00 for the ‘‘stiff’’ edges. After
may be present as well as a potentially higher risk of failure in this modification, each structure was checked again for full
cases of unexpected overloads. Thus the observed substantial
differences in ductility demands between the opposite edges of
the examined buildings, especially in the beams, point to the need
Table 4
for a design modification that would eliminate or reduce these Edge displacements for the one-story frames (in m). Initial and modified design.
differences. Such a modification is implemented in the present
paper on the basis of results obtained from the elastic analyses of Eccentricity Initial design Modified design
the buildings. It aims at increasing the strength of the structural Direction X
members (columns, beams and braces) at the ‘‘flexible’’ edges and
Frame–X01 Frame–X04 Frame–X01 Frame–X04
reducing the strength of the braces at the ‘‘stiff’’ edges without
affecting the strength of the other structural elements (columns, e ¼0.10 0.02656 0.02441 0.02096 0.01908
beams). A very similar modification proved quite effective for e ¼0.20 0.03201 0.02940 0.01947 0.01999
torsionally stiff buildings [15,17]. Direction Y
The first step for application of this modification is to obtain
the top story displacements at the ‘‘flexible’’ and ‘‘stiff’’ edges of Frame–Y01 Frame–Y04 Frame–Y01 Frame–Y04
e ¼0.10 0.02631 0.04496 0.02501 0.03749
the buildings in both horizontal directions due to the earthquake
e ¼0.20 0.02686 0.05109 0.02561 0.03107
loading considered and then compute the following factors in

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

AXIAL STRAIN DUCTILITY IN BRACES
symm FR-X02(flex) symm FR-Y03(flex)
FR-X03(stiff) FR-Y02(stiff)
5 5

4 4
Story

Story

3 3

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
symm FR-X01(flex) symm FR-Y04(flex)
FR-X04(stiff) FR-Y01(stiff)
5 5

4 4
Story

Story

3 3

2 2

1 1
1 2 3 4 1 2 3 4
Rotational ductility factor Rotational ductility factor

Fig. 12. Member ductility demands of 5-story building with em ¼ 0.20 and comparison with the symmetric building (FR-X01,X02, & FR-Y03,Y04: ‘‘flexible’’ sides, FR-
X03,X04 & FR-Y01,Y02: ‘‘stiff’’ sides).
152 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

Table 5 compliance with the applicable codes. The new, modified

Ductility demands for braces and beams in the 1-story frames. Initial and modified structures were again subjected to the same two component
design.
motion earthquake set and their responses were again com-
Eccentricity Initial design Modified design puted as before.
Direction X–braces Results for the modified one-story frames are summarized in
Frame–X02 Frame–X03 Frame–X02 Frame–X03 Tables 4 and 5. In the results presented in Table 4, we can see
that the differences in displacements between ‘‘flexible’’ and
e ¼0.10 2.12 1.85 1.81 1.67
e ¼0.20 2.10 1.76 1.40 1.43
‘‘stiff’’ edges have been significantly reduced. The same hap-
Direction Y–braces pens with the ductility demands of the braces and the beams as
Frame–Y02 Frame–Y03 Frame–Y02 Frame–Y03 the difference in the values of demands has decreased. Figs.
e ¼0.10 2.73 3.62 2.71 3.32 13–16 show displacement demands for the 3 and 5 story
e ¼0.20 2.36 3.59 2.09 2.41
frames for the initial and the modified designs. Figs. 17–20
Initial design Modified design show ductility demands for the 3 and 5 story frames for the
Eccentricity Direction X–beams initial and the modified designs. If we compare the results
Frame–X01 Frame–X04 Frame–X01 Frame–X04
e ¼0.10 1.76 1.26 1.23 1.09
obtained from the modified designs with that of the original
e ¼0.20 1.49 1.26 1.11 1.02 design, we see a substantial improvement of response in all
Direction Y-beams cases: the overall maximum ductility demand factor in each
Frame–Y01 Frame–Y04 Frame–Y01 Frame–Y04 group is reduced and so are the differences between ‘‘flexible’’
e ¼0.10 1.08 2.16 1.04 1.72
and ‘‘stiff’’ edges, producing more uniform distribution of such
e ¼0.20 1.00 1.31 1.00 1.32
demands.

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

TOTAL DISPLACEMENTS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3

Story
Story

2 2

1 1
0.00 0.05 0.10 0.00 0.05 0.10
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
Flex(mod) Stiff(mod) Flex(mod) Stiff(mod)
Flex(init) Stiff(init) Flex(init) Stiff(init)
3 3

2 2
Story
Story

1 1

0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04
Relative displacement (m) Relative displacement (m)

Fig. 13. Total displacements and interstory drifts of 3-story modified building with em ¼ 0.10 and comparison with the initial building.

DIRECTION X, εm=0.20 DI RECT I ON Y , ε m=0.20

TOTAL DISPLACEMENTS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3
Story

Story

2 2

1 1
0.00 0.05 0.10 0.00 0.05 0.10
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
Flex(mod) Stiff(mod) Flex(mod) Stiff(mod)
Flex(init) Stiff(init) Flex(init) Stiff(init)
3 3

2 2
Story
Story

1 1

0 0
0.01 0.02 0.03 0.04 0.05 0.01 0.02 0.03 0.04 0.05
Relative displacement (m) Relative displacement (m)

Fig. 14. Total displacements and interstory drifts of 3-story modified building with em ¼ 0.20 and comparison with the initial building.
 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156 153

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

TOTAL DISPLACEMENTS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4

Story
Story
3 3

2 2

1 1
0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
Flex(mod) Stiff(mod) Flex(mod) Stiff(mod)
Flex(init) Stiff(init) Flex(init) Stiff(init)
5 5

4 4

3 3

Story
Story

2 2
1 1
0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04 0.05
Relative displacement (m) Relative displacement (m)

Fig. 15. Total displacements and interstory drifts of 5-story modified building with em ¼ 0.10 and comparison with the initial building.

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

TOTAL DISPLACEMENTS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4
Story

Story

3 3

2 2

1 1
0.00 0.05 0.10 0.15 0.00 0.10 0.20
Total displacement (m) Total displacement (m)
INTERSTORY DISPLACEMENTS
Flex(mod) Stiff(mod) Flex(mod) Stiff(mod)
Flex(init) Stiff(init) Flex(init) Stiff(init)
5 5

4 4

3 3
Story
Story

2 2

1 1

0 0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04 0.05 0.06
Relative displacement (m) Relative displacement (m)

Fig. 16. Total displacements and interstory drifts of 5-story modified building with em ¼ 0.20 and comparison with the initial building.

6. Some comments on the proposed modification
To get more insight about the consequences of the proposed
modification, the new stiffness centers of the modified buildings
were computed using eqs. 1 and 2 and compared with those of
the original designs. Results for the floors of the 3-story building
are presented in Fig. 21. We can see that the proposed modifica-
tion brings the approximate stiffness center of each story sub-
stantially closer to the mass center and thus the torsional motions
are substantially reduced. This reduction is obviously greater for
the ‘‘flexible’’ edge and hence the reduction in the observed
differences of ductility demands between ‘‘flexible’’ and ‘‘stiff’’
edges. We must note here that bringing the stiffness center as
close as possible to the mass center, in other words trying to
minimize the physical eccentricity, is a well known design
objective in earthquake engineering as it minimizes torsional
motion. The proposed modification is thus a ‘‘blind’’ way of
achieving this without significant extra effort.
7. Conclusions
In the present work the earthquake response of eccentric,
torsionally flexible, steel, braced frame buildings designed in
accordance with Eurocodes EC3 and EC8 was examined and
similar, overall, results were obtained compared to earlier find-
ings for torsionally stiff steel and reinforced concrete frame
buildings. More specifically, it was found that under the action
154 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156

DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10

AXIAL STRAIN DUCTILITY IN BRACES
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3

Story

Story
2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3
Story

Story
2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 17. Member ductility demands of 3-story modified building with em ¼0.10 and comparison with the initial building.

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

AXIAL STRAIN DUCTILITY IN BRACES
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3
Story

Story

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
3 3
Story

Story

2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 18. Member ductility demands of 3-story modified building with em ¼0.20 and comparison with the initial building.

of two horizontal component earthquake loadings, compatible
with the design spectra, ductility demands at the ‘‘flexible’’ edges
were significantly greater than ductility demands at the
‘‘stiff’’ edges.
Subsequently the original design was modified by changing
the diagonal braces so that those on the ‘‘flexible’’ sides were
made stronger while those on the ‘‘stiff’’ sides were made
weaker. In addition, beams and columns at the ‘‘flexible’’ edge
were strengthened, while the members at the ‘‘stiff’’ edge
were left unchanged. For a given frame, the modification
factor, constant for all stories in the same edge, is the ratio
of the respective top story edge displacement, to the mean
value of this and the displacement of the opposite edge. By
computing again the inelastic response of the modified
designs for the same groups of motions used with the original
buildings, it was found that the response of the new designs
was improved: ductility demands at the ‘‘flexible’’ edges gen-
erally decreased and the differences between the two sides
diminished, so that a more uniform distribution of ductility
demands was achieved.
On the basis of these findings, a code modification may appear
desirable. However, additional studies covering other types of
irregular buildings and a wider spectrum of parameters will be
required, before any firm recommendation is put forward.
 DIRECTION X, εm=0.10 DI RECT I ON Y , εm=0.10
AXIAL STRAIN DUCTILITY IN BRACES
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4

Story
Story
3 3

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4

Story
Story

3 3

2 2

1 1
1 2 3 1 2 3
Rotational ductility factor Rotational ductility factor

Fig. 19. Member ductility demands of 5-story modified building with em ¼0.10 and comparison with the initial building.

DIRECTION X, εm=0.20 DI RECT I ON Y , εm=0.20

AXIAL STRAIN DUCTILITY IN BRACES
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4
Story
Story

3 3

2 2

1 1
1 2 3 4 1 2 3 4
Axial strain ductility factor Axial strain ductility factor
ROTATIONAL DUCTILITY FACTOR IN BEAMS
Flex-mod Stiff-mod Flex-mod Stiff-mod
Flex-init Stiff-init Flex-init Stiff-init
5 5

4 4
Story

Story

3 3

2 2

1 1
1 2 3 4 1 2 3 4
Rotational ductility factor Rotational ductility factor

Fig. 20. Member ductility demands of 5-story modified building with em ¼0.20 and comparison with the initial building.

Fig. 21. Approximate stiffness centers and floor eccentricities of the original and modified 3-story with em ¼ 0.20.
156 M.T. Kyrkos, S.A. Anagnostopoulos / Soil Dynamics and Earthquake Engineering 47 (2013) 144–156
Acknowledgments
The authors would like to thank Mr. Nikos Chroneas, for
making available to them the commercial program NEXT used
in the design of all the buildings.
References
[1] Chopra AK, Goel RK. Evaluation of torsional provisions in seismic codes.
Journal of the Structural Division, (ASCE) 1991;117(12):3762–82.
[2] De La Colina J. Assessment of design recommendations for torsionally
unbalanced multistory buildings. Earthquake Spectra 2003;19(1):47–66.
[3] Stathopoulos, KG, Anagnostopoulos, SA. Inelastic earthquake response of
buildings subjected to torsion. In: 12th World conference on earthquake
engineering id.0781; 2000.
[4] Stathopoulos KG, Anagnostopoulos SA. Inelastic earthquake response of
single-story asymmetric buildings: an assessment of simplified shear-beam
models. Earthquake Engineering and Structural Dynamics 2003;32:1813–31.
[5] Stathopoulos KG, Anagnostopoulos SA. Inelastic torsion of multistory build-
ings under earthquake excitations. Earthquake Engineering and Structural
Dynamics 2005;34:1449–65.
[6] Rutenberg A. EAEE Task Group (TG)8: Behavior of irregular and complex
structures—state of the art report. Seismic nonlinear response of code-
designed asymmetric structures. In: 11th European Conference on Earth-
quake Engineering; 1998.
[7] Rutenberg A. EAEE Task Group (TG)8: Behavior of irregular and complex
structures—asymmetric structures progress since 1998. In: 12th European
Conference on Earthquake Engineering; 2002.
[8] De Stefano M.& Pintucchi (2008).A review of research on seismic behavior of
irregular building structures since 2002Bulletin of Earthquake Engineering
6:285-308.
[9] Anagnostopoulos SA, Alexopoulou Ch, Stathopoulos K. An answer to an
important controversy and the need for caution when using simple models
to predict inelastic earthquake response of buildings with torsion. Earth-
quake Engineering and Structural Dynamics 2010;39:521–40.
[10] Chandler AM. Building damage in Mexico City earthquake. Nature
1986;320:497–501.
[11] Chandler AM, Duan XN. Evaluation of factorsinfluencing the inelastic seismic
preformance of torsionally asymmetrical buildings. Earthquake Engineering
and Structural Dynamics 1991;20:87–95.
[12] Tso, WK, Zhu, TJ. Strength Distribution for Torsionally Unbalanced Structures.
In: 10th World Conference on Earthquake Engineering, vol. 7; 1992. p. 3865–
3868.
[13] Humar, JL, Kumar, P. A new Look at the Tortion Design Provisions in Seismic
Building Codes. In: 12th World Conference on Earthquake Engineering
id.1707; 1998.
[14] Ghersi A, Marino E, Rossi PP. Inelastic response of multi-storey asymmetric
buildings. In: Proceedings of the 12th World Conference on earthquake
Engineering, no 1716; 2000.
[15] Kyrkos MT, Anagnostopoulos SA. Towards earthquake resistant design of
steel buildings for uniform ductility demands. Advances and trends in
structural Engineering, Mechanics and Computation. In: Zingoni (Ed), Cape-
town: SEMC 2010; 2010.
[16] Kyrkos MT, Anagnostopoulos SA. An assessment of code designed torsionally
stiff asymmetric steel buildings under strong earthquake excitations. Earth-
quakes and Structures 2011;2:109–26.
[17] Kyrkos MT, Anagnostopoulos SA. Improved earthquake resistant design of
torsionally stiff asymmetric steel buildings. Earthquakes and Structures
2011;2:127–47.
[18] Kyrkos MT. (2011). Improved earthquake resistant design of eccentric steel
buildings with diagonal braces, PhD thesis. University of Patras (in Greek).
[19] Karabalis DL, Cokkinides GJ, Rizos DC, Mulliken JS. Simulation of earthquake
ground motions by a deterministic approach. Advances in Engineering
Software 2000;31:329–38.
[20] Carr AJ, (2005). Ruaumoko Manual. Volume 3. User manual for the 3-dimen-
sional Version-Ruaumoko 3D.
[21] Anagnostopoulos SA. Inelastic Beams for Seismic Analyses of Structures.
Journal of Structural Division, ASCE 1981;107(7):1297–311.