BEHAVIOUR AND DESIGN OF SET-BACK STEEL FRAMES

Federico M. Mazzolani1, Vincenzo Piluso2

Abstract

The presence of set-backs is usually considered a geometrical feature of

buildings which can lead to the worsening of the seismic inelastic response with
respect that experienced by geometrically regular frames. In this paper, a
sophisticated design procedure aiming at the control of the failure mode is
presented. This procedure is the extension, to the case of geometrically irregular
frames, of a new approach, based on second order rigid-plastic analysis, which
has been successfully applied in the case of regular steel frames.
The main items of the proposed design procedure are briefly recalled, while the
attention is focused on the solution of the additional difficulties arising in the
case of set-back steel frames.
The reliability of the proposed design procedure is evidenced through its
application to 14 set-back steel frames, all derived from the same six bay-five
storey regular frame. The occurrence of a pattern of yielding according to the
global mechanism is verified by means of static push-over inelastic analyses. The
obtained results show that the proposed design procedure leads not only to the
control of the failure mode, but also to the control of the local ductility demand.
Finally, the inelastic behaviour of the designed frames is compared to that of the
regular frame from which they have been derived. This comparison points out
that set-back frames cannot be considered as irregular, provided that they are
properly designed.

Introduction
Despite the physical meaning of the structural regularity concept is quite intuitive, its
quantitative definition is particularly difficult. In fact, there is a great number of parameters
which influence the energy dissipation mechanism and, therefore, the structural regularity.
Moreover, for any given degree of regularity the failure mode is significantly affected by the
adopted design criteria.

1 Department of Structural Analysis and Design, University of Naples, Italy

2 Department of Civil Engineering, University of Salerno, Italy

279
 The term plan regularity is referred to the ability of a building to vibrate separately in
two vertical planes without torsional coupling. This feature is considered favourable because it
provides both a reduction of the damage concentration in the peripheral elements and it allows
the use of simplified assumptions in structural modelling and analysis. This paper is devoted to
the plastic design of buildings fulfilling the above requirement.
However, also the vertical configuration of buildings can significantly affect the seismic
inelastic behaviour. The term vertical regularity is used with reference to those vertical
configurations for which damage concentrations are not expected.
With reference to the volume of the building, it is generally accepted that closed and
compact shapes are the most favourable ones. Therefore, under this point of view, the ideal
configuration is obtained when the building has the shape of a parallelepiped. Irregular
configurations are characterized by the variation of the boundary lines along the height giving
rise to set-backs or off-sets; they are considered irregular because the corresponding
distributions of mass and stiffness are not uniform. Moreover, even if the boundary lines of a
building are regular, it has to be considered as irregular, when the vertical distribution of the
stiffness is characterized by abrupt variations. All these structural situations can lead to an
undesirable concentration of the earthquake input energy.
Therefore, from the practical point of view, a very important aspect is represented by
the need to provide design criteria able to avoid such damage concentrations and, therefore, a
poor seismic inelastic behaviour of geometrically irregular structures which can be more prone
to local failure modes.
With reference to structural schemes characterized by the presence of set-backs,
commonly considered as irregular, this paper presents a new design approach able to assure,
also in these structural situations, the development of a collapse mechanism of global type.
The concept of a desirable member hierarchy in the energy dissipating mechanisms, to
be exploited in ductile moment resisting multistorey frames during severe earthquakes, is
universally recognized [1-6]. In particular, it is usually required that plastic hinges develop in
beams rather than in columns. The aim of this design goal is to avoid collapse mechanisms
characterized by poor energy dissipation capacity, such as "soft-storey" mechanisms. In
particular, irregular frames are more prone to these undesired failure modes, so that the
adoption of design criteria aiming at the control of the collapse mechanism are even more
important than in the case of regular frames.
Significant research efforts have been spent in order to establish simple design criteria
to avoid undesired collapse mechanisms which undermine the global ductility supply and
energy dissipation capacity of structures. Unfortunately, these are empirical rules rather than
theoretical approaches so that the common design practice still requires the use of engineering
judgement in quantifying the member hierarchy to be applied to control the failure mode.
The primary aim of the capacity design of columns is to eliminate the likelihood of the
simultaneous formation of plastic hinges at both ends of all columns of a storey [4]. This
design goal is obtained by imposing that the sum of the flexural strength of the columns
connected to any joint is greater than the sum of the flexural strength of the beams connected
to the same joint. Even though with some small differences depending also on the
contructional material, this member hierarchy criterion is suggested in all modern codes [7-10].
With reference to moment resisting steel frames, this simple design criterion can be
sufficient to ensure that a "soft-storey" will not develop, but it does not lead to frames failing
in a global mode [11,12]. For this reason, a more sophisticated design procedure, assuring the
development of a collapse mechanism of global type has been recently proposed with reference

280
to regular steel frames. The proposed design procedure, based on the kinematic theorem of
plastic collapse and on second order plastic analysis, covers an important gap in the design
tools for seismic resistant steel frames [13].
The plastic methods of structural analysis can be particularly useful to evaluate the
inelastic behaviour of seismic resistant frames provided that a suitable distribution of horizontal
forces, increasing according to a unique multiplier, is selected [14,15]. In addition, the use of
second order plastic analysis, i.e. of mechanism equilibrium curves, has been recently
suggested as an effective method to include P- effects into the design process of seismic
resistant steel frames [16].
The structural design oriented to the failure mode control is a relatively recent problem
arisen from seismic design needs, which up-to-now has been mainly faced through simplified
rules provided in seismic codes. As already stated, modern seismic codes [7-10] require only

Fns Fns

Fk Fk hns

im him
F2 F2

h2
F1 F1

h1

GLOBAL MECHANISM TYPE-1 MECHANISM

Fns Fns

Fk Fk hns

im im hi m
F2 F2

h2
F1 F1

h1

TYPE-2 MECHANISM TYPE-3 MECHANISM

Figure 1 - Examined collapse mechanism typologies

281
the fulfilment of the member hierarchy criterion which is sufficient to avoid storey
mechanisms, but it does not allow the complete development of a global mechanism.
Starting from this consideration, the proposed design procedure aims at the control of
the failure mode of seismic resistant steel frames. It is based on the observation [14,15] that the
collapse mechanisms of frames under horizontal forces can be considered belonging to three
main typologies (Fig.1). The collapse mechanism of the global type is a particular case of
type-2 mechanism. The control of the failure mode can be performed through the analysis of
3ns mechanisms (where ns is the number of storeys) [17]. It is assumed that the beam sections
are already designed to resist vertical loads. Therefore, the values of the plastic section
modulus of columns only have to be defined so that the kinematically admissible multiplier of
the horizontal forces corresponding to the global mechanism is less than those corresponding to
the other 3ns-1 kinematically admissibile mechanisms. It means that, according to the upper
bound theorem, the above stated multiplier is the true collapse multiplier and, therefore, the
true collapse mechanism is the global failure mode.
From the theoretical point of view, a significant improvement can be obtained by
accounting for second order effects by a procedure based on the equilibrium curves of the
considered mechanisms, without any significant increase of the computational effort.
In this paper, the proposed procedure is extended to the case of geometrically irregular
frames. Therefore, only the main items of the design approach will be briefly recalled, while
the attention will be focused on the solution of the additional difficulties arising in the case of
set-back steel frames.
Starting from a six bay-five storey regular frame, 14 set-back steel frames are derived.
All frames are designed with the proposed procedure for failure mode control and the
occurrence of a pattern of yielding according to the global mechanism is verified by means of
static push-over inelastic analyses, showing the reliability of the proposed design procedure. In
addition, the obtained results show that not only the control of the failure mode, but also the
control of the local ductility demand is obtained with the suggested design approach.
Finally, the comparison between the seismic inelastic performance of the designed
frames and that of the regular frame from which they have been derived is presented in order
to underline that properly designed set-back frames cannot be considered as irregular.

Proposed design procedure

The seismic action is modelled through a system of horizontal forces whose distribution
can be selected according to a proper combination of the eigenmodes. The magnitude of these
horizontal forces is governed by the multiplier , while vertical loads are assumed to be
constant. In addition, it is assumed that the beams are already designed to resist vertical loads,
so that the unknowns of the design problem are the column sections.
In order to design frames failing in a global mode, the cross-sections of columns have
to be dimensioned so that, according to the upper bound theorem, the kinematically admissible
horizontal force multiplier corresponding to the global type mechanism is the minimum among
all kinematically admissible multipliers.
This condition is sufficient to assure the desired collapse mechanism provided that the
structural material behaves as rigid-plastic so that the horizontal displacements are equal to
zero up to the complete development of the collapse mechanism. On the contrary, the actual
behaviour is elasto-plastic with significant displacements before the complete development of

282

(t)
im
(t)
i generic mechanism
m

(g)

(g)

global mechanism

u
Figure 2 - Design conditions

the collapse mechanism. These displacements give rise to second order effects which cannot be
neglected in the design process.
From the practical point of view, the influence of second order effects can be taken into
account by imposing that the mechanism equilibrium curve corresponding to the global
mechanism has to lie below those corresponding to all other mechanisms. However, the
fulfilment of this requirement is necessary only up to a selected ultimate displacement u which
has to be compatible with the plastic rotation capacity of members (Fig.2).
Therefore, the following design conditions have to be imposed [13,16]:
(g) (g) (t) (t)
u im im u im = 1,2,3,.....,ns t = 1,2,3 (1)
where:
(g) and (g) are, respectively, the kinematically admissible multiplier of the horizontal forces
(rigid-plastic theory) and the slope of the softening branch of the curve, corresponding
to the global type mechanism;
(imt) and (imt) have the same meaning of the previous symbols, but they are referred to the imth
mechanism of the tth typology (t=1,2,3).
A very important point of the design procedure for failure mode control is the selection
of the maximum top sway displacement up to which it is desired to assure that the collapse
mechanism cannot be different from the global one. In fact, the value of this displacement
governs the magnitude of second order effects accounted for into the design procedure. In
addition, the complete development of the collapse mechanism could be prevented by the
occurrence of plastic rotation demands exceeding the local ductility supply. Therefore, a good
criterion to choose the design ultimate displacement u is to relate it to the plastic rotation
supply of beams or beam-to-column connections by assuming u = p H (where p is the plastic
rotation supply).

283
 Concerning the plastic rotation supply of steel moment connections, it is traditionally
measured by cyclic moment rotation tests. In the past, some researchers proposed that if a
connection can reach a plastic rotation of 0.02 radian under cyclic loading, the connection can
be considered sufficiently ductile to be used for seismic resistant frames [18]. Other researchers
have established that, in the case of severe earthquakes, the required plastic rotation can reach
more than 0.03 radian [19-21].
This topic is particularly important, as confirmed by the recent earthquakes of
Northridge (January 17, 1994) and Hyogoken-Nambu (Kobe, 17 January, 1995), and deserves
further investigations; however, for the purpose of failure mode control, the value 0.04 radian
can be suggested aiming to assure an increased safety level against the influence of second
order effects under seismic loads [13].
In order to transform the design goal expressed by equation (1) into design conditions
which the column sections have to satisfy to assure the failure mode control, the first step is
the evaluation of the parameters defining the equilibrium curve of all the possible mechanisms.
With reference to the generic mechanism, the kinematically admissible multiplier of
horizontal forces can be expressed through the following relationship:
[ tr ( C R c ) + 2 tr ( B R b) tr ( q Dv ) ]
T T T (2)
=
FT s
where:
C is the matrix of order nc ns (being nc the number of columns and ns the number of
storeys) whose elements are equal to the column plastic moments reduced for the presence
of axial forces;
R c is the matrix (order nc ns) of the coefficients R c.ik accounting for the participation of the
ith column of the k th storey to the collapse mechanism, being R c.ik = 2 when the column is
yielded at both ends, R c.ik = 1 when only one column end is yielded and R c.ik = 0 when the
column does not participate to the collapse mechanism;
B is a matrix of order nb ns (being nb the number of bays) whose elements are equal to the
plastic moments of beams;
R b is the matrix (order nb ns) of the plastic rotations developed by the beams for a unit
virtual plastic rotation of column hinges;
q is the matrix (order nb ns) of the uniform loads acting on the beams;
Dv is a matrix (order nb ns) whose elements Dv.jk represent the area of the first order
vertical displacement diagram of the jth beam of the k th storey for a unit virtual plastic
rotation of column hinges;
F is the vector of the design horizontal forces;
s is the shape vector of the storey horizontal virtual displacements;
tr denotes the trace of the matrix.
In addition, the slope of the linearized mechanism equilibrium curve is given by [13]:
1
VT s (3)
Ho
=
FT s
where:
V is the vector of the resultant vertical load acting at the different storeys;
Ho is the sum of the interstorey heights of the storeys involved by the generic mechanism.

284
 The design requirements given by equation (1) provide the limitations which the column
plastic moments have to satisfy. These design conditions are expressed through the parameters:
nc

M c.iim
M Tc.i I i=1 (4)
im = T m = nc
M c1 I
M c,i1
i=1

where im is the mechanism index and M c.ik is the plastic moment, reduced for the presence of
the axial internal force, of the ith column of the k th storey. Therefore, the parameter im repre-
sents the ratio between the sum of the reduced plastic moments of the imth storey columns and
the same sum corresponding to the first storey columns. All design conditions will be ex-
pressed by means of these ratios.
By denoting with (im1) the values of the ratios (4) which have to be assured to prevent
failure according to type-1 collapse mechanisms, the imth condition to be satisfied to avoid
these collapse mechanisms (equation (1) for t = 1) can be written in the following form [13]:
nc
1 im (g) (im1)
1
im M c.i1 + 1
im
+ (g)


(1)
i 1

u + (g)
1
im
m
(1) i=1 (5)
im nc
1
im M c.i1
i=1

which has to be applied for im=1,2,3,.....ns.

For a unit virtual plastic rotation of column hinges and with reference to the global
mechanism, the parameter (g) represents the internal work developed by beams, the parameter
(g) is related to the second order work due to vertical loads while the parameter (g) represents
the external work due to the uniform vertical loads acting on the beams.
The function im represents the ratio between the internal work which the beams develop
in the imth mechanism of type-1 and that developed in the global mechanism. The function im
represents the ratio between the external work which the horizontal forces develop in the imth
mechanism of type-1 and that developed in the global mechanism. In addition, the function (im1)
represents the ratio between the external work which the uniform vertical loads develop in the
imth mechanism of type-1 and that developed in the global mechanism. Finally, the parameter
(im1) represents the ratio between the slope of the equilibrium curve of the imth mechanism of
type-1 and that of the global mechanism.
All these parameters are known quantities, because both the beam plastic moments and
the magnitude of both vertical loads and design seismic forces are data of the design problem.
By denoting with (im2) the values of the ratios (4) corresponding to the fulfilment of the
ns design conditions to avoid collapse mechanisms of type-2 (equation (1) for t = 2), the
following relationship is obtained [13]:

285
 nc
(g) im (g) (im2)
Mc.i1 + 1
im
+
im
1 + (g) (im2) 1 u

(2) i=1 (6)
im nc
1
im M c.i1
i=1

which has to be applied for im=1,2,3,.....ns.

where the function im represents the ratio between the internal work done by the beams in the
imth mechanism of type-2 and that done in the global mechanism; the function im represents the
ratio between the external work due to the horizontal forces in the imth mechanism of type-2
and that in the global mechanism. In addition, (im2) has the same meaning, but with reference to
the external work due to the uniform vertical loads. Finally, (im2) represents the ratio between
the slope of the equilibrium curve of the imth mechanism of type-2 and that of the global
mechanism.
All these functions are still known quantities, because the plastic moments of beams, the
horizontal forces and the uniform vertical loads are input data of the design problem.
Finally, denoting with (im3) the values of the ratios (4) corresponding to the fulfilment of
the ns design conditions to avoid collapse mechanisms of type-3, the following relationship is
obtained [13]:
nc

Mc.i1 + (g) (g) + (g)(im3) 1 u

(3)

i=1 (7)
im nc
2
im M c.i1
i=1

which has to be applied for im=1,2,3,.....ns.

In relationship (7) im represents the ratio between the external work developed by the
horizontal forces in the imth mechanism of type-3 and that done in the global mechanism, while
the parameter (im3) represents the ratio between the slope of the equilibrium curve of the imth
mechanism of type-3 and the one of the global mechanism.
The above design conditions have been derived directly from the extension of the upper
bound theorem to the mechanism equilibrium curves [13]. In particular, for each storey there
are three design conditions to be satisfied, because three collapse mechanism typologies have
been considered. As these design conditions have to be contemporaneously satisfied for each
storey, the ratios (4) between the sum of the reduced plastic moments of columns of the imth
storey and the same sum corresponding to the first storey columns allow to satisfy all the
design requirements provided that the following relationship is verified:
(1) (2) (3)
im = max im , im , im

(8)

As the section of columns can only decrease along the height of the frame, the values of
im (with im=1,2,....,ns) obtained from the above conditions have to be modified in order to
satisfy the following technological limitation:
1 2 3 ........ ns (9)

286
 The evaluation of the parameters (4) satisfying all design requirements, given by
relationships (5), (6), (7) and (9), can be performed by means of the algorithm described in
[13,16], providing also the value of the sum of the first storey column design moments.
In the case of geometrically regular frames, it was proposed to derive the design value
of the bending moment of the single columns by assuming that, at any given storey, the storey
nc

bending moment (k M c.i1) is subdivided in equal parts among the storey columns [13,16].
i=1
An additional difficulty arises in the case of geometrically irregular frames, because the
number of columns varies along the frame height so that the above approach cannot be
applied. In fact, in the case of geometrically irregular frames, at any given storey, the column
axial forces can be significantly different each other. For this reason, in the case of such
structural schemes, it can be suggested to distribute, at any given storey, the sum of column
plastic moments proportionally to the corresponding axial forces. In reality, the repartition of
moments is governed by the column inertia, but these data are not available a priori as the
column sections are the unknowns of the problem. Therefore, it is necessary to use the existing
data, i.e. the axial forces corresponding to the complete development of the global mechanism.
According to this criterion, the design value of the bending moment of each column is
computed through the following relationship:
nc (10)
N ik
M c.ik = k M c.i1 nc
i=1
N ik
i=1

where N ik is the axial force in the ith column of the k th storey.

In order to evidence the reliability of the proposed design procedure also in the case of
geometrically irregular frames, a wide number of set-back steel frames has been designed. The
inelastic behaviour of such frames will be presented in the following section.

DESIGN OF SET-BACK FRAMES

The presence of set-backs is commonly considered a geometrical property of structures
leading to a worsening of the seismic inelastic behaviour with respect to the case of regular
frames. In addition, set-back frames are more prone to local failure modes. For this reason, the
structural design oriented to the failure mode control is, for such structural schemes, even more
important than in the case of regular frames.
It is worthwhile mentioning that a very important feature of the proposed design
procedure, which has been briefly described in the previous section, is its ability to account for
any structural configuration, including the presence of set-backs.
The application of the proposed design procedure to set-back steel frames will be
presented in this section with reference to all the structural schemes which can be derived from
the same six bay-five storey regular frame. However, only symmetrical schemes will be
considered in order to exclude any interaction with the torsional inelastic response arising in
the case of plane irregular structures. Therefore, starting from the same regular frame (FRAME
1), 14 set-back frames are obtained (Fig.3).

287
 FRAME 1 FRAME 2

FRAME 3 FRAME 4

FRAME 5

FRAME 6 FRAME 7

FRAME 8 FRAME 9

FRAME 10

FRAME 11 FRAME 12

FRAME 13 FRAME 14

FRAME 15

Figure 3 - Analysed frames and corresponding pattern of yielding for < 64 cm

288
 The bay span of the analysed frames is equal to 6.50 m; the interstorey height is equal
to 3.20 m. The characteristic value of the uniform vertical load acting on the beams is equal to
15 kN/m (Gk) and 10 kN/m (Qk) for dead and live loads, respectively. The beams have been
designed considering the load combination 1.3 Gk + 1.5 Qk. All the beams have IPE330 shape
made of Fe430 steel.
The design values of the seismic horizontal forces have been computed considering the
design spectrum given by Eurocode 8 for stiff soil conditions with a peak ground acceleration
equal to 0.35 g and by assuming a value of the q-factor equal to 6. The period of vibration has
been computed through the following relationship:
T = 0.085 H3 4 (11)

suggested by Eurocode 8 for preliminary design [7], where H is the overall height of the
building.
For each frame, the design base shear has been distributed along the building height by
means of the classical formula for simplified modal analyses [7].
Starting from the above defined distribution of the seismic horizontal forces and by
considering the vertical loads corresponding to the load combination Gk + 0.3 Qk + E (where E
represents the design seismic action), the column sections have been designed by means of the
procedure briefly described in the previous section. In particular, the design value of the top
sway plastic displacement has been choosen by assuming as limit value a plastic rotation
supply of members and/or connections equal to 0.04 rad, so that the corresponding top sway
plastic displacement is equal to 0.64 m. As a result of the proposed design approach, the
column sections given in table 1 have been obtained. Also the columns are made of Fe430
steel.

INELASTIC BEHAVIOUR OF SET-BACK FRAMES

In order to evidence the reliability of the proposed design procedure for failure mode
control, the inelastic behaviour of the designed frames has been investigated by means of
push-over static inelastic analyses, including both geometrical and mechanical non-linearities.
The main result of these inelastic analyses is the pattern of yielding developed before
the occurrence of a displacement exceeding the design value of the top sway displacement
(0.64 m). These patterns of yielding are represented in Fig.3 for all the analysed frames. From
this figure, it can be observed that plastic hinges develop according to the global failure mode.
However, some column hinging occurs, but the corresponding plastic hinges are not able to
undermine the feature of the failure mode which is basically a collapse mechanism of global
type. This result shows the reliability of the proposed design procedure also in the case of
set-back frames. In fact, the only exception is represented by FRAME 15 which is particularly
prone to partial mechanisms. However, the development of the partial mechanism involving
the tower above the large base characterizing this frame practically occurs when the
design value of the ultimate top sway displacement is reached. It means that the mechanism
equilibrium curve corresponding to the mechanism shown in Fig.3 practically intersects that
corresponding to the global mechanism when the design top sway displacement is reached.
The seismic inelastic performances of the designed frames have been furtherly
examined by computing the q-factor through an energy approach based on push-over static
inelastic analysis [14,22], leading to the following relationship:

289
 Table 1 - Member sections of designed frames

FRAME COLUMN STOREY 1 STOREY 2 STOREY 3 STOREY 4 STOREY 5

A HEB400 HEB340 HEB340 HEB320 HEB280
1 B HEB360 HEB340 HEB320 HEB320 HEB280
C HEB360 HEB340 HEB320 HEB320 HEB280
D HEB360 HEB340 HEB320 HEB320 HEB280
A HEB340 HEB300 HEB280 HEB260
2 B HEB400 HEB340 HEB340 HEB320 HEB280
C HEB360 HEB340 HEB340 HEB320 HEB260
D HEB360 HEB340 HEB340 HEB320 HEB260
A HEB340 HEB300 HEB300 HEB260
3 B HEB340 HEB300 HEB300 HEB260
C HEB400 HEB360 HEB360 HEB340 HEB260
D HEB400 HEB340 HEB340 HEB340 HEB260
A HEB280 HEB240 HEB220
4 B HEB360 HEB320 HEB320 HEB320 HEB280
C HEB360 HEB320 HEB320 HEB300 HEB280
D HEB360 HEB320 HEB320 HEB300 HEB280
A HEB300 HEB260 HEB220
5 B HEB320 HEB300 HEB300 HEB240
C HEB360 HEB340 HEB340 HEB320 HEB260
D HEB360 HEB340 HEB340 HEB320 HEB260
A HEB280 HEB240 HEB220
6 B HEB280 HEB240 HEB220
C HEB360 HEB340 HEB340 HEB280 HEB260
D HEB360 HEB340 HEB340 HEB280 HEB260
A HEB240 HEB180
7 B HEB340 HEB320 HEB320 HEB320 HEB280
C HEB340 HEB300 HEB300 HEB300 HEB260
D HEB340 HEB300 HEB300 HEB300 HEB260
A HEB240 HEB200
8 B HEB320 HEB280 HEB280 HEB240
C HEB360 HEB320 HEB320 HEB320 HEB260
D HEB360 HEB320 HEB320 HEB320 HEB260
A HEB240 HEB180
9 B HEB280 HEB240 HEB200
C HEB360 HEB320 HEB300 HEB280 HEB260
D HEB340 HEB320 HEB300 HEB280 HEB260
A HEB240 HEB200
10 B HEB240 HEB200
C HEB360 HEB340 HEB340 HEB280 HEB260
D HEB360 HEB340 HEB340 HEB280 HEB260

290
 Table 1 - Member sections of designed frames (continued)

FRAME COLUMN STOREY 1 STOREY 2 STOREY 3 STOREY 4 STOREY 5

A HEB180
11 B HEB340 HEB320 HEB320 HEB320 HEB280
C HEB340 HEB320 HEB320 HEB320 HEB260
D HEB340 HEB320 HEB320 HEB320 HEB260
A HEB180
12 B HEB300 HEB280 HEB280 HEB240
C HEB340 HEB320 HEB320 HEB320 HEB260
D HEB340 HEB320 HEB320 HEB320 HEB260
A HEB180
13 B HEB280 HEB240 HEB200
C HEB340 HEB320 HEB300 HEB280 HEB260
D HEB340 HEB320 HEB300 HEB280 HEB260
A HEB180
14 B HEB240 HEB200
C HEB360 HEB360 HEB320 HEB280 HEB260
D HEB340 HEB340 HEB300 HEB280 HEB260
A HEB220
15 B HEB220
C HEB500 HEB500 HEB400 HEB260 HEB260
D HEB500 HEB500 HEB400 HEB260 HEB260

A B C D C B A
Figure 4 - Scheme for column identification

291
 Table 2 - Main results of push-over static inelastic analyses

FRAME y max 6 q6 Wy W1 q6 p
1 1.989 3.829 8.882 6.398 3.028 11.13 0.033 0
2 2.375 4.148 8.268 5.687 4.667 12.29 0.032 0.2444
3 2.627 4.610 8.162 5.632 5.658 13.40 0.032 0.3556
4 2.327 4.144 7.937 5.722 4.350 11.93 0.033 0.3778
5 2.643 4.637 7.942 5.668 5.615 13.43 0.034 0.3667
6 2.689 4.659 7.543 5.397 5.887 13.09 0.035 0.4889
7 2.122 3.953 8.597 6.015 3.605 11.42 0.044 0.5111
8 2.496 4.603 8.715 6.027 4.985 13.46 0.041 0.4148
9 2.581 4.733 8.039 5.619 5.494 13.17 0.042 0.5185
10 2.121 4.533 7.943 5.859 3.685 11.25 0.029 0.6222
11 2.043 3.804 8.171 5.790 3.463 10.77 0.035 0.6444
12 2.302 4.246 8.278 5.828 4.418 12.25 0.037 0.4722
13 2.487 4.487 8.141 5.629 5.213 12.85 0.040 0.5667
14 2.098 4.466 7.834 5.796 3.634 11.05 0.032 0.6611
15 2.298 5.413 7.502 5.789 4.377 12.11 0.031 0.7555

12
W u
q = (12)
W 1

where W u is the work done by horizontal forces up to the attainment of the ultimate displace-
ment and W 1 is the elastic work under the design horizontal forces ( = 1).
This value of the q-factor accounts also for the overstrength which unavoidably occurs
with respect to the design base shear. In fact, it can be divided into two parts according to the
following relationship:
12 12 (13)
Wy W u
q =q with q=
W 1 Wy

where W y is the work done by the horizontal forces up to the development of the first plastic
hinge.
The parameter ( W y W 1 )1 2 represents the overstrength with respect to the design base
shear, while the parameter q represents the actual frame energy dissipation capacity due to
plastic redistribution and global ductility.
The computed values are given in Tab.2 where also the main parameters of the curve
relating the multiplier of horizontal forces to the top sway displacement are given. These are:
- y multiplier of horizontal forces corresponding to the occurrence of the first plastic hinges;
- max maximum value of the horizontal forces multiplier;
- 6 global ductility corresponding to a rotation capacity equal to 6;
- q6 value of the q-factor computed according to equation (12) and by assuming a rotation
capacity equal to 6;

292
- d maximum plastic rotation corresponding to the attainment of the design value (0.64 m) of
the top sway displacement.
In the same table also the geometrical irregularity index computed as proposed in
[23] is given.
From the observation of Fig.5, where the q-factor computed according to equation (12)
is given as a function of the geometrical irregularity index , it can be stated that, by
designing set-back frames according to the proposed procedure, the seismic inelastic behaviour
is practically independent of the geometrical configuration. It means that properly designed
set-back frames cannot be considered as irregular.
An important result evidenced by numerical analyses is that the adoption of the
proposed design procedure leads also to the local ductility control. In fact, all the frames have
been designed by assuming that the ultimate top sway displacement is attained when the plastic
rotation demand is equal to 0.04 rad. The displacement corresponding to this assumption is
equal to 0.64 m. The maximum plastic rotation demand corresponding to the occurrence of the
above displacement is given in Fig.6 as a function of the geometrical irregularity index . This
figure shows that the plastic rotation demands corresponding to the design value of the top
sway displacement are always less than or very close to the value 0.04 rad which has been
considered in the design procedure.
In other words, independently of the geometrical configuration of the frame, the
proposed design procedure leads not only to the control of the failure mode, but also to the
local ductility control.

16
MODIFIED q-FACTOR

14

12

10

6
0 0.2 0.4 0.6 0.8
GEOMETRICAL IRREGULARITY INDEX
Figure 5 - Influence of geometrical configuration on inelastic behaviour

293
 0.05
design value
PLASTIC ROTATION
0.04

0.03

0.02

0.01

0
0 0.2 0.4 0.6 0.8
GEOMETRICAL IRREGULARITY INDEX
Figure 6 - Plastic rotation demands corresponding to the attainment of the design top sway displacement

CONCLUSIONS
A new method for designing moment resisting steel frames failing in a global mode has
been presented in this paper with reference to set-back frames which are usually considered as
irregular. The method is based on the extension of the kinematic theorem of plastic collapse to
the concept of mechanism equilibrium curve. This allows to include into the design process the
influence of second order effects and, in addition, leads to the control of the local ductility
demands corresponding to the development of a pattern of yielding according to the global
failure mode.
The reliability of the proposed design procedure has been also demonstrated through its
application to 15 frames, leading to the fulfilment of the design objective, as evidenced by the
push-over static inelastic analyses.
Finally, the examination of the seismic performance of the set-back frames designed
according to the proposed design procedure has clearly evidenced that properly designed
set-back structures cannot be considered as irregular, because their geometrical configuration
does not lead to any worsening of the seismic inelastic behaviour.

REFERENCES
[1] H. Akiyama: Earthquake Resistant Limit State Design for Buildings, University of
Tokyo Press, 1985.
[2] E. Rosenblueth (editor): Design of Earthquake Resistant Structures, Pentech Press,
London, 1980.
[3] D.J. Dowrick: Earthquake Resistant Design: A Manual for Engineers and Architects,
John Wiley & Sons, 1977.

294
[4] T. Paulay, M.J.N. Priestley: Seismic Design of Reinforced Concrete and Masonry Buil-
dings, John Wiley & Sons, 1995.
[5] M. Wakabayashi: Design of Earthquake Resistant Buildings, Mc Graw Hill Book Co.,
New York, p.229, 1986.
[6] V.V. Bertero, E.P. Popov: Seismic Behaviour of Ductile Moment-Resisting Reinforced
Concrete Frames, in Reinforced Concrete Structures in Seismic Zones, ACI Publica-
tion SP-53, American Concrete Institute, Detroit, pp. 247-291, 1977.
[7] Commission of the European Communities: Eurocode 8: European Code for Seismic
Regions - Design, Part.1.3: Buildings, ENV, November 1994.
[8] UBC: Uniform Building Code, International Conference of Building Officials, Whit-
tier, CA, 1990.
[9] AISC-LRDF: Seismic Provisions for Structural Steel Buildings - Load and Resistance
Factor Design, American Institute of Steel Construction, November 15, 1990.
[10] AIJ: Standard for Limit State Design of Steel Structures (Draft), Architectural Institute
of Japan, 1990 (English version, October 1992).
[11] V.V. Bertero, S.W. Zagajeski: Optimal Inelastic Design of Seismic-Resistant Reinforced
Concrete Framed Structures, in Non-Linear Design of Concrete Structures, CSCE-
ASCE-ACI-CEB International Symposium, University of Waterloo, Ontario, Canada,
pp.219-272, 1979.
[12] Han-Seon Lee: Revised Rule for Concept of Strong-Column Weak-Girder Design,
Journal of Structural Engineering, ASCE, Vol.122, No.4, pp.359-364, April, 1996.
[13] F.M. Mazzolani, V. Piluso: Plastic Design of Seismic Resistant Steel Frames, Ear-
thquake Engineering and Structural Dynamics, paper accepted for publication on August
1996 (in press).
[14] M. Como, G. Lanni: Elementi di Costruzioni Antisismiche, Cremonese, Roma, 1979.
[15] E. Cosenza: Duttilit Globale delle Strutture Sismo-Resistenti in Acciaio, PhD Thesis,
Universit di Napoli, Aprile 1987.
[16] F.M. Mazzolani, V. Piluso: A new method to design steel frames failing in global mode
including P- effects, in Behaviour of Steel Structures in Seismic Areas edited by
F.M. Mazzolani and V. Gioncu, Proceedings of the International Workshop (Timisoara,
Romania, 26 June - 1 July, 1994), E & FN SPON, an Imprint of Chapman & Hall, pp.
300-309, 1995.
[17] F.M. Mazzolani, V. Piluso: Failure Mode and Ductility Control of Seismic Resistant
MR-Frames, Costruzioni Metalliche, N.2, pp. 11-28, 1995.
[18] K.C. Tsai, E.P. Popov: Steel Beam-Column Joints in Seismic Moment Resisting Fra-
mes, Earthquake Engineering Research Center, University of California at Berkeley,
Report No. UCB/EERC-88/19, November, 1988.
[19] V.V. Bertero, J.C. Anderson, H. Krawinkler: Performance of Steel Building Structures
during the Northridge Earthquake, Earthquake Engineering Research Center, University
of California at Berkeley, Report No. UCB/EERC-94/09, 1994.

295
[20] M.N. Nader, A. Astaneh-Asl: Seismic Behaviour and Design of Semirigid Steel Fra-
mes, Earthquake Engineering Research Center, University of California at Berkeley,
Report No. UCB/EERC-92/06, April, 1992.
[21] A. Astaneh-Asl: Seismic Design of Bolted Steel Moment-Resisting Frames, Structural
Steel Educational Council, Technical Information and Product Service, July, 1995.
[22] M. Como, G. Lanni: Aseismic Toughness of Structures, Meccanica, 18, pp. 107-114,
1983.
[23] C.A. Guerra, F.M. Mazzolani, V. Piluso: Le Conseguenze delle Irregolarit Strutturali:
Problematiche e Prospettive, IV Convegno Nazionale, LIngegneria Sismica in Italia,
Milano 5-7 Ottobre, 1989.

296