Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/engstruct
Abstract
New relations to estimate maximum seismic displacement profiles of plane steel moment resisting frames are proposed. These expressions
associate maximum floor displacements with local member deformation seismic demands and are based on statistical analysis of the results
of several hundred nonlinear dynamic analyses of steel frames. The influence of specific parameters, such as the number of stories, the
number of bays, the joint capacity design factor and the level of inelastic deformation induced by the seismic excitation, is studied in detail.
It is concluded that the main structural characteristic that affects the shape of the displacement pattern is the number of stories. Furthermore,
the present study reveals that a differentiation between the profile of a frame in the elastic and the inelastic range of response is necessary. A
comparison between the proposed displacement patterns and other existing ones is also made to demonstrate the merits of the former.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Maximum displacement profile; Interstory drift ratio; Performance based seismic design; Steel frames
0141-0296/$ - see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2005.06.021
10 T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22
relations dealing with the strength reduction factor, the new hybrid methodology for the performance-based seismic
ductility and the period of vibration of SDOF systems [5–8], design of plane steel moment resisting frames (MRF)
or by using directly the results of a statistical analysis on the has been developed by the authors [23]. Although this is
ratio of the maximum inelastic over the maximum elastic an adaptation of force-based design in the sense that a
displacement [9–13]. The second method is based on the strength reduction factor is used, it focuses on damage
concept of equivalent linearization in which the maximum and displacements rather than forces. The first step of this
deformation of an inelastic SDOF system is approximated displacement/force hybrid (DFH) method needs a fairly
by the maximum deformation of an elastic SDOF system accurate estimation of the maximum floor displacements of
with stiffness smaller than the initial stiffness of the inelastic the frame.
SDOF system and with a damping ratio larger than that The abovementioned expressions for maximum displace-
of the inelastic system. Many research works have been ment profiles are somewhat limited in that they have been
devoted in the past to the study of equivalent linearization, derived for concrete frames in the linear elastic range and
as is evident in Iwan and Gates [14] and Miranda and are functions of only one parameter, the number of frame
Ruiz-Garcia [15]. The translation of the maximum SDOF stories. In this work expressions for maximum displacement
displacement to maximum floor displacements of the profiles of plane steel MRF undergoing elastic and inelas-
building structure can be achieved by using appropriate tic deformation are provided as functions of geometrical pa-
conversion factors which are based, either on statistical rameters of the frame. This was achieved by an extensive in-
analysis of a large number of dynamic analyses [16], or on vestigation involving several hundreds of nonlinear dynamic
the concept of the constant deformed shape of the structure analyses of steel frames. The results of these analyses were
during the seismic excitation [8]. post-processed in order to identify the main parameters that
Well known procedures adopted by seismic codes affect the displacement profile of these frames. These pa-
and provisions, which follow the abovementioned general rameters consist of basic structural characteristics of a plane
three-stage approach (pushover analysis of the MDOF — steel frame, such as the number of stories, the number of
substitute SDOF — maximum floor displacements of the bays and the joint capacity design factor used during the
MDOF) for the estimation of maximum floor displacements, phase of the selection of sections. Furthermore, the shape
are the capacity spectrum method [17], the NSP FEMA-273 of the displacement profile was checked at different levels
method [18] and the N2 method of Fajfar [8,1]. Maximum of damage of the structure and this checking showed that a
floor displacements of building frames can also be computed differentiation between the shape of the profile in the linear
by utilizing methods which are different than the usual three- and nonlinear range of the response is necessary. The pro-
stage approach [19,20]. The common characteristic of all posed relations directly relate the maximum floor displace-
the abovementioned methods is that a pre-designed structure ments with the maximum IDR and are given as functions
is needed in order to perform a pushover or a dynamic of the abovementioned structural characteristics. Thus, they
analysis. Thus, these procedures are suitable for the seismic are suitable for explicitly performance based design meth-
evaluation of existing structures or for the iterative design of ods, such as the DBD or the DFH methods.
new structures.
In order to use maximum displacements as the starting 2. Computational foundation of the present work
design variables of a design methodology, a simple relation
for the maximum floor displacement pattern is needed. This In this section, the computational methodology for the
relation should be independent of the sections of the frame derivation of the new maximum displacement patterns of
and moreover, should correlate displacements with damage plane steel moment resisting frames is described, in detail.
in order to be applicable to various limit states. According In order to identify the structural parameters of a steel frame
to the authors’ knowledge, the only relations that describe that significantly affect the shape of the profile, a set of forty-
the maximum displacement profile of plane frames are those two plane steel moment resisting frames was employed for
proposed by Loeding et al. [21]. These profiles relate the the parametric studies of this work. A typical such frame
maximum story displacements with the maximum interstory with two bays and three stories is shown in Fig. 1. These
drift ratio (IDR) and their derivation is based on the frames have the following characteristics: number of stories
results of linear multi-modal dynamic analyses of concrete (n s ) with values 3, 6, 9, 12, 15, 18, 20; number of bays (n b )
frames. It should be emphasized that displacement profiles with values 3, 6; joint capacity factor (acd ) used in the design
of this kind do not describe real deformed shapes, since the phase with values 1.1, 1.3, 1.5. The joint capacity factor acd
maximum story displacements are in general asynchronous is defined as
during seismic excitation. Furthermore, the story where the
M Rc
maximum IDR will be first reached is not known. The use acd = (1)
of the maximum displacement profile has been adopted M Rb
by the SEAOC seismic design manual (blue book) [22] in where M Rc is the sum of the plastic moment of resistance
connection with the first step of the direct displacement of columns framing a joint and M Rb is the sum of
based design procedure (DBD) proposed therein. Recently, a the plastic moment of resistance of beams framing the
T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22 11
Table 1
Steel moment resisting frames considered in the parametric studies
presented in various forms, depending on the selection of values is the same. The second one calculates the central
the statistical central value, e.g., mean or median and on the value by using the formula
method that this central value is calculated with. The median n
is the central value of the log-normal distribution, which
j =1 ln U j,i
is an appropriate description for earthquake response [25].
Um,i = exp (2)
Thus, if one assumes that one’s data are sampled from n
a population that follows the log-normal distribution,
the central value can be calculated by employing two
different procedures which belong to the theory of statistical where Um,i is the central value of the maximum
inference [26]. The first procedure calculates the median as displacement of the floor i of a frame and n is the number
the number below and above of which the number of sample of seismic excitations. The results of the nonlinear dynamic
T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22 13
Table 2
Information pertinent to the ground motions used in the parametric studies
analyses showed small differences between the values of the influence is obvious at the upper stories in the case of the
median calculated by the two abovementioned procedures. nonlinear range of the response.
Moreover, a measure of dispersion is needed in order 4. The number of stories is the main structural characteristic
to complete the descriptive statistics of the maximum that affects the shape of the displacement profile. Tall
displacement profile of steel MRF. The measure of frames show a parabolic profile, while short ones show
dispersion adopted with respect to the central value of Eq. an almost linear profile.
(2) is the standard deviation of the logarithm of the n sample 5. The shape of the profile is different in the linear and the
values, defined as nonlinear range of the response.
6. An increase in the number of bays produces a small
n reduction on the maximum floor displacement. This
(ln U j,i − ln Um,i )
i=1 influence is lost in the linear range of the response.
σm,i = . (3)
n−1
It can be proved [26] that for a log-normal distribution, Eq. 4. Proposed relations for the maximum displacement
(3) takes the form profile of steel moment resisting frames
σm,i = ln(1 + COV 2 ) (4) Based on the general conclusions presented in Section 3
where COV is the coefficient of variation, defined as the ratio of the paper, the whole data-bank is separated into two parts.
of the standard deviation over the mean. Thus, according to The first part consists of central value profiles associated
Eq. (4), σm,i is almost identical to COV for small values with the linear range up to the first plastic hinge limit
of dispersion, e.g., COV < 30%. In Figs. 3–16, central and state, while the second consists of central value profiles
dispersion values, as defined by Eqs. (2) and (3) respectively, associated with the nonlinear range of the response. The
are illustrated for the maximum floor displacements of the equation for the mathematical representation of the physical
frames considered in the parametric studies. relationship between damage, e.g. IDR, and maximum floor
The following conclusions can be drawn from the results displacements Ui at floor i was selected to have the form:
(Figs. 3–16) of the present study: hi
Ui = P1 · IDR · h i · 1 − P2 · . (5)
1. Frames of lower than six stories show a small dispersion H
on maximum displacements, while frames of more than The calculation of the parameters P1 and P2 is done
six stories show large values of dispersion. with the aid of Table 3 as a function of the number
2. The heightwise variation of the dispersion depends upon of stories of the frame and the desired response range
the level of inelastic deformation. In the linear range of (elastic or inelastic). Please note that the three values of P1
the response, the larger value of dispersion appears at separated by a comma in the first column of the inelastic
the upper stories, while in the nonlinear range of the response case of Table 3 correspond to the three values
response, the larger value of dispersion is located at the of the joint capacity design factor, namely 1.1, 1.3 and
lower stories. 1.5. This influence of the joint capacity design factor on
3. The joint capacity design factor used in the design the value of maximum floor displacements comes as a
of the frame affects the value of the maximum floor physical consequence. Thus, for example, a structure with
displacement, especially for tall frames. Larger values of a high value of acd , which distributes inelasticity uniformly
acd produce larger maximum floor displacements. This along its height, can achieve higher values of maximum
14 T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22
Fig. 5. Central values of the maximum floor displacements of 6 story frames Fig. 6. Dispersion values of the maximum floor displacements of 6 story
associated with (a) first plastic hinge; (b) maximum interstory drift ratio frames associated with (a) first plastic hinge; (b) maximum interstory drift
equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
Fig. 7. Central values of the maximum floor displacements of 9 story frames Fig. 8. Dispersion values of the maximum floor displacements of 9 story
associated with (a) first plastic hinge; (b) maximum interstory drift ratio frames associated with (a) first plastic hinge; (b) maximum interstory drift
equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
HEB280 sections and beams with IPE360 sections. The nonlinear time history analysis. The response spectra of
fundamental period of vibration of the frame is equal to these motions, in comparison with the design spectrum, are
1.1 s. The expected ground motion is defined by the elastic depicted in Fig. 20.
acceleration design spectrum of the EC8 seismic code [1] According to the N2 method [1,8], the following
with peak ground acceleration equal to 0.60g and soil class procedure is employed in order to compute the maximum
C. Three semi-artificial accelerograms compatible with the floor displacements of the steel frame of Fig. 19: a linear
EC8 design spectrum were generated via a deterministic displacement shape (inverted triangle), unchanged during
approach [27], in order to effectively compare the estimated the action of the ground motion, is assumed of the form
floor displacements with those derived from the rigorous Φ T = [0.166, 0.333, 0.5, 0.666, 0.833, 1]. The lateral force
T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22 17
Fig. 9. Central values of the maximum floor displacements of 12 story Fig. 10. Dispersion values of the maximum floor displacements of 12 story
frames associated with (a) first plastic hinge; (b) maximum interstory drift frames associated with (a) first plastic hinge; (b) maximum interstory drift
ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
Pi at each floor i with mass m i can then be directly obtained displacement of the curve of Fig. 21 by a transformation
as Pi = m i Φi . Because the masses concentrated at each factor Γ equal to (Σ m i Φi )/(Σ m i Φi2 ), which in this example
floor are equal, the load pattern has a constant inverted is equal to 1.38. Afterwards, the idealized elastic–perfectly
triangular shape. With this force pattern, the DRAIN-2DX plastic force–displacement curve (shown in Fig. 22 by the
program [24] performs a pushover analysis and produces broken line) is determined by assuming that the yield force
the base shear V — top displacement Ut curve which is Fy is equal to the force that results in the formation of the
depicted in Fig. 21. The force–displacement curve (shown plastic collapse mechanism of the building and the yield
in Fig. 22 by the solid line) of the equivalent SDOF system displacement d y is such that the areas under the actual
is then determined by dividing the base shear and top- and the idealized curves of Fig. 22 are equal. Thus, for the
18 T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22
Fig. 11. Central values of the maximum floor displacements of 15 story Fig. 12. Dispersion values of the maximum floor displacements of 15 story
frames associated with (a) first plastic hinge; (b) maximum interstory drift frames associated with (a) first plastic hinge; (b) maximum interstory drift
ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
SDOF one has Fy = 774 kN and d y = 0.14 m. The displacement, which was used for the calculation of the
mass of the SDOF system is calculated as m = Σ m i Φi = area under the actual force–displacement curve, was almost
176.5 kN m−1 s2 . The elastic period of the SDOF system equal to 0.32 m, thus no iteration is needed. The seismic
is calculated as T = 2π(md y /Fy )0.5 = 1.12 s. Because demands are then transformed back to the MDOF system
this period is larger than the characteristic period Tc of by performing a pushover analysis up to a top displacement
the design spectrum, the equal displacement rule applies. equal to 0.32Γ = 0.44 m.
This assumption leads to a target displacement equal to According to Loeding et al. [21], the maximum floor
0.32 m, calculated as the design spectral ordinate that displacements of a six story frame are calculated through
corresponds to a period equal to 1.12 s. The initial SDOF the expression Ui = IDRh i (1 − 0.0625h i /H ), where
T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22 19
Fig. 13. Central values of the maximum floor displacements of 18 story Fig. 14. Dispersion values of the maximum floor displacements of 18 story
frames associated with (a) first plastic hinge; (b) maximum interstory drift frames associated with (a) first plastic hinge; (b) maximum interstory drift
ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
h i and H are the height of floor i and the height of for the maximum IDR is 0.0343, calculated at the step
the roof from the base, respectively. The maximum floor of the pushover analysis which corresponds to the target
displacement profile of the test steel frame proposed in displacement. A graphical comparison of the maximum floor
the present work is expressed as Ui = 0.9IDRh i (1 − displacements is presented in Fig. 23. This graph shows that
0.2h i /H ). The results of the three dynamic analyses in terms the displacement profiles proposed by Loeding et al. [21]
of the maximum IDR are 0.024, 0.019 and 0.025. These and the N2 method [8,1] overestimate floor displacements.
values are used in the abovementioned two relations for On the other hand, the proposed relations seem to predict
the prediction of maximum floor displacements. It should fairly accurately the maximum floor displacements at
also be mentioned that the prediction of the N2 method various levels of inelastic local deformation.
20 T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22
Fig. 15. Central values of the maximum floor displacements of 20 story Fig. 16. Dispersion values of the maximum floor displacements of 20 story
frames associated with (a) first plastic hinge; (b) maximum interstory drift frames associated with (a) first plastic hinge; (b) maximum interstory drift
ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%. ratio equal to 2.5%; (c) maximum interstory drift ratio equal to 4%.
Table 3
Values of the parameters of the proposed maximum displacement profile
Fig. 20. Response spectra of the ground motions used in the test example.
(4) The new profiles are section independent and reflect the Fig. 21. Pushover curve of the test example steel frame.
physical relation between damage, expressed through the
maximum interstory drift ratio, and the maximum floor
displacements. Thus, they can be directly used at the via a rigorous nonlinear dynamic analysis. On the other
first step of performance based design methods such as hand, other proposed procedures or relations are found
the DBD or the HFD, where the sections are not known to overestimate the true displacements.
as yet. (6) The proposed relations are valid for regular plane steel
(5) The maximum floor displacements of plane steel moment resisting frames. The authors are currently
moment resisting frames obtained by utilizing the extending the proposed equations to the case of irregular
proposed relations are found to be close to those obtained steel frames.
22 T.L. Karavasilis et al. / Engineering Structures 28 (2006) 9–22