Professional Documents
Culture Documents
Refered Journals
Refered Journals
Refered Journals
net/publication/262260068
CITATIONS READS
25 797
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Roberto Realfonzo on 23 November 2016.
ABSTRACT
The aim of the paper is to establish rational and, at the same time, easy to determine, criteria
for regularity in elevation; furthermore, these criteria should make a clear boundary between
regular and irregular in elevation structures. The first step to do in order to pursue this aim is
to analyse the seismic response of irregular structures in the nonlinear range of behaviour;
this is carried out by a comparison with reference cases, i.e. regular structures. Mass, stiffness
and strength irregularity in elevation are studied separately. A large number of parametric
analyses is carried out on very simple shear type frames, designed according to EC8. A
bilinear with 5% hardening monotonic curve, and no stiffness and strength degradation with
cycling, are assigned as storey shear – interstorey drift relationship. The number of storeys,
the seismic input and the entity and location of mass, stiffness and strength discontinuities in
elevation are changed to take into account many real cases. The total masses are modified in
order to compare the seismic behaviour of frames having the same first period.
INTRODUCTION
It is well known that under seismic excitations buildings characterised by an irregular in
elevation distribution of the geometrical and mechanical characteristics show - with respect to
regular buildings - increments of the structural and non structural damage, which can also be
large; this is due to the coming up of unfavourable dissipation mechanisms, associated to the
concentration of the plastic demand in few sections.
The first studies on the seismic behaviour of buildings characterised by discontinuities along
the height in terms of mass, stiffness and strength are dated back to the 1980s [1],[2].
In the nineties the interest for the seismic behaviour of irregular structures reaches the top; in
1994 Task Group 8 was established under the auspices of the European Association of
Earthquake Engineering (EAEE) with the mandate to study the “Seismic Behaviour of
Irregular and Complex Structures”.
A wide parametric investigation on the seismic behaviour of irregular in elevation frame
structures was recently carried out by Al-Ali and Krawinkler [3]. Shear type plane frames
with discontinuities in terms of mass, stiffness and strength were taken into account. The
results show that large irregularities of the mass distribution cause negligible variations in
terms of interstorey drifts and ductility demands with respect to the reference regular case; on
the contrary, a soft storey, due to large stiffness variations, causes, at its level, large
increments of the interstorey drifts; these ones represent, according to the authors, a reliable
index of the structural and non structural damage. Finally, as consequence of an irregular
strength distribution due to more or less large increments or reductions at a certain storey
with respect to the regular distribution, large variations with respect to the regular case of the
ductility demand associated to increments of the interstorey drifts and dissipated energy were
observed.
The development of the scientific research during the course of the years has determined the
contemporaneous development of the international seismic codes. In the following the
Eurocode 8 (CEN [4]) and the Uniform Building Code (UBC [5]) provisions regarding mass,
stiffness and strength irregularities are briefly presented.
The Eurocode 8 (EC8) provides 3 criteria for regularity in elevation. All of them have to be
contemporaneously satisfied for defining a building as regular: all lateral load resisting
systems run without interruption from their foundations to the top of the building; both the
lateral stiffness and the mass of the individual storeys remain constant or reduce gradually,
without abrupt changes, from the base to the top; in framed buildings the ratio of the actual
storey resistance to the resistance required by the analysis should not vary disproportionately
between adjacent storeys. The multi-modal analysis and a 20% reduction of the behaviour
factor (q) are prescribed for the design of structures, which do not respect the aforesaid
criteria. Even though they are clearly defined from a qualitative point of view, these criteria
are not quantitatively defined and consequently do not seem objective.
On the contrary, the Uniform Building Code (UBC) provides “quantitative” criteria for
defining structural irregularities. These criteria, concerning the here examined types of
irregularity, can be summarised as follows; mass irregularity: the effective storey mass is
grater than 150% of the mass of an adjacent storey; stiffness irregularity: a soft storey is
defined when its stiffness is less than 70% of the storey above or less than 80% of the mean
stiffness of the three storeys above; strength irregularity: a weak storey is defined when its
strength is less than 80% of the storey above.
Some available research studies give the opportunity to verify the reliability of code criteria
for regularity and related design provisions. An evaluation of the design provisions
prescribed by UBC was recently carried out by Valmundsson and Nau [6]. The Authors
analysed the seismic behaviour of shear type frames which, according to the aforesaid
provisions, were in a threshold condition between regularity and irregularity. The obtained
results allowed to conclude that the UBC criteria are not well balanced and therefore an
evaluation of new more reliable methods is necessary. Furthermore the Authors underlined
the high sensitivity of the response to irregular strength distributions.
Recently Magliulo et al. [7] studied the problem of the determination of quantitative criteria
for regularity. Concerning mass irregularity, they noticed that large variations of the mass
distribution in elevation cause negligible modifications of the seismic response with respect
to the reference regular case. Based on the results obtained on the frames with stiffness
irregularities, an alternative to the existing ones criterion was proposed, characterised by the
evaluation of interstorey drifts obtained by elastic analysis. Finally, the high sensitivity of the
response to strength irregular distributions and the unreliability of the quantitative criteria
provided by UBC were confirmed.
In this paper the already performed parametric analysis is extended to frames with different
geometry. The already proposed criterion for the determination of the stiffness irregularity in
elevation is further verified using the results of the new analyses and, furthermore, the non
linear dynamic behaviour of frame buildings characterised by overstrength irregular
distributions is studied more in detail.
EXAMINED FRAMES
In the following the seismic behaviour of shear type plane frames characterised by five and
nine storeys is analysed. For each type of frame, different irregular in elevation distributions
of mass, stiffness and strength are taken into account and the comparison with the reference
regular frame is always carried out. Their geometry is shown in Figure 1. In order to study
the stiffness irregularity in elevation the behaviour of not tapered 5 storey frames is also
analysed; in this case, with respect to the geometry shown in Figure 1 regarding the tapered
frame, the cross section of all the columns is constant and equal to 30x60 centimetres.
30x50 30x60 30x60 30x50 3.5 m
5m 5m 5m 5m 5m
At each floor the seismic mass of the tapered and not tapered 5 storey regular frames, is equal
to 36.74 ton, whilst the mass of the 9 storey regular frame is equal to 49.34 ton.
The beams are always assumed with unlimited stiffness and strength. The column yielding
strength is calculated by multi-modal analysis according to the EC8 (CEN [4]) design
spectrum, soil type B (medium); the assumed peak ground acceleration (PGA) is 0.4g, the
damping equal to 5% and the behaviour factor equal to 5.
Not taking into account the shear flexibility of the columns, the first period of the reference
frames is: 0.529 sec for tapered, 0.522 sec for not tapered five storey frames; 0.82 sec in the
case of nine storey frame.
The first period of the irregular frames, obtained varying the mass or stiffness distribution
with respect to the reference frame ones, is always kept constant, in order to not have
different spectral values of each accelerogram, corresponding to the first period of different
frames. This is obtained varying the total mass of the structure; obviously the design take into
account this variation, assigning to the elements of each frame the yielding strength according
to the multi-modal analysis.
NUMERICAL ANALYSES
As regards the numerical model used in the non linear analyses, the shear-type frames are
characterised by a bilinear, with 5% hardening, storey shear – interstorey drift relationship;
the hysteretic behaviour does not take into account the strength and stiffness degradation both
in loading and unloading branches and it is not conditioned by the axial force variation due to
the action of the horizontal forces.
Five real accelerograms are used as seismic input in the numerical analyses: the main
characteristics of the used accelerograms are presented in Table 1. These earthquakes are
chosen because they represent different types of seismic excitation, with different entities of
the peak ground acceleration, different duration and different response spectrum shape. It is
to be noted that the mean response spectrum, obtained by the average, for each period, of the
spectral accelerations characterising the 5 accelerograms, well fits the EC8 elastic spectrum,
for PGA equal to 0.4 g, soil type B and damping equal to 5%, which is the one used for the
design of the frames (Magliulo et al. [7]).
TABLE 1
ACCELEROGRAMS
Earthquake Date Place Duration [sec] PGA [g]
Imperial V. 18.05.40 El Centro 53.40 0.348
Kern County 21.07.52 Taft 54.40 0.179
Montenegro 15.04.79 Petrovac 19.60 0.438
Valparaiso 03.03.85 El Almendral 72.02 0.284
Northridge 17.01.94 Newhall 59.98 0.590
The non linear numerical analyses are performed scaling the accelerograms, in order to make
their elastic spectra values, corresponding to the first period of the analysed frames, equal to
the one of the aforesaid EC8 elastic spectrum. This is done in order to minimise the
differences of the effects on the frames of different seismic excitations, at least until the
structural behaviour is elastic. The frame non linear seismic response is evaluated in terms of
interstorey drifts and ductility demand. The analysed values are the maximum reached during
the relevant time history. In the following for each effect the shown value is the average of
the ones obtained by the five earthquakes. Analysing the effects of each accelerogram, a not
negligible dependence on the typology of the seismic input has been noted.
By the results of the numerical analyses, the irregular cases are clearly defined, that is those
ones characterised by a response very different from the reference structure one. In this way
the EC8 qualitative criteria can be evaluated and the reliability of the UBC quantitative ones
analysed. Finally, the analysis of the dynamic elastic response permits to associate to the
irregular response simple and objective criteria, able to separate the irregular in elevation
structures from the regular ones.
The analysed cases are shown in Figure 2, where “Up”, “Central” and “Down” clearly refer
to the position of the floor where the mass which gives the irregularity is located (in the “Up”
case at the fifth floor, in the “Central” at the third and in “Down” at the first).
Considering that the geometry, and so the stiffness, of the four frames is the same, in order to
have the value of the first period always equal to the reference frame one (T1 = 0.529 sec), the
total masses of the four frames are different one from the other. Referring to the Figure 2:
M1=0.78M, M2=1.05M, M3=1.19M, where M is the total mass of the reference frame.
It is to be underlined that the different mass distribution and, above all, the different value of
the total masses lead to an internal force distribution different for one case from the other.
Nevertheless, these differences are appropriately taken into account by design assigning to
each storey the strength evaluated by multi-modal analyses.
REGULAR UP
M/5 M /3
1
M/5 M /6
1
M/5 M /6
1
M/5 M /6
1
M/5 M /6
1
CENTRAL DOWN
M /6
2 M /6 3
M /6
2 M /6 3
M /3
2 M /6 3
M /6
2 M /6 3
M /6
2 M /3 3
In Figure 3 the four frames responses are compared considering the average – on the five
aforesaid earthquakes – both of the interstorey drifts divided by the interstorey height in
percentage and of the ductility demands. In the figure can be noted as, regarding the
interstorey drifts, the distributions of the masses, irregular according to the aforesaid codes,
do not cause evident differences with respect to the response obtained for the reference case:
only light increments at the fifth storey in the case of the “Up” frame, at the first storey for
the “Down” frame and at the second and third storey for the “Central” are obtained. These
considerations are confirmed analysing the ductility demands: the differences are even
lighter.
5 5
Regular
Up
4 4
Central
Down
Regular
Floor
Floor
3 3
Up
Central
Down
2 2
1 1
0.0 0.5 1.0 1.5 0 5
interstorey drift [%] Ductility demand 10 15
Figure 3: Mass supposed irregular frames: interstorey drifts and ductility demands
In Magliulo et al. [7] the results of non linear analyses on the here shown frames designed by
ELF procedure were analysed. This study was carried out in order to understand if the light
differences between the reference case response and the supposed irregular cases one are due
to the capacity of the multi-modal analysis to mitigate the effects of the mass discontinuities
assigning a more appropriate strength distribution. Nevertheless, was noted that, even
designing by ELF procedure, the differences between the seismic responses of the reference
frame and the supposed irregular ones are almost negligible.
In conclusion, all the obtained results show that “reasonable” variations of the mass do not
lead to very different responses from the reference regular building one. Probably it is
possible to obtain irregular behaviours with very irregular mass distributions, but does not
seem reasonable to suppose, for real cases, mass variations in elevations larger than the
already considered ones. After all, the codes regularity criteria concerning the mass
distribution do not seem to be justified.
In the following the results concerning the 5 storey frames already presented in Magliulo et
al. [7] regarding 2 different typologies (tapered and not tapered) are presented and the
parametric analyses are extended to the shown 9 storey model. For the sake of brevity, the
reported non linear dynamic analysis results only concern to the “D” cases; however, similar
indications are obtained by the frames with stiffness variation at the central storey (“C”).
In Figures 4, 5 and 6 the results, in terms of interstorey drifts and storey ductility demand, of
the non linear analyses of all the examined “D” cases and the comparison with the respective
reference regular frames are shown. It is to be underlined that the stiffness variations,
assigned in order to make irregular the reference model, modify the yielding interstorey drift,
thus conditioning the values of the ductility demand, which, therefore, does not seem to be
reliable parameter for evaluating the plastic demand (Al-Ali & Krawinkler [3]). For this
reason, the following evaluations will be carried out, only considering the interstorey drifts.
It is preliminarily to be noted that, according to the UBC criteria, are irregular only the
frames D20 and D40 in the tapered case (Figure 4) and D20, D40 and D60 in the not tapered
case (Figure 5), while in the case of the 9 storey frame (Figure 6) are irregular the frames
D20 and D40.
Analysing the results shown in the figures, a general accordance with the UBC provisions can
be noted. All the cases defined irregular according to the US code indications show
increments larger than 20% of the interstorey drift with respect to the reference frame.
Nevertheless, in Magliulo et al. [7] was noted that the UBC irregularity criterion is
characterised by not negligible discrepancies. Indeed, the five storey tapered frame D60,
regular according to UBC, shows an increment, with respect to the regular tapered frame, of
the interstorey drift equal to 34%, larger of the 29% found for the not tapered D60, irregular
according to the aforesaid code.
In order to better fit the results of the non linear dynamic analyses, a new criterion for the
stiffness irregularity in elevation is proposed, which is based on the evaluation of the
interstorey drifts obtained by elastic analysis (Magliulo et al. [7]); it is to be underlined that
these are easier to calculate with respect to the storey stiffness. Herein this criterion is tested
in the case of the 9 storey frames for a further validation.
5 D20 5
D40 D20
D60 D40
4 D80 4 D60
Reg D80
D120 Reg
Floor
Floor
3 D140 3 D120
D140
2 2
1 1
0 0.5 1 1.5 2 0 5 10 15
Interstorey drift [%] Ductility demand
Floor
Floor
3 3 D140
D140
2 2
1 1
0 0.5 1 1.5 2 0 5 10 15
interstorey drift [%] Ductility demand
Figure 5: 5 storey not tapered frames: interstorey drifts and ductility demand
9 9
D20 D20
8 8
D40 D40
7 D60 7 D60
D80 D80
6 6
Reg Reg
Floor
Floor
5 D120 5 D120
D140 D140
4 4
3 3
2
2
1
1
0.0 0.5 1.0 1.5 2.0
0 5 10 15
Interstorey drift [%] Ductility demand
For the 5 storey (tapered and not tapered) frames and 9 storey ones, the results of the elastic
analyses in terms of interstorey drifts δi are shown in Table 2:
δi=(∆i-∆i-1)/Hi*100;
where ∆i is the i-th storey displacement and Hi is the interstorey height.
Furthermore, the values αi, obtained as ratio between the interstorey drifts are shown:
αi = δi / δi + 1 = (∆ i − ∆ i − 1 ) / (∆ i + 1 − ∆ i ) ;
obviously, these ratios cannot be calculated at the upper storey.
It can be noted that, both for the 5 storey frame (tapered and not tapered) and for the 9 storey
frame, as regards the regular frame the trend of these ratios is monotonically decreasing from
up to down. This does not happen for frames obtained by the reference frame varying the
storey stiffness distribution. For all the very irregular frames, i.e. D20 and D40, a sharp
increment of the parameter α can be observed at the storey characterised by the stiffness
reduction. Calculating the ratios αi/αi+1, shown in the fourth column of each table, this not
uniform variation of the interstorey drifts along the height can be more clearly highlighted.
TABLE 2
INTERSTOREY DRIFTS OBTAINED BY ELASTIC ANALYSIS (5 STOREY FRAMES)
D20 – tapered D40 – tapered D60 – tapered D80 – tapered Regular - tapered
FN δi αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1
1 0.238 4.080 4.85 0.166 2.100 2.59 0.111 1.285 1.62 0.086 0.953 1.21 0.070 0.756 0.96
2 0.058 0.836 0.93 0.079 0.811 0.93 0.086 0.795 0.93 0.090 0.786 0.93 0.092 0.781 0.93
3 0.070 0.896 0.79 0.097 0.868 0.79 0.108 0.852 0.79 0.114 0.845 0.79 0.118 0.840 0.80
4 0.078 1.131 0.112 1.094 0.127 1.072 0.135 1.063 0.140 1.058
5 0.069 0.102 0.119 0.127 0.132
D20 – not tapered D40 – not tapered D60 – not tapered D80 – not tapered Reg. – not tapered
FN δI αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1 δi αi αi/αi+1
1 0.303 5.949 4.65 0.240 2.879 2.24 0.195 1.878 1.52 0.162 1.387 1.14 0.139 1.096 0.90
2 0.051 1.284 0.87 0.083 1.252 0.87 0.104 1.232 0.86 0.117 1.218 0.86 0.126 1.209 0.87
3 0.040 1.459 0.74 0.067 1.433 0.73 0.084 1.415 0.74 0.096 1.403 0.73 0.105 1.394 0.73
4 0.027 1.965 0.046 1.943 0.060 1.925 0.069 1.915 0.075 1.905
5 0.014 0.024 0.031 0.036 0.039
It is to be underlined as such ratios are almost coincident at each storey for all the frames
having the same typology, while they are very different at the lower storey, where the
irregularity is artfully assigned. Larger is the stiffness reduction, larger is the ratio αi/αi+1 at
that storey. A value of αi/αi+1 larger than 1.5 seems to be able to define the same threshold for
regularity as given by UBC, with advantage to eliminate the lack of uniformity in judging the
2 frames D60, which as already said, UBC judges differently. Furthermore, it is to be
underlined that for αi/αi+1 larger than 1.50 at the storey characterised by the sharp stiffness
reduction (soft storey), increments of the interstorey drifts larger than 20% with respect the
reference regular cases can be always observed.
Considering the Figure 7 (5 storey tapered frames), is to be noted that, applying the UBC
irregularity criteria, only four frames are irregular: “r60(3)”, “r40(2-5)”, “r60(2-5)” and
“r60(1-2,4-5)”. Nevertheless, observing the diagrams in the figure, can be clearly noted that
some other cases, regular according to UBC, present a very different behaviour from the
regular one, with not negligible increments of interstorey drifts. For example, the frame
“r40(1-2/4-5)”, obtained increasing the strength of the regular frame of the 40% at each
storey less than the third one, shows increments of the interstorey drifts equal to about 100%
at the third storey, where a weak storey behaviour can be clearly seen. Nevertheless, the
aforesaid frame is regular according to the American code. The frames “r20(1-2,4-5)”, r20(2-
5) show other contradictions between the non linear analyses results and the UBC irregularity
criterion.
5 5
regular regular r20 (1-2,4-5)
r20 (2-5) r40 (1-2,4-5) r60 (1-2,4-5)
r40 (2-5) r20 (3) r40 (3)
4 4
r60 (2-5) r60 (3)
r20 (1)
Floor
2 2
1 1
0 0.5 1 1.5 0 0.5 1 1.5
interstorey drift [%] interstorey drift [%]
When the overstrength is assigned to a limited number of storeys, the weak storey effect and
all the consequent response anomalies, are attenuated. This is confirmed by the results shown
in Figure 8, regarding the 5 storey frames: the weak storey behaviour shown at the first storey
by the frame “r60(2-5)” is much more evident than for the case “r60(3-5)”, characterised by
overstrength from the third to the fifth storey. In this last case, the interstorey drifts
increments obtained with respect to the regular case at the first and second level are clearly
smaller. The effect, as shown in the figure, is definitively attenuated in the case “r60(4-5)”
where the overstrength is only assigned at the fourth and fifth storey. Identical comments can
be done comparing the frames “r20”, even though in this case the limited assigned
overstrength (20%) leads to smaller variations of the response with respect to the reference
case.
5 5
regular regular
r20(2-5) r60(2-5)
4 r20(3-5) 4 r60(3-5)
r20(4-5) r60(4-5)
r20(5) r60(5)
Floor
Floor
3 3
2 2
1 1
0 0.5 Interstorey drift [%] 1 1.5 0 0.5 Interstorey drift [%] 1 1.5
Floor
5 5
4 4
3 3
2 2
1 1
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
Interstorey drift [%] Interstorey drift [%]
REFERENCES
1. Moehle, J.P. Seismic response of vertically irregular structures. Journal of Structural
Engineering ASCE, 110: 2002-2014: 1984.
2. Moehle, J.P., Alarcon, L.F. Seismic analysis methods for irregular buildings. Journal of
Structural Engineering 112: 35-52: 1986.
3. Al-Ali, A.A.K., Krawinkler, H. Effects of vertical irregularities on seismic behavior of
building structures, Dept. of Civil and Env. Eng., Stanford, Rep. No.130: 1998.
4. CEN. Eurocode 8. Design provisions for earthquake resistance of structures: 1994.
5. Uniform Building Code. Volume 2. Structural engineering design provisions: 1997.
6. Valmundsson, E.V., Nau, J.M. Seismic response of building frames with vertical
structural irregularities. Journal of Structural Engineering: 123 (1): 1997.
7. Magliulo, G., Realfonzo, R., Ramasco, R. Sul comportamento sismico di telai piani in
c.a. caratterizzati da irregolarità in elevazione. X Convegno ANIDIS: 2001 (in italian).