E.

Cosenza (ed), Eurocode 8 Perspectives from the Italian Standpoint Workshop, 109-117, © 2009 Doppiavoce, Napoli, Italy

ASYMMETRIC-PLAN BUILDINGS:
IRREGULARITY LEVELS AND NONLINEAR SEISMIC RESPONSE

Andrea Lucchini a, Giorgio Monti b, Enrico Spacone c

a
Sapienza Università di Roma, Roma, Italy, andrea.lucchini@uniroma1.it
b
Sapienza Università di Roma, Roma, Italy, giorgio.monti@uniroma1.it
c
Università degli Studi G. D’Annunzio, Pescara, Italy, espacone@unich.it

ABSTRACT
Results from both linear and nonlinear dynamic analyses exploring the seismic torsional
behaviour of a multi-storey RC frame structure asymmetric in plan and regular in elevation
are reported. Based on the obtained findings, the conceptual gaps in the current prescriptions
given by the OPCM3431 and the EC8 on the evaluation method for the irregularity level
produced by the mass and stiffness asymmetry and on the definition of the range of
applicability of the linear analyses methods are identified. Comments on the code
prescriptions relative to the number of accelerograms needed in the nonlinear time-history
analyses and on the proposed method for modifying the seismic response obtained with a
pushover analysis in order to account for the torsional behaviour of asymmetric-plan
buildings are also presented. Possible proposals for a code provision are finally suggested.

KEYWORDS
Asymmetric-plan, irregularity level, evaluation, nonlinear dynamic response.

1 INTRODUCTION

Asymmetric-plan buildings, namely buildings with in-plan asymmetric mass and strength
distributions, are systems characterized by a coupled torsional-translational seismic response.
This particular behaviour under earthquake excitation usually leads to classify such type of
buildings as irregular systems. Difficulties related to the study of their seismic response can
be identified as follows: first, the identification and measure of the irregularity level produced
by the asymmetry; second, the selection of the appropriate linear or nonlinear analysis method
that can be used to evaluate the seismic demand; finally, the improvement and the extension
of the pushover methods, originally proposed for regular structures, to the case of systems
characterized by torsional seismic behaviours. The prescriptions on these issues proposed by
the Italian code OPCM3431 “Technical Code for Seismic Design, Assessment and Retrofit of
Buildings” (2005) are basically the same as those given in EC8 part 3 “Assessment and
Retrofitting of Buildings” (2004). More specifically, as for the identification of the
asymmetry, only general and qualitative rules are given by OPCM3431, which classifies
buildings as regular only if both lateral stiffness and mass distributions are “approximately
symmetrical” in plan with respect to two orthogonal axes. Furthermore, EC8 prescribes a
limit on the value, both, of the structural eccentricity e0 between the centre of mass and the
centre of stiffness, and of the ratio between the torsional radius r and the radius of gyration l
110 A. Lucchini, G. Monti, E. Spacone

of the floor mass in plan. As for the applicability of the linear analysis methods, both
OPCM3431 and EC8 prescribe that the strengths should be distributed so as to produce
uniform inelastic demands in the resisting elements of the structure. An estimation of the
inelastic demand can be obtained by evaluating, for all the primary elements of the building,
the ratio ρi between the bending moment demand calculated from a linear analysis and the
corresponding capacity. In the evaluation of such uniformity, EC8 includes all yielded
elements (with ρi values greater than 1), while OPCM3431 only those that go well into the
nonlinear range (with ρi values greater than 2). It is important to note that this applicability
condition for linear analysis methods applies to all buildings, regardless of the shape of their
plan; as a matter of fact, no specific requirements for the case of asymmetric-plan buildings
are given by either code.
Finally, on the use of pushover methods, while OPCM3431 does not apply any kind of
restriction to asymmetric-plan buildings, EC8 prescribes to increase the displacements of the
stiff/strong side obtained with a pushover by amplification factors based on the results of an
elastic analysis. This prescription was supported by studies reported in Peruš and Fajfar
(2005), Marušić and Fajfar (2005), and Fajfar (2005), which demonstrated that the elastic
amplification of the displacements at the edges of the plan with respect to the displacement at
the centre of mass can be used as a rough estimate also in the inelastic range.
The objective of the present paper is to show, through the analysis of a selected case study,
the conceptual gaps in the current prescriptions given in OPCM3431 and EC8 on asymmetric-
plan buildings. Suggestions for possible improvements of the identification and evaluation
methods are also presented.

2 CASE STUDY

The case study selected for the investigation is the two-storey RC building shown in Figure 1,
comprising a shear-beam frame structure with rigid diaphragms at each floor. The system is
regular in elevation and asymmetric in plan because of the eccentricity in the x-direction at
each floor between the centre of stiffness CS, located at the geometrical centre of the plan,
and the centre of mass CM. Periods and participating mass ratios of the elastic modes of the
system, whose shapes are shown in Figure 2, are reported in Table 1. Because of the
asymmetry in the y-direction only, the first and third modes are coupled, while the second one
is pure translational in the x-direction.

CS
CM x

Figure 1. Studied two-storey frame RC structure: 3D view and plan.

Asymmetric-Plan Buildings: Irregularity Levels and Nonlinear Seismic Response 111

The strengths of the resisting elements are designed for the seismic load combination so as to
generate a uniform ρi distribution, characterized by a ratio between the maximum and
minimum values ρmax and ρmin lower than the limit of 2.5 prescribed by OPCM3431. More
specifically, the two ρmax/ρmin values corresponding to earthquakes exciting the structure in the
x and y directions are equal to 1.89 and 1.73, respectively.

Mode 1 Mode 2 Mode 3

Figure 2. Deformed shapes at the roof of the first three modes of the structure.

Table 1. Periods and modal participating mass ratios.

Mode 1 Mode 2 Mode 3
T [s] 0.70 0.62 0.56
Mux [%] 0 93 0
Muy [%] 52 0 41
Mrz [%] 78 0 15

Table 2. Natural accelerograms selected for the dynamic analyses.

Primary Secondary
Earthquake Date Station component component
PGA [g] PGA [g]
Friuli 1976-05-06 Tolmezzo-Diga Ambiesta 0.34 0.30
Montenegro 1979-04-15 Ulcinj-Hotel Albatros 0.35 0.28
Campano Lucano 1980-11-23 Bagnoli-Irpino 0.28 0.21
Campano Lucano 1980-11-23 Sturno 0.31 0.21
Umbria-Marche 1997-09-26 Nocera Umbra 0.51 0.45
Izmit 1999-08-17 Izmit-Meteoroloji Istasyonu 0.32 0.23
Duzce 1999-11-12 LDEO Station n° C0375 VO 0.74 0.48

In the nonlinear analyses, performed to evaluate the inelastic response of the structure, the
resisting elements are modelled as beams with plastic hinges at the ends, with elasto-plastic
hysteretic behaviours. The yielding bending moment values of the column hinges in the x and
y directions are considered as non-interacting between them and with the axial force.
In both linear and nonlinear analyses, a Raleigh model is used for the viscous damping
matrix, whose mass and tangent stiffness matrices coefficients are set to have modal damping
ratios equal to 5% at the first and third period of the elastic system. The acceleration time-
histories selected for the dynamic analyses are the seven unscaled natural accelerograms
112 A. Lucchini, G. Monti, E. Spacone

defined in Iervolino et al. (2007), compatible with the response spectrum type A with PGA
intensity equal to 0.35g. The main properties of the records are briefly reported in Table 2.

3 ANALYSES RESULTS

On the selected case study two sets of analyses are carried out. First, linear modal response
spectrum analyses are used to evaluate the coupling level between the rotational and
translational response of the system produced by the in-plan asymmetry. Then, linear and
nonlinear time-history analyses are used to study the seismic behaviour evolution from elastic
to inelastic range, and also to evaluate the response sensitivity to the exciting accelerograms.

3.1 Response spectrum analyses

In general, the asymmetric-plan building irregularity results in the coupling of the torsio-
translational behaviour, producing non-uniform seismic demands in the resisting elements of
the system. Such coupling depends, in turn, by the significance of the higher modes
contribution: in fact, it can be easily observed that, with the increase of the coupling, more
than one mode is needed to excite the translational mass of the system.
However, it is worth noticing that, on the basis of this evidence, the irregularity in the demand
does not depend solely on the properties of the building, as indicated both in OPCM3431 and
EC8, but also on the properties of the seismic excitation. The single mode contribution, in
fact, depends both on the associated participating mass ratio and on the frequency content of
the seismic excitation as well.
This can be easily seen in the plot of Figure 3, where the OPCM3431 response spectra of
three different ground types are shown. Looking at the two periods T2 and T3 of the coupled
modes of the studied building, it is observed that for two different spectra, e.g., type D and
type A, the relative contribution of the modes is different: for type D the two modes have the
same contribution, while for type A the two modes have different contributions.

3.5
Ground type A
3 Ground type B,C,E
2.5
Ground type D
Se/PGA

1.5

1
T3 T2
0.5

0
0 0.5 1 1.5 2 2.5 3

T [s]
Figure 3. OPCM3431 response spectra with the periods of the coupled modes of the studied building.
Asymmetric-Plan Buildings: Irregularity Levels and Nonlinear Seismic Response 113

One way for estimating the importance of the higher modes in global terms could be to
evaluate the ratio between the seismic base shear associated to the first coupled torsional-
translational mode in one direction and the total seismic base shear acting in the same
direction.
Thus, a simple modal response spectrum analysis would suffice to measure the irregularity
level of the building, while weighing the seismic excitation effects on the asymmetry of mass
and stiffness distributions.
In the studied case, values of the base shear ratio ranging between 65% and 75%, depending
on the ground type considered, are obtained.

3.2 Time-history analyses

According to EC8, if ρmax/ρmin is lower than a certain value (2.5 for OPCM3431), then the
distribution of the inelastic demands, as measured by the ρi, is considered as uniform. This
allows to use a linear analysis to evaluate the seismic demand in the resisting elements of the
building. In this prescription, there is the implicit assumption that, if the inelastic demands are
uniformly distributed, then the building behaves regularly, in the sense that the elastic
response shape does not significantly change when the building enters the nonlinear range.
This implies that the seismic demand evaluation (in terms of deformations) obtained from the
linear analysis can be considered as representative of that in the nonlinear range.
One of the peculiarities of the seismic response of asymmetric-plan buildings is that the
maximum demand is not attained at the same time in all the elements of the structure: this is
due to the higher modes effects, that is, to the torsio-translational behaviour characterizing
their seismic motion. For such systems, even if their ρi distribution is uniform, the structure
does not enter the nonlinear range in a uniform fashion. In such case, it can be seen that the
seismic response cannot be identified as regular. This is the case of the investigated
asymmetric-plan building, which, notwithstanding its ρmax/ρmin largely lower than 2.5, shows a
seismic response shape in the nonlinear range different from the linear one.
In Figure 4, the response of the building to the Montenegro earthquake (Yugoslavia, April 15,
Ulcinj-Hotel Albatros station) is shown. In the plots, the envelopes of the maximum
displacements in the direction of asymmetry (the y-direction) of the four frames at the second
floor of the structure are reported. More specifically, in the two plots the absolute
displacements uy and the normalised values with respect to the displacement at the centre of
mass uy, CM are shown, respectively. In addition to the nonlinear displacements, in each plot
the elastic response of the building to the same accelerogram is also reported for comparison.

0.2 1.4

0.16
1.2
uy [uy, CM]

0.12
uy [m]

1
0.08
TH NL TH NL
TH L 0.8 TH L
0.04

0 0.6
0 6 12 18 0 6 12 18
x [m] x [m]

Figure 4. Response of the system to the Montenegro earthquake: envelopes of the roof maximum
displacements in the y-direction obtained from nonlinear (TH NL) and linear (TH L) analyses, shown as
absolute values (left) and normalised with respect to the centre of mass displacement (right).
114 A. Lucchini, G. Monti, E. Spacone

1.6 1.4

1.4
1.2
uy [u y, CM]

uy [uy, CM]
1.2
1
1
0.8 TH NL
0.8 TH NL
TH L
TH L
0.6 0.6
0 6 12 18 0 6 x [m] 12 18
x [m]

Figure 5. Comparison of the responses under different earthquakes normalised with respect to the centre
of mass displacement: Campano Lucano (left) and Umbria Marche (right).

On the basis of these results, general observations on the behaviour of asymmetric-plan

buildings can be done. First, in some cases the inelastic deformations of the system can result
higher than the elastic ones, meaning that the linear analysis can underestimate the actual
inelastic demands in the elements of the structure. Second, the shape of the system response
changes between linear and nonlinear range. The elastic one, in fact, is usually more affected
by in-plan rotations (see, for example, Lucchini et al., 2009): the more curved envelope shows
amplifications with respect to the displacements at the centre of mass on both the rigid and
flexible side of the plan, which are usually absent in the nonlinear response.
From such considerations, it can be concluded that, for asymmetric-plan buildings, if the
inelastic demand is estimated as significant (e.g., as proposed by OPCM3431, if the ρi are
greater than 2), regardless its distribution (as measured by ρmax/ρmin), a nonlinear analysis
should always be used to evaluate the seismic demand.
In Figure 5, the envelopes of the maximum displacements produced by the Campano Lucano
and the Umbria Marche earthquakes, respectively, are reported. From the comparison
between the two curves the dependence of the shape of the system response on the selected
exciting accelerogram can be clearly noticed: on the flexible side of the plan, in fact, either
amplifications or deamplifications can be experienced by the system in the nonlinear range.
Such dependence can be also observed in the plot of Figure 6 that reports, both for the linear
and the nonlinear response (denoted with solid and dashed lines, respectively), the mean and
the mean plus/minus one standard deviation values of the normalised maximum
displacements (which shows the displacement shape), evaluated in the y-direction for all the
seven earthquakes in Table 2.

1.6 1.6

1.4 1.4
uy [uy, CM]

uy [uy, CM]

1.2 1.2

1 1

0.8 0.8

0.6 0.6
0 6 12 18 0 6 12 18
x [m] x [m]
Figure 6. Mean and mean plus/minus one standard deviation values of the normalised maximum
displacements in the y-direction evaluated for the seven earthquakes in Table 2,
as obtained from linear (left) and nonlinear (right) analyses.
Asymmetric-Plan Buildings: Irregularity Levels and Nonlinear Seismic Response 115

Even if the response shape tends to show less variability in the nonlinear range – i.e., its
standard deviations decreases – however, it still depends on the excitation characteristics: it is
important to notice, in fact, that for regular buildings characterized by pure translational
behaviours the standard deviation of the normalised displacements is always equal to zero, as
the in-plan response shape does not depend on the excitation.
Based on this finding, it can be stated that when nonlinear time-history analyses are used to
evaluate the seismic demand of asymmetric-plan buildings, because of the significant
sensitivity of the response shape to the selected record, a larger number of accelerograms than
that requested for regular buildings should be prescribed by the code.
Observing the plot of Figure 6, another interesting result can be found: the mean value of the
normalised displacement evaluated with the linear analysis is always higher than the nonlinear
one. This means that the inelastic displacements amplifications at the edges of the plan can be
conservatively approximated by the elastic ones. This finding actually confirms the EC8
prescription that requests to increase the displacement of the stiff/strong side obtained with a
pushover with an amplification factor based on the results of an elastic analysis. However,
based on this code prescription, while the displacements are clearly obtained through the
amplification factors, it is not clear how other quantities of interest (e.g., element forces or
chord rotations), nonlinearly dependent on the displacement, can be obtained.

4 POSSIBLE PROPOSALS FOR A CODE PROVISION

Dynamic analyses carried out on a selected case study of a mass-asymmetric multi-storey

frame building led the authors to highlight conceptual gaps of both OPCM3431 and EC8
seismic codes in the following general issues related to the evaluation of asymmetric-plan
buildings:
• the estimation of the irregularity level of the seismic behaviour of the building;
• the use of the ρi uniformity as a parameter evaluating the regularity in the resistance
distribution and identifying the range of applicability of linear analyses;
• the evaluation of the number of accelerograms to be used in nonlinear time-history
analyses;
• the modification procedure of the pushover method to account for the torsional
behaviour.
Based on the results obtained in the investigations, the following suggestions are proposed by
the authors for improving the code prescriptions:
• the irregularity produced by the asymmetric elastic properties of the building could be
measured by evaluating the ratio between the contribution Vb,1 given by the first
coupled mode in one direction and the total seismic shear Vb,tot acting at the base of the
structure in the same direction; this parameter, which can be calculated with a simple
modal response spectrum analysis, can measure both the coupling level of the building
behaviour and the influence of the earthquake properties on the significance of the
higher modes effects;
• when significant nonlinearity is expected (e.g., as proposed by OPCM3431, if ρi ≥ 2),
regardless the uniformity of the estimated inelastic demand distribution (as measured
by ρmax/ρmin), the building should be evaluated with a nonlinear analysis; apart from
the regularity of the resistance distribution with respect to the seismic demand, mass
and stiffness asymmetric configurations produce an irregular response, namely an
irregular evolution of the nonlinear damage that cannot be evaluated with a simple
linear analysis;
116 A. Lucchini, G. Monti, E. Spacone

• the minimum number of accelerograms requested in the nonlinear time-history

analyses should be larger than that prescribed by the code for regular symmetric-plan
buildings; in fact, unlike symmetric systems, those asymmetric in plan show a
response shape significantly sensitive to the exciting record; however, the authors
believe that further investigations are still needed to define such minimum number;
• more exhaustive explanations on how to modify the pushover analysis results based on
the elastic analyses should be given by the code when dealing with response quantities
that do not depend linearly on the displacements (e.g., the element forces or the chord
rotations) need to be evaluated.

A possible procedure for selecting the analysis type based on the first two points listed above,
is shown as a flow-chart in Figure 7.

ASYMMETRIC-PLAN BUILDINGS

Low expected
Inelastic Demand

Yes ρi < 2 No

L
Low contribution
of the Higher Modes

Yes Vb,1 / Vb,tot > 75% No

NL
Uniformity of the
Inelastic Demand

Yes ρmax / ρmin < 2.5 No

L NL

Figure 7. Proposal of requirements for selecting linear (L) or nonlinear (NL) analysis methods
in the evaluation of asymmetric-plan buildings.

5 ACKNOWLEDGEMENTS

This work has been partially carried out under the program “Dipartimento di Protezione
Civile – Consorzio Reluis”, signed on 2005-07-11 (n. 540), Research Line 2, whose financial
support is greatly appreciated.
Asymmetric-Plan Buildings: Irregularity Levels and Nonlinear Seismic Response 117

6 REFERENCES

European Committee for Standardization (2004). “Assessment and Retrofitting of Buildings”, Eurocode
8 Part3, prEN 1998-3.
Fajfar P., Marušić D., Peruš I. (2005). “The extension of the N2 method to asymmetric buildings”,
Proceedings of the 4th European Workshop on the Seismic Behaviour of Irregular and Complex
Structures, Thessaloniki, Greece.
Marušić D., Fajfar P. (2005). “On the inelastic seismic response of asymmetric buildings under bi-axial
excitation.” Earthquake Engineering and Structural Dynamics, 34(8), pp. 943-963.
Iervolino I., Maddaloni G., Cosenza E. (2007). “Accelerogrammi naturali compatibili con le specifiche
dell’OPCM 3431 per l’analisi sismica delle strutture”, XII Convegno Nazionale L’Ingegneria Sismica
in Italia, Pisa, Italy.
Lucchini, A., Monti, G., Kunnath, S. (2009). “Seismic behavior of single-story asymmetric-plan
buildings under uniaxial excitation”. J. of Earthquake Engineering and Structural Dynamics, John
Wiley, New York, (in press).
Peruš I., Fajfar P. (2005). “On the inelastic torsional response of single-storey structures under bi-axial
excitation”, Earthquake Engineering and Structural Dynamics, 34(8), pp. 931-941.
Presidente del Consiglio dei Ministri (2005). “Norme Tecniche per il Progetto, la Valutazione e
l’Adeguamento Sismico degli Edifici”, OPCM 3431.