You are on page 1of 25

Bulletin of Earthquake Engineering 1: 371395, 2003.

2004 Kluwer Academic Publishers. Printed in the Netherlands.


371
Seismic Vulnerability Assessment of Gravity Load
Designed R/C Frames
ANGELO MASI
Department of Structures, Geotechnics and Geology applied to Engineering, Faculty of
Engineering, University of Basilicata, Potenza, Italy
(Tel:+39 0971 205061; Fax +39 0971 205070; E-mail: masi@unibas.it)
Received 6 april 2003; accepted 10 October 2003
Abstract. The seismic vulnerability of some frame structures, typical of existing Reinforced Con-
crete buildings designed only to vertical loads, has been evaluated. They are representative of building
types widely present in the Italian building stock of the last 30 years. A simulated design of the
structures has been made with reference to the codes in force, the available handbooks and the
current practice at the time of construction. The seismic response is calculated through non linear
dynamic analyses with articial and natural accelerograms. Three main types have been examined:
bare frames, regularly inlled frames and pilotis frames. The results show a high vulnerability for the
pilotis buildings: they can be assigned to the class B of the European Macroseismic Scale of 1998
(EMS98). On the contrary, a low vulnerability (class D of EMS98) can be attributed to the regularly
inlled buildings: in this case collapse can be considered unlikely also with strong earthquakes.
An intermediate seismic behavior is shown by buildings without inlls, whose vulnerability can be
placed between the classes B and C of EMS98.
Key words: assessment, existing buildings, frames, masonry inlls, reinforced concrete, seismic
vulnerability, simulated design
1. Introduction
Reinforced Concrete (R/C) buildings make up an increasing proportion of the
building stock of many countries all over the world. In Italy and in other earthquake-
prone Mediterranean countries they currently represent about 50% of the total.
Many of them were built before the advent of seismic codes or with the utilization
of old and inadequate antiseismic design criteria. During past earthquakes (South-
ern Italy 1980, USA 1994, Japan 1995, Turkey 1999, Greece 1999, Taiwan 2001)
R/C buildings often displayed unsatisfactory seismic behaviour, particularly when
their design included only vertical loads and ductile detailing was not explicitly
provided. Thus, the evaluation of seismic vulnerability of R/C building structures
has a key role in the determination and reduction of earthquake impact.
Although the most dangerous consequences of earthquakes in the near future are
likely to come from existing buildings, until now the attention of the community in
its various components (researchers, professionals, policy makers, . . . ) has largely
372 ANGELO MASI
focused on the seismic design of new buildings (Fardis, 1998). There are a number
of socio-economic reasons for this, including the high costs frequently requested
by the retrot of R/C buildings and the lack of community concern with safety in
general, not only in the seismic eld, as experience in other elds corroborates (see
for example safety on job sites).
Along with socio-economic factors, there are also major technical problems.
With the exception of some guidelines in Japan (JBDPA, 1977) and USA (FEMA,
1992), there is little guidance in the codes for the determination of the seismic
resistance of R/C existing buildings. In Europe, the codes are just beginning to
tackle the problem (see for instance Eurocode 8 - part 3 (CEN, 2002a)).
There is no doubt that the assessment of an existing R/Cbuilding is a much more
difcult task than the design of a new one, because it requires work on structures of
which only a limited knowledge can be obtained. There are difculties in determ-
ining possible deterioration conditions as well as in obtaining sufciently accurate
knowledge of some structural data (e.g. amount and location of reinforcement)
or completeness (e.g. materials strength), as appropriate technical documentation
is rarely available. The specicity of the problem calls for the setting up of ad
hoc methods to assess the seismic vulnerability of R/C buildings in a sufciently
reliable, as well as inexpensive way.
A crucial aspect of knowledge of the structural characteristics of a building is
the period of its construction. When technical documentation is either not available
or insufcient, most valuable data can be obtained through reference to the codes,
the design methods and the typical current practice at the time of construction.
Using all the information obtainable from the sources above, a group of structural
characteristics typical of the buildings of a certain period and of a certain region
can be determined. Specically, through an examination of the codes in force, spe-
cications on the prescribed values of loads and material strengths, the minimum
values of the dimensions of structural elements and of reinforcement amounts can
be drawn up. More difculties occur for the evaluation of the values of internal
forces actually used in the safety verications, of the location of reinforcement
and of the detailing solutions. For this reason, reference has to be made to the
handbooks commonly adopted in the period and to the technical documentation
of real buildings found in the archives of public administrations, building rms
and professional ofces. From the handbooks more accurate indications can be
obtained regarding the design methods and the arrangement of reinforcement in the
structural elements. Finally, the technical documentation of real buildings enables
verication of the reliability of the data obtained from codes and handbooks, as it
shows the design and construction rules actually adopted in practice.
The present study aims at evaluating the seismic vulnerability of some framed
structural types representative of post-1970 R/C existing buildings designed only
to vertical loads. To obtain a realistic evaluation of their vulnerability, a great deal
of work has been carried out to determine the most important structural char-
acteristics of these buildings including a thorough examination of the technical
SEISMIC VULNERABILITY ASSESSMENT 373
documentation of the period. It is worth noting the great importance of this matter,
for example in using in the European context the above cited approaches (JB-
DPA, 1977; FEMA, 1992) or a now available powerful approach based on the
software package HAZUS (FEMA-NIBS, 1999) developed in USA. Performance
evaluations reported in these approaches cannot be directly adopted in other coun-
tries, as they are relevant to specic R/C style constructions (Kappos, 2001) and
are sometimes based on empirical formulas calibrated on observed behaviour and
damage data from local earthquakes.
2. Procedure
The evaluation of the seismic vulnerability of gravity-load designed R/C buildings
has been carried out on the basis of a purposely set up procedure made up of ve
main steps (Masi, 2000; Masi et al., 2001a):
(1) some congurations of R/C buildings, typical of the period under examination,
are selected;
(2) the structures are designed taking into account only vertical loads, on the basis
of the codes in force, of the available handbooks and of the current practice of
the period;
(3) the seismic response is calculated through non linear dynamic analyses with
articial and natural accelerograms;
(4) the seismic resistance is evaluated by means of fragility curves relevant to some
structural and non-structural damage parameters (drift, ductility demands, etc.);
(5) the vulnerability class of each type, according to the European Macroseismic
Scale 1998 (ESC, 1998), is dened taking into account the increasing damage
degrees computed by applying increasing seismic intensities.
2.1. STRUCTURAL TYPE SELECTION
The review of the technical documentation reveals that moment resisting frames
are the most widespread structural type. The majority are medium-rise buildings
particularly in the 35 storey range, with internal beams spanning in one direction
only. In the other direction the horizontal stiffness of the vertical resisting elements
is more or less symmetrically distributed.
Based on these observations a reference structure has been dened. It is a four
storey plane frame, with two bays of 5.0 m span. Interstorey height is equal to
3.0 m. Starting from this reference system, three frame types have been dened on
the basis of the position in the building (exterior and interior frames) and of the
beam stiffness (rigid beams, exible beams and no beams):
(1) RB frame: exterior frame with Rigid Beams (0.30 m 0.50 m);
(2) FB frame: exterior frame with Flexible Beams (0.70 m 0.22 m);
(3) NB frame: interior frame with No Beams.
Typically observed dimensions of the beams have been adopted, whereas, in
case of NB frames, frequently present as interior frames in vertical load designed
374 ANGELO MASI
Figure 1. Layout of frame types.
buildings, a strip of the tile lintel oor diaphragm connecting the columns is con-
sidered. The strip width has been computed on the basis of the torsional and exural
stiffness of structural elements around the columns and is equal to about 1.0 m.
In the exterior frames, the presence and position of inll masonry walls have
also been considered, thus obtaining the types BF (Bare Frame), IF (Inlled Frame)
and PF (Pilotis Frame). In conclusion, seven frame types have been dened (Fig-
ure 1).
BF-RB and BF-FB types are representative of exterior frames without inlls.
While inlls are generally present, they sometimes have many and/or very large
openings or they are badly connected to the structure and thus their contribution
to the strength and stiffness of the structure can be neglected. On the contrary,
IF and PF types are relevant to buildings where inlls are well conned by the
surrounding frames and have good mechanical and construction characteristics
without large openings: IF type is representative of frames with regularly arranged
masonry inlls, while PF type is representative of frames without masonry inlls
at the ground oor (soft storey).
SEISMIC VULNERABILITY ASSESSMENT 375
2.2. STRUCTURAL DESIGN OF MODELS
A simulated design has been carried out, where internal force values have been
computed on the basis of the characteristic values of dead and live loads. Live loads
have been assumed equal to 2.0 kN/m
2
, as prescribed for residential buildings.
Safety verications have been performed according to the allowable stress method,
assuming mechanical properties of materials according to post-1970 standards, i.e.,
medium quality concrete (R
ck
250) with strength close to the C20/25 strength and
deformed steel (FeB 38K) with grade close to S400 type.
The columns have been designed taking into account only axial load and ad-
opting the minimum requirements provided in the Italian code of the period (D.M.
LL.PP., 1972) regarding reinforcement. The beams have been designed on the basis
of the simplied model of continuous beam resting on simple supports. Where
the dimensions of structural elements and the reinforcement amounts could not
be inferred by code prescriptions or by internal forces values, reference has been
made to the most prominent handbooks (Santarella, 1968; Pagano, 1968) and to the
current practice of the period. In this context an examination of local characteristics
reveals many deciencies generally due to a lack of attention to detailing: low
amounts of longitudinal and, mainly, of transverse reinforcement with respect to
the correspondent anti-seismic frames, are commonly present.
Beam reinforcement is constant along the height with percentages equal to
about 0.5% in the Rigid Beams and in the range 0.550.9% in the Flexible Beams.
Shear reinforcement is formed by 45

inclined steel bars resulting from the re-


inforcement arranged according to the bending moment diagram, and adding the
hoops requested by either shear values or minimum code requirements.
In the types relevant to external frames (RB, FB) the columns always have
square section with 0.30 m dimension. The same occurs in the type relevant to
interior frames (NB), with the exception of the middle column which has 0.30
0.40 m and 0.30 0.35 m dimensions, respectively at I and II oor level. All
the columns have four reinforcement bars with diameter equal to 16 mm, 14 mm
and 12 mm at I, II and III-IV oor levels, respectively. Percentage values are in
the 0.500.68% range. Transverse reinforcement is made up of 6 mm hoops with
constant spacing equal to 0.15 m.
2.3. NON LINEAR RESPONSE EVALUATION
The seismic response has been evaluated by means of inelastic time-history ana-
lyses. Models able to represent the selected structural types in a sufciently reliable
way, both in terms of stiffness and resistance, have been implemented by using the
DRAIN-2D+ code (Tsai, 1994), in a purposely upgraded version.
As regards the actions, only a limited share of the live loads has been considered
in the evaluation of the seismic response. According to (CEN, 1993, 1994a), the
376 ANGELO MASI
following values of the vertical loads for the static situation (V
d
) and for the calcu-
lation of the seismic effects (S
d
) have been, respectively, assumed:
V
d
= G
k
+0.3 Q
k
S
d
= G
k
+0.3 0.85 Q
k
where G
k
and Q
k
are the characteristic values of the dead and live loads.
In addition to the energy dissipation due to hysteresis, viscous damping has
also been considered, by assuming the mass-proportional damping and stiffness-
proportional damping constants, and , function of the frequencies of vibration
and of the equivalent damping ratios of the 1
st
and 3
rd
mode. The values of damping
ratios,
1
and
3
, are taken equal to 2%, with the exception of the frames without
effective inlls (BF types) where, to take into account the presence of inlls not
effective in sustaining seismic loading but able to dissipate some energy,
1
and
3
have been assumed equal to 5% (Chopra, 2001).
2.3.1. Modeling of R/C members
The hysteretic modeling of R/C members has been carried out by taking into
account that the structures being examined are relevant to buildings with medium-
good construction characteristics. Moreover, R/C members are rather slender, have
low amounts of longitudinal reinforcement and are reinforced with deformed bars.
In all members the shear strength is always larger than the maximum shear force,
computed also on the basis of the maximum yield moments at the ends of the
member. Anchorage lengths typically available in the members of the examined
buildings (Santarella, 1968) are able to prevent bond failure (Masi, 2001a). As a
consequence, the most probable failure mechanism will be a exural one, because
shear or bond failures, even if possible, appear unlikely (Cosenza et al., 1999).
Thus, a hysteresis model for exure-dominated members has been considered.
Specically, for both beams and columns, a Takeda-type model has been used,
due to its ability to provide simple and numerically stable as well as sufciently
realistic hysteretic rules. The degrading stiffness properties are modelled through
the parameters
1
and
2
(Figure 2), controlling the amount of degradation and
energy dissipation during cyclic loading. Values for R/C members are normally in
the range 00.4 and 00.6 for
1
and
2
, respectively. In the analyses, considering
the degrading properties of the members under study,
1
= 0.2 and
2
= 0 have
been assumed.
To account for the effects of cracking due to non-seismic actions, the initial
stiffness of R/C members has been taken equal to 70% of the uncracked gross
section.
For material strength, realistic values have to be considered in the evaluation
of seismic resistance, therefore the mean rather than the characteristic values have
been used in the computation of the member strength capacities. A mean strength
f
cm
= 28 MPa has been assumed for the concrete, according to the expression
SEISMIC VULNERABILITY ASSESSMENT 377
Figure 2. Role of
1
and
2
parameters in the shape of the hysteresis cycle.
f
cm
= (f
ck
+ 8) MPa provided in EC2 (CEN, 1991). The mean strength of steel
has been taken equal to f
ym
= 400 MPa. The yield moment has been computed
according to EC2 and considering, in the columns, the effect of the axial force due
to vertical loads.
2.3.2. Modeling of masonry inlls
In R/C existing buildings, inlls are usually made of two layers of hollow brick
masonry with a total thickness equal to about 200 mm and scarce mechanical
characteristics. Taking into account the modeling capabilities of the program used
in the analyses (DRAIN-2D+), each masonry panel has been modelled by using
plane elements having an elastoplastic force-deformation relation dened by the
ultimate resistance, F
y
, the yield displacement,
y
, and the ultimate displacement,

u
.
The total resistance and stiffness capacity of each panel has been dened by
considering an equivalent diagonal strut, whose area has been determined by mul-
tiplying the panel thickness t
w
by an equivalent width b
w
. The following expression
due to Mainstone (1974), relevant to rectangular masonry panels, has been used to
compute b
w
:
b
w
d
w
= 0.20 sin(2)

E
w
t
w
h
3
w
sin(2)
E
c
I
p

(1)
378 ANGELO MASI
where h
w
= panel height, d
w
= diagonal length, = angle of the diagonal with
the horizontal, E
w
=modulus of elasticity of masonry, E
c
=modulus of elasticity
of concrete and I
p
= moment of inertia of columns.
The characteristics of the equivalent strut are the area A
s
=b
w
t
w
, the stiffness
K
s
=(E
w
A
s
/d
w
) and the ultimate resistance F
us
= A
s
f
w
. Atypical value has been
assumed for the compressive strength of masonry f
w
, equal to 1.2 N/mm
2
(D.M.
LL.PP., 1987; Calvi and Recla, 2000). On the basis of the angle of the diagonal
with the horizontal , the horizontal components of the stiffness K
s,h
= K
s
cos
2

and of the ultimate resistance F


us,h
= F
us
cos can be calculated. Finally, the
characteristics of the equivalent shear panel are as follows:
Area A
p
= L
w
t
w
Ultimate Resistance F
yp
= F
us,h
Equivalent Shear Modulus of elasticity G
p,eq
= (K
s,h
h
w
)/A
p
Yield Displacement
y
= F
yp
/K
s,h
Yield Drift
y
/h
w
where L
w
is the panel length equal to the bay width.
As already mentioned, given the capabilities of the code at hand, the degrading
behaviour of masonry inlls has been taken into account through a purposely set
up model (Masi et al., 2001a) which has simple, numerically stable and sufciently
realistic hysteretic rules. Each masonry panel is modelled by means of 4 sub-panels
connected in parallel with a total resistance equal to that of the panel. Each sub-
panel has the same yield displacement
y
, while the ultimate resistance is 3/10 of
the total for three of them and 1/10 for the fourth. The progressive deterioration is
obtained assuming a suitable value of the ultimate displacement for each sub-panel:
all four sub-panels have different displacement ductilities and the weakest sub-
panel has the largest ductility. In Figure 3 the monotonic relation force-deformation
of a panel is shown.
2.4. SEISMIC INPUT
Two sets of input motions have been used in carrying out the inelastic time-history
analyses: articial accelerograms generated from the elastic response spectrum for
site B of EC8 (CEN, 1994b), and natural accelerograms mainly drawn from the
1997 Umbria-Marche (Central Italy) seismic sequence.
Articial accelerograms have a trapezoidal envelope, with a rst part with in-
creasing intensity t
1
secs long, a second part with constant intensity t
2
secs long
and, nally, a third part with decreasing intensity t
3
secs long, where t
1
, t
2
and
t
3
are dependent on the considered PGA value (CEN, 1994b; Min.LL.PP., 1997).
Seven PGA values, in the range 0.050.35 g, have been considered for the articial
seismic input (Table I). In Table I also the average values of the effective duration
t
d
(Trifunac, 1975) and of the Arias Intensity AI (Arias, 1970) are reported for
each seismic intensity. The numerical simulations have been carried out by using
SEISMIC VULNERABILITY ASSESSMENT 379
Figure 3. Force-deformation relationship of a masonry panel as a combination of 4
sub-panels.
Table I. Characteristics of the articial accelerograms used in the
analyses
PGA
[g]
t
1
[sec]
t
2
[sec]
t
3
[sec]
t
t ot
[sec]
t
d
[sec]
AI
[cm/sec]
0.05 2.6 10 2 14.6 11.5 5.7
0.1 2.6 10 2 14.6 11.5 22.7
0.15 3.3 12.5 2.5 18.3 14 60.4
0.2 4 15 3 22 18 141.5
0.25 4.6 17.5 3.5 25.6 21 265.3
0.3 5.3 20 4 29.3 24 413.5
0.35 6 22.5 4.5 33 26.5 664.2
8 articial accelerograms for each intensity. The values of the response parameters
are obtained as averages of the results of the 8 step-by-step analyses.
The sample of natural accelerograms used in the analyses (Masi et al., 2001a)
is made up of more than 20 recordings with PGA values in the range 0.040.58 g.
It is worth noting that, in the range PGA = 0.05-0.35 g, the natural accelerograms
have mean values of AI in the range 595 cm/s, far lower than the correspondent
ones of the articial accelerograms.
380 ANGELO MASI
Figure 4a. Normalised base shear NBS vs PGA/g for articial accelerograms.
3. Non linear dynamic response analyses
The non linear seismic response of models has been analyzed on the basis of the
following parameters:
Normalised Base Shear NBS (base shear divided by the total weight);
Interstorey Drift ID (relative storey displacement divided by the storey height);
Curvature ductility demand.
The ductility demands have been evaluated in the columns (DC) and in the beams
(DB) by considering the maximum values in the frame (DC
max
and DB
max
). The
average values in each storey have also been computed and the maximum values
among the storeys (DC
s,max
and DB
s,max
) have been considered. In the diagrams
of Figures 47 the variation of the maximum values of some response parameters
with the seismic intensity in terms of PGA are shown both for articial and natural
accelerograms.
The results obtained with the natural accelerograms, given their large scatter,
are represented by means of their regression curves referring to the same interval
PGA=0.050.35 g used for the articial accelerograms. Low values of the correl-
ation coefcient R are sometimes obtained (R values in the range 0.30.6), due to
the poor ability of PGA to effectively represent the damage potential of a ground
motion. Better correlations (Masi et al., 2001b) are obtained when the results are
reported in terms of Arias Intensity (R values in the range 0.50.7).
By examining the diagram relevant to the normalised base shear NBS relevant to
articial accelerograms (Figure 4a), three different groups of curves corresponding
to the three main structural types under study (BF, IF and PF), can be clearly seen.
BF types show the lowest values of base shear: the maximum values of NBS are
in the range 0.1050.135 g, going from the type without beams (BF-NB) to the
type with rigid beams (BF-RB). IF types show the highest values of base shear: the
SEISMIC VULNERABILITY ASSESSMENT 381
Figure 4b. Normalised base shear NBS vs PGA/g for natural accelerograms.
maximum values of NBS are in the range 0.360.39 g, being the maximum value
relevant to the frames with rigid beams (IF-RB). Intermediate values are shown by
the PF types: the maximum values are almost the same for frames with rigid and
exible beams and are about equal to 0.18 g. All curves show an increasing trend
and a decreasing gradient with seismic intensity, as the structures approach their
maximum resistances.
The diagram relevant to the natural accelerograms (Figure 4b) shows similar
trends from a qualitative point of view, but the results are lower than those obtained
by applying articial accelerograms. This difference is mainly a consequence of
the far lower values of Arias Intensity (up to 56 times) of natural accelerograms
compared to those of articial accelerograms. In fact, the curves obtained by ap-
plying natural accelerograms are almost coincident with the initial part (PGA up to
0.150.20 g) of the curves relevant to the articial accelerograms.
In the diagrams of the maximum interstorey drifts ID relevant to articial accel-
erograms (Figure 5a), the IF type curves and the BF-NB type curve, represent the
lower and the upper bounds respectively. Only for the highest values of PGA the
curves of PF types show values larger than those of BF-NB type. The maximum
computed values of ID for PF types are equal to about 4% (Figure 5a), values
for which structural and non-structural elements should be completely collapsed.
Also, ID values relevant to moderate-heavy damage states (0.7%1.5%) are already
shown for PGA > 0.1 g. These results clearly highlight how inadequate the seis-
mic behaviour of frames with soft storey is: on one hand they have shown NBS
values up to 4050% larger than frames without effective inlls, on the other hand
they undergo disproportionately large effects with increasing seismic intensity. The
results relevant to IF types corroborates the strong contribution provided by good
quality and regularly arranged inlls in reducing seismic effects: ID shows values
relevant to damage states ranging from slight (ID = 0.2%) up to moderate (ID <
382 ANGELO MASI
Figure 5a. Interstorey drift ID (%) vs PGA/g for articial accelerograms.
Figure 5b. Interstorey drift ID (%) vs PGA/g for natural accelerograms.
0.5%), for PGA values equal to 0.15 g and 0.20 g, respectively, up to a maximum
value equal to about 1.5%.
These considerations are also more or less valid, from a qualitative point of
view, for the results relevant to natural accelerograms (Figure 5b) but the values
are far lower than the articial ones. As already stated, the difference can be ex-
plained by considering the very different values of AI in the two sets of adopted
accelerograms. Being represented by means of their regression curves, in this case
the results show a more regular trend. Consequently, the peculiar behaviour shown
by the pilotis frames under the articial input motion does not appear: their curves
are always below the curves relevant to BF-NB type.
SEISMIC VULNERABILITY ASSESSMENT 383
Figure 6a. Ductility demand on columns DC
s,max
vs PGA/g for articial accelerograms.
Figure 6b. Ductility demand on columns DC
s,max
vs PGA/g for natural accelerograms.
Figures 6a and 6b show the curves of the ductility demands on columns DC
s,max
.
The lowest values are again relevant to the frames with regularly arranged inlls.
For these structural types, the curves relevant to natural accelerograms (Figure 6b)
are almost coincident, with slightly higher values for the IF-RB type. On the con-
trary, large differences can be seen in the case of articial accelerograms (Fig-
ure 6a) with high values of PGA: when PGA = 0.35 g, the maximum values of
DC
s,max
are equal to about 11 and 3, respectively, for IF-RB and IF-FB. Such a
difference can be explained by looking at the different global damage pattern. In
both types, the highest values of DC
s,max
occur at the ground oor. However, with
increasing PGA the inlls on the rst oor suffer damage to a greater extent in
384 ANGELO MASI
Figure 7a. Ductility demand on beams DB
s,max
vs PGA/g for articial accelerograms.
the IF-FB type, making this oor much more deformable than in the IF-RB type.
Due to the lower constraint action of the joints between the ground and rst oor,
lower moment values are detected in the ground oor columns of the IF-FB type.
It should be noted that this behaviour does not occur in the bare frames (BF types):
in this case the frame with exible beams shows signicantly higher values of
ductility demands than the BF-RB type and, furthermore the highest of all types.
Pilotis frames show ductility demands on columns, obtained by applying articial
accelerograms (Figure 6a), which are almost coincident between RB and FB types,
and near to the ones of BF-FB type.
The curves of ductility demands on beams DB
s,max
(Figures 7a and 7b) show
values far higher than the curves of the ductility demands on columns, in all types.
In this case the upper bound is relevant to the bare frame with rigid beams followed
by the bare frame with exible beams. When articial accelerograms are applied,
high values of DB
s,max
can be observed in BF types for PGA > 0.2 g (Figure 7a),
reaching a maximum value equal to 37 in the BF-RB type. This is mainly due to
two reasons:
according to the common practice, in the gravity load dominated design a
very small amount of reinforcing steel is placed (usually only two small-medium
diameter rebars) on the bottom side of beam ends;
for PGA > 0.2 g, moment values due to seismic loads become larger than
negative moment values caused by gravity loads, thus beams yield deeply at their
ends for positive moment.
Low values of DB
s,max
can be observed in PF types, although the lowest values
are again relevant to IF types, corroborating the favourable contribution of regularly
arranged inlls.
SEISMIC VULNERABILITY ASSESSMENT 385
Figure 7b. Ductility demand on beams DB
s,max
vs PGA/g for natural accelerograms.
4. Denition of damage degrees
The evaluation of global damage has been carried out by involving the following
damage variables, which refer to structural and non structural damage:
DC
max
Ductility demand on Columns (max value);
DC
s,max
Ductility demand on Columns (max among the story average values);
DB
max
Ductility demand on Beams (max value);
DB
s,max
Ductility demand on Beams (max among the story average values);
ID Maximum Interstorey Drift (%).
Furthermore, the ductility demands have been compared to the correspondent ulti-
mate values of the ductility capacities, thus obtaining the following ratios:
RDC
max
= DC
max
/ DC
u
(max Ductility Ratio in the Columns);
RDC
s,max
= DC
s,max
/ DC
u
(max Ductility Ratio in the Columns among the story
average values);
RDB
max
= DB
max
/ DB
u
(max Ductility Ratio in the Beams);
RDB
s,max
= DB
s,max
/ DB
u
(max Ductility Ratio in the Beams among the story
average values).
There are several expressions for the evaluation of the ductility capacities of R/C
elements available in the literature. They are drawn from analytical formulations
supported either by the analysis of experimental results or by expert judgement.
Generally, these expressions (e.g., Park and Paulay, 1975; Park and Ang, 1985)
are relevant to R/C structures designed according to antiseismic criteria where,
for example, high connement actions due to transverse reinforcement are avail-
able. Only a few cases are relevant to members without seismic design. Of these,
expressions drawn from statistical elaborations of different experimental investiga-
tions (e.g. Fardis, 1998) or obtained from analytical formulations adopting suitable
386 ANGELO MASI
values of the ultimate deformations of materials (e.g. Penelis and Kappos, 1997;
Calvi and Priestley 1998) are available.
In the present study the latter approach has been used assuming as ultimate
values, respectively,
cu
= 0.005 for the concrete compression strain and
su
=
0.02 for the reinforcement tensile strain. Regarding concrete, reference to poorly or
unconned conditions has been made (Mander et al., 1988; Paulay and Priestley,
1992), while with regard to reinforcement, the deterioration in presence of high
ductility demands suggests that only a limited part of the ultimate tensile strain in
the rebars can actually be exploited (Calvi and Priestley, 1998). By applying this
procedure, ultimate values of the curvature ductility capacities equal to about 30
in the rigid beams and 20 in the exible beams, are obtained. The ultimate values
of the curvature ductility capacities in the columns are equal to about 10 at the
rst storey (value signicant for the PF types) and increase up to about 25 at the
third and fourth storey (value signicant for the BF and IF types). The values of
the ductility ratios in the beams have been computed taking into account that the
beam characteristics remain constant along the frame height. On the contrary, in
the columns the ultimate ductility is strongly inuenced by the axial force and thus
the ductility ratios have to be computed in the storey where this ratio is the larger
one.
Many damage indices proposed in the literature can be used to evaluate the
global damage in a R/C structure. A very comprehensive classication, mainly
aimed at clarifying the different approaches that can be used for dening damage
indices, is reported in (Kappos, 1997). Generally speaking, a damage index is a
quantity ranging from 0 (no damage) to 1 (collapse), whose computation should
involve more than one damage variable. Given the nal objective of this study,
global damage has been described in terms of damage degrees L
d
as dened in
the European Macroseismic Scale (ESC, 1998), ranging from no damage (L
d
= 0)
to complete destruction (L
d
= 5). Following these denitions, the assignment of
damage degrees has been carried out on the basis of suitable values of the selected
damage variables (Table II). The relationship assumed between the damage vari-
ables and the global damage degrees is mainly based on indications drawn from the
literature (Park and Ang, 1985; Naeim, 1989; Kunnath et al., 1990; FEMA-NIBS,
1999; Ghobarah et al., 1999). Due to the good construction characteristics as-
sumed for the structural types under examination, the evaluation of damage degrees
is essentially ductility-based. However, reduced values of some capacity-demand
ductility ratios have been assumed, as the fullest exploitation of the monotonic
deformation capacity of structural members might not occur due to a limited contri-
bution of the cumulative energy absorption on damage. As regards the drift limits,
the peculiar brittle characteristics of masonry inlls have been taken into account.
SEISMIC VULNERABILITY ASSESSMENT 387
Table II. Evaluation of damage degrees L
d
(EMS 98) on the basis of damage variables
L
d
Denition DC
max
RDC
max
RDB
max
DC
s,max
RDC
s,max
RDB
s,max
ID [%]
(EMS98) DB
max
DB
s,max
0 SD = null
NSD = null
<1 <0.1
1 SD = null
NSD = slight
1 <1 0.10.25
2 SD = slight
NSD = moderate
0.10.4 0.20.5 <0.1 <0.2 0.250.5
3 SD = moderate
NSD = heavy
0.40.8 0.51 0.10.4 0.20.5 0.51.0
4 SD = heavy
NSD = very heavy
>0.8 >1 0.40.8 0.51 1.01.5
5 Destruction >0.8 >1 >1.5
(SD = Structural Damage, NSD = Non Structural Damage)
5. Evaluation of damage degrees
The damage degrees resulting from the application of the articial accelerograms
are reported in Figure 8 for all structural types. High values are generally shown:
for PGA = 0.35 g all types are completely collapsed (L
d
= 5), with the exception
of inlled types where a partial collapse is expected (L
d
= 4). IF types show no
or low damage also in the inlls up to PGA values equal to about 0.2 g. The types
relevant to buildings without inlls (BF types) collapse when PGA 0.2 g. The
worst behaviour is shown by the pilotis frames, which would collapse for PGA
0.15 g. A similar high vulnerability is shown by the frames without beams (BF-NB
type). Also for this type, structural damage can be expected already for the lowest
intensity (PGA = 0.05 g) and the damage degrees rapidly increase, leading to a
possible collapse for PGA 0.15 g. Signicant differences do not appear between
rigid beam and exible beam types, particularly in case of inlled frames.
The mean and maximum values of the damage degrees obtained by applying
natural accelerograms are reported in Figure 9. As already stated, in this case dam-
age variable values, given their large scatter, are represented by means of regression
curves referred to the same PGA interval used for the articial accelerograms.
For this reason, besides the values of damage degrees obtained from the curves,
assumed as mean values (L
d
values marked with med in Figure 9), an estimation
of the maximum predictable values has also been made (L
d
values marked with
max in Figure 9) based on the variability of the results.
The exam of the mean values of damage degrees shows much lower values
than those obtained from the articial accelerograms. In fact, also when PGA =
0.35 g, all types are not collapsed (L
d
3), with the exception of PF-FB and BF-
NB types, which are near to partial collapse (L
d
= 34). The maximum values of
388 ANGELO MASI
Figure 8. Damage degrees with EC8-B articial accelerograms.
damage degrees are closer to those obtained by applying articial accelerograms,
but generally higher in the range PGA = 0.050.10 g and lower in the range PGA
= 0.250.35 g. This alternate trend is partly due to the large scatter of the results
obtained with the natural seismic input.
The results in Figure 9 show that the highest damage degrees to be expected in
the inlled frame types are moderate or slight, respectively, for the non structural
and for the structural elements. On the other hand, signicant damage is to be
expected in the PF-FB and BF-NB types already for PGA = 0.05 g, up to a total
collapse when PGA> 0.25 g. Intermediate behaviour can be observed for the other
types. The absence of signicant differences between types with rigid or exible
beams is corroborated. The role of inlls again appears important even if, due to
a general tendency to a reduction of effects, the differences among the types are
smaller than those relevant to the articial accelerograms.
It should be noted that, whereas a comparison of the results based on the same
values of PGA shows damage degrees computed by applying natural accelero-
grams to be remarkably lower than those relevant to articial accelerograms, the
results are almost coincident when the comparison is based on the Arias Intensity,
AI. In fact, articial accelerograms have generally much higher values of AI, but
if similar values of AI are considered in the comparison (e.g. AI

= 60), similar
values of damage degrees are also obtained. This conrms that an energy based
seismic parameter, like the Arias Intensity, can be used in a more effective way
than a peak ground motion parameter (e.g. PGA) to represent the damage potential
of an accelerogram.
SEISMIC VULNERABILITY ASSESSMENT 389
Figure 9. Mean and maximum damage degrees with natural accelerograms.
Table III. PGA values and corrispondent EMS98 intensity degrees
PGA [g] 0.05 0.10 0.15 0.20 0.25 0.30 0.35
I_EMS V VI VII VIIVIII VIII VIII VIIIIX
6. Assessment of seismic vulnerability
The seismic vulnerability of buildings can be assessed through the use of different
techniques. Direct typological and direct mechanical techniques are among those
widely used in Italy (Dolce, 1996). These techniques have been used in this study
for the assignment of the vulnerability class of the building types under consid-
eration, in line with the classication of the EMS98. In this way, the Damage
Probability Matrices provided by Braga et al. (1982), upgraded in (Dolce et al.,
2002), can be used to predict the probability of damage, given a seismic intensity.
Seismic vulnerability has been assigned on the basis of a comparison between
the damage degrees computed with the numerical simulations and those reported
in the EMS98. For this comparison to be carried out, a correlation between PGA
values (used in the numerical simulations) and EMS98 intensities is necessary,
even if it is well known that these correlations have a large inherent scatter. To this
end, assuming the EMS98 intensity degrees (I_EMS) to be substantially coincident
with the intensity degrees of MSK scale, as suggested in the manual of EMS98, the
following empirical relationship due to Margottini et al. (1985) can be used:
I_EMS = (1/0.258) LOG
10
[(PGA/g)(981/2.279)] (2)
In Table III the values of I_EMS related to the PGA values considered are reported.
390 ANGELO MASI
Figure 10. Vulnerability classes for framed R/C buildings according to EMS98.
By examining the denitions of the seismic intensity degrees in the EMS98,
the expected mean and maximum damage degrees can be assigned to each vulner-
ability class, when varying the seismic intensity. In the EMS98 six vulnerability
classes are considered, arranged in decreasing order from A to F, according to
the structural type and the level of earthquake resistant design (Figure 10). For
each case, together with the most likely vulnerability class (mean value) a range
including probable cases and less probable or exceptional cases, is provided. Only
the classes from A to D are herein considered, taking into account that this is
the vulnerability range where the framed R/C buildings designed without seismic
criteria are certainly placed.
In Table IV the mean and maximum values of damage degrees L
d
for each vul-
nerability class are reported. They have been obtained on the basis of the denitions
of the EMS98 intensity degrees in the range from I_EMS = V (strong) to I_EMS
= IX (destructive).
Table IV. Mean and maximum values of damage levels L
d
provided by EMS98 for the vulnerability
classes AD
I_EMS Class A Class B Class C Class D
Ld,med Ld,max Ld,med Ld,max Ld,med Ld,max Ld,med Ld,max
V 0 1 0 1 0 0 0 0
VI 1 2 1 2 0 1 0 0
VII 23 4 12 3 01 2 0 1
VIII 34 5 23 4 12 3 01 2
IX 45 5 34 5 23 4 12 3
In Tables V and VI the damage levels computed through the numerical analyses
for each of the considered structural types, are reported. The values of damage de-
SEISMIC VULNERABILITY ASSESSMENT 391
Table V. Damage degrees L
d
of the examined structural types computed by applying articial
accelerograms
Frames with rigid beams Frames with exible beams Frames with no beams
BF-RB IF-RB PF-RB BF-FB IF-FB PF-FB BF-NB
I_EMS Ld Ld Ld Ld Ld Ld Ld
V 12 0 2 2 0 2 2
VI 23 1 3 23 12 3 3
VII 3 1 34 3 2 4 4
VIII 4 23 5 5 23 5 5
VIIIIX 5 4 5 5 4 5 5
Table VI. Mean and maximum values of damage degrees L
d
of the examined structural types
computed by applying natural accelerograms
Frames with Frames with Frames with
rigid beams exible beams no beams
BF-RB IF-RB PF-RB BF-FB IF-FB PF-FB BF-NB
I_EMS Ld Ld Ld Ld Ld Ld Ld
med max med max med max med max med max med max med max
V 12 2 0 1 2 23 2 23 0 12 2 3 2 23
VI 2 23 1 12 23 3 2 3 12 12 23 3 3 3
VII 2 23 12 12 23 3 23 3 12 2 3 34 3 4
VIII 23 3 12 2 3 34 23 34 12 2 3 4 3 45
VIIIIX 3 34 2 2 3 4 3 4 2 2 34 5 34 5
grees are drawn from Figures 8 and 9 making reference to the relationship between
PGA values and I_EMS reported in Table III.
By comparing the values in Table IV with those in Tables V and VI, some in-
teresting observations can be made. Given that, according to EMS98, C is the most
likely vulnerability class for the structural types under examination, the damage
degrees computed by applying the articial accelerograms appear clearly over-
estimated.
A more realistic evaluation can be made on the basis of the results computed by
applying natural accelerograms. In this case, vulnerability levels matching those
relevant to framed buildings without or with low earthquake resistant design (L-
ERD in Figure 10) can be estimated. Specically, in accordance with EMS98, the
results of the numerical analyses show that the possible vulnerability classes for
these buildings are placed in the BD range. Within this range, class C can be
considered the most likely vulnerability class. As less probable cases, the regularly
392 ANGELO MASI
inlled framed buildings have the lowest vulnerability (class D), whereas the pilotis
frames and the frames whose columns are connected only by a strip of the tile lintel
oor diaphragm have the highest vulnerability (class B). It should be noted that,
according to EMS98, class D is considered probable for the framed R/C buildings
with moderate earthquake-resistant design (M-ERD) and not very likely for the L-
ERD buildings. The results of the present study thus corroborate the thesis that the
frames with good quality regularly arranged masonry inlls have a good seismic
resistance even if designed to carry only vertical loads.
7. Conclusion
A simple and reliable procedure to evaluate the seismic resistance of some struc-
tural types of existing R/C buildings has been set up. It is mainly based on an
accurate recognition of the main and widespread structural characteristics of the
post-1970 building stock designed only to gravity loads. To this purpose, the codes
in force, the handbook typically in use and, principally, a wide database of original
design drawings relevant to real typical buildings of the period under considera-
tion have been examined. Consequently, the examined structural models have been
selected and described in such a way as to be representative of real existing R/C
buildings.
The results conrm the strong role of the seismic input. As regards the selection
of accelerograms, the choice of articial ones generated according to some codes
(e.g., Eurocode 8, CEN, 1994a) appears inadequate for vulnerability assessment
evaluations. They are too onerous when compared to natural accelerograms, as-
suming equal PGAvalues, because of their damage potential and frequency content
as the variations in the preStage 49 draft of Eurocode 8 (CEN, 2002b) implicitly
suggest. Moreover, their conventional characteristics makes them suitable only for
design purposes. By comparing the results obtained with the two types of seismic
input but assuming equal Arias Intensity values, almost coincident values of the
seismic response are obtained, emphasizing that AI is a more effective seismic
parameter than PGA to represent the damage potential of an accelerogram.
When using articial accelerograms unrealistic levels of vulnerability should be
assigned, particularly to the BF and PF types. Results close to the behaviour shown
by R/C buildings in past earthquakes (i.e. Southern Italy 1980) are obtained by
applying natural accelerograms. Due to the high scatter of the results, in this case
both mean and maximum values have been taken into account. Maximum values
explain the numerous partial or total collapses observed in past strong earthquakes
(I_EMS VIII). Specically, by comparing the seismic behaviour of the examined
types, a high vulnerability is shown by the pilotis frames and by the bare frames
without beams. On the contrary, a low vulnerability (class D of EMS98) is shown
by the frames with regularly arranged good quality masonry inlls. In this case
the failure probability can be considered unlikely also after strong earthquakes. An
SEISMIC VULNERABILITY ASSESSMENT 393
intermediate behaviour (vulnerability class BC of EMS98) is shown by the frames
without inlls or with ineffective inlls.
Finally, it is worth noting that the evaluations carried out are relevant to plane
models, even if representative of real buildings. Consequently, it is deemed that
where the relevant three-dimensional models are analyzed, the assigned vulner-
ability classes can change to some extent. Real buildings, made of plane frames
having different dynamic characteristics, strengths and available ductility, show a
complicated non linear dynamic behaviour. As a result, further studies aimed at
a better evaluation of their vulnerability which take into account the interactions
between the different plane types are currently in progress.
Acknowledgements
This work was partly supported by Italian National Seismic Service (SSN) within
the framework of the research Contract Seismic Vulnerability of R/C Buildings
Part I carried out by a research group at University of Basilicata formed by the
writer and by Prof. M. Dolce, Dr F. Telesca and Ph. Dr M. Vona. The valuable
contribution of Prof. M. Dolce and some stimulating communications with Prof.
A. J. Kappos of University of Thessaloniki are gratefully acknowledged.
References
Arias, A. (1970) A measure of earthquake intensity, in Seismic design for nuclear power plants (ed.
R. J. Hansen), MIT Press, Cambridge, Massachusetts, 438483.
Braga, F., Dolce, M. and Liberatore, D. (1982) Southern Italy November 23, 1980 Earthquake: A
Statistical Study on Damaged Buildings and an Ensuing Review of the M.S.K.-76 Scale, CNR-
PFG n. 503, Rome.
Calvi, G.M. and Priestley, N. (1998) Assessment of existing buildings, in Seismic design of reinforced
concrete structures for controlled inelastic response, Bulletin CEB 240, Thomas Telford LTD,
London, 133166.
Calvi, G.M. and Recla, M. (2000) Assessment of current prediction capacities of the response of ex-
isting reinforced concrete buildings. Proc. of 12
th
World Conference in Earthquake Engineering,
Auckland.
Chopra, A.K. (2001) Dynamics of Structures Theory and Application to Earthquake Engineering,
2
nd
edition. Prentice-Hall, New Jersey.
CEN (1993) Eurocode 1 Basis of design and actions on structures Part 1: Basis of design, ENV
1991-1, Brussels.
CEN (1991) Eurocode 2 Design of concrete structures Part 1: General rules and rules for
buildings, ENV 1992-1-1, Brussels.
CEN (1994a) Eurocode 8 Design provisions for earthquake resistance of structures Part 1-1:
General rules - Seismic actions and general requirements for structures, ENV 1998-1-1, Brussels.
CEN (1994b) Eurocode 8 Design provisions for earthquake resistance of structures Part 1-2:
General rules General rules for buildings, ENV 1998-1-2, Brussels.
CEN (2002a) Eurocode 8 Design of structures for earthquake resistance Part 3: Strengthening
and repair of buildings, prEN 1998-3, Brussels.
CEN (2002b) Eurocode 8: Design of structures for earthquake resistance of structures Part 1:
General rules, seismic actions and rules for buildings, prEN 1998-1, Brussels.
394 ANGELO MASI
Cosenza, E., Manfredi, G. and Verderame, G.M. (1999) Problemi di verica sismica di telai pro-
gettati per crichi verticali. Proc. of 9

Convegno Nazionale LIngegneria Sismica in Italia, Torino


(in Italian).
D.M. LL.PP. 30/4/72 (1972) Norme tecniche alle quali devono uniformarsi le costruzioni in
conglomerato cementizio, normale e precompresso ed a struttura metallica (in Italian).
D.M. LL.PP. 20/11/87 (1987) Norme tecniche per la progettazione, esecuzione e collaudo degli
edici in muratura e per il loro consolidamento(in Italian).
Dolce, M. (1996) Seismic vulnerability evaluation and damage scenarios. Proc. of US Italian
Workshop Seismic Evaluation and Retrot, Columbia University, New York City.
Dolce, M., Masi, A., Marino, M. and Vona, M. (2002) Earthquake damage scenarios of the building
stock Potenza(Southern Italy) including site effects, Bulletin of Earthquake Engineering, N. 1
(accepted for publication).
ESC Working Group Macroseismic Scales (1998) European Macroseismic Scale 1998. Geo-
ForschungsZentrum Potsdam, Germany.
Fardis, M.N. (1998) Seismic assessment and retrot of RC structures. Proc. of 11
th
European
Conference on Earthquake Engineering (invited lecture), Paris.
Federal Emergency Managenent Agency (FEMA) (1992) NEHRP handbook for the seismic evalu-
ation of existing reinforced concrete buildings, FEMA Report 178, Washington DC.
FEMA-NIBS (1999) Earthquake Loss Estimation Methodology, HAZUS99 Technical Manual,
Vol. 13, Washington DC.
Ghobarah, A., Abou-Elfath, H. and Biddah, A. (1999) Response-based damage assessment of
structures, Earthquake Engineering and Structural Dynamics, 78, 789104.
Japan Building Disaster Prevention Association (JBDPA) (1977) Standard for seismic capacity
evaluation of existing reinforced concrete buildings, Tokyo.
Kappos, A.J. (1997) Seismic damage indices for RC buildings: evaluation of concepts and proced-
ures. Progress in Structural Engineering and Materials, 1(1), 7887.
Kappos, A.J. (2001) Seismic vulnerability assessment of existing buildings in Southern Europe
(invited lecture). Atti del 10

Convegno Nazionale LIngegneria Sismica in Italia, Potenza, Italy.


Kunnath, S.K., Mander, J.B. and Reinhorn, A.M. (1990) Seismic Response and damageability of
gravity-load (non-seismic) designed buildings, Proc. of 9
th
European Conference on Earthquake
Engineering, Moscow, pp. 323332.
Mainstone, R.J. (1974) Supplementary note on the stiffness and strength of inlled frames, Current
Paper CP13/74, Building Research Establishment, London.
Mander, J.B., Priestley, M.J.N. and Park, R. (1988) Theoretical stress-strain model for conned
concrete, J. Am. Soc. Civ. Engrs, 114(8), 18041826.
Margottini, C., Molin, D., Narcisi, B. and Serva, L. (1985) Intensity vs. acceleration: italian data,
Proc. of the Conference on Historical Seismicity of Central-eastern Mediterranean Region,
pp. 213226.
Masi, A. (2000) Valutazione della resistenza sismica di strutture intelaiate in c.a. progettate negli
anni 70, Proc. of Final Workshop su Protezione sismica delledilizia esistente e di nuova
edicazione mediante sistemi innovativi, Napoli (in Italian).
Masi, A., Dolce, M., Vona, M. and Telesca, F. (2001a) Valutazione della vulnerabilit sismica di
edici in c.a. a struttura intelaiata realizzati dopo il 1970 Parte I: denizione delle tipologie
strutturali e metodologia di analisi. Atti del DiSGG, N. 4, Potenza (in Italian).
Masi, A., Dolce, M., Goretti, A., Telesca, F. and Vona, M. (2001b) Resistenza sismica di telai in
c.a. relativi ad edici esistenti con e senza tamponature, Proc. of 10

Convegno Nazionale
LIngegneria Sismica in Italia, Potenza (in Italian).
Min. LL. PP. Presidenza del Consiglio Superiore Servizio Tecnico Centrale (1997) Linee guida
per progettazione, esecuzione e collaudo di strutture isolate dal sisma, Ingegneria Sismica, n. 1
(in Italian).
Naeim, F. (ed.) (1989) The seismic design handbook, 1
st
edition, Van Nostrand Reinhold, New York.
SEISMIC VULNERABILITY ASSESSMENT 395
Pagano, M. (1968) Teoria degli edici Edici in cemento armato, Edizione Liguori, Napoli (in
Italian).
Park, R. and Paulay, T. (1975) Reinforced Concrete Structures, J. Wiley & Sons, New York.
Park, Y.J. and Ang, A.H.S. (1985) Mechanistic seismic damage model for reinforced concrete,
Journal of Structural Engineering, 111(4), 722739.
Paulay, T. and Priestley, M.J.N. (1992) Seismic design of reinforced concrete and masonry buildings,
J. Wiley & Sons, New York.
Penelis, G.G. and Kappos, A.J. (1997) Earthquake resistant concrete structures, E & F Spon,
London.
Santarella, L. (1968) Il cemento armato Le applicazioni alle costruzioni civili ed industriali, II
volume, Edizione Hoepli (in Italian).
Trifunac, M.D. and Brady, A.G. (1975) A study on the duration of strong earthquake ground motion,
Bulletin of Seismological Society of America, 65(3), 581626.
Tsai, K.C. and Li, J.W. (1994) DRAIN2D+ and VIEW2D user guide, Department of Civil Engineer-
ing, National Taiwan University.

You might also like