17-Seismic Response and Reliability Index (2022) - Jara
17-Seismic Response and Reliability Index (2022) - Jara
17-Seismic Response and Reliability Index (2022) - Jara
A R T I C L E I N F O A B S T R A C T
Keywords: For more than 50 years, buildings with soft ground floors continue to suffer severe damage and
Weak first-soft story buildings collapse when subjected to seismic events. During the earthquakes of September 19, 1985 and
Failure probability 2017 that occurred in Mexico, a significant number of collapsed buildings had this type of
Passive energy systems structure. Currently, many existing buildings have a parking zone on the ground floor and
Reliability index apartments for residential use on the upper floors. The parking area is supported by columns and
Real strong motions
beams, whereas the slabs of the upper floors are supported by masonry walls, creating a weak
story at the ground level. In order to explain the collapses of weak story buildings observed during
the earthquake of September 19, 2017, this study evaluates the seismic response and determines
the reliability indices of a typology of buildings with flexible ground floors in Mexico City.
Additionally, passive control devices with friction energy dissipation, hysteric behavior and
viscous dampers, were proposed to retrofit the buildings and increase the reliability indices of the
structures. Results showed that the reliability indices of existing buildings with a weak ground
floor are very low, which explains the observed damages and collapses during the 2017 earth
quake. Reliability indices of the retrofitted buildings were significantly increased and the failure
probabilities of the structures were strongly reduced, showing the effectiveness of passive control
systems to mitigate the seismic risk of this type of structures.
1. Introduction
Buildings with a weak and flexible ground floor are identified in the architectural solutions of the 60’s and 70’s to use open ground
floors as parking. These buildings, combined with the common design and construction practices of the time, make these structures
seismically vulnerable [16]. After the occurrence of destructive earthquakes, it is very common to find collapsed soft-story buildings
[1,2,12,24,36,39].
First soft-story buildings have large displacement demands in the ground floor that justify the observed damages and collapses [13,
24]. From decades, retrofit schemes have been proposed by adding passive control systems, bracing systems, shear walls and RC
jacketing, among others [4,16,20,22,23,38,40]. Different retrofit proposals showed important reductions in interstory drift ratio
demands, improving the seismic response of the buildings [38,41]. Most of the studies found that a stiffness increase in the soft story
improves the expected behavior but it also increases the base shear demands requiring to evaluate the foundation capacity in existing
buildings.
In general, the published studies that incorporate proposals to retrofit buildings with a flexible ground floor, use traditional
* Corresponding author.
E-mail address: [email protected] (J.M. Jara).
https://doi.org/10.1016/j.jobe.2022.104199
Received 7 December 2021; Received in revised form 31 January 2022; Accepted 8 February 2022
Available online 17 February 2022
2352-7102/© 2022 Elsevier Ltd. All rights reserved.
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
iterative processes that consider as main design parameters the first story displacement, drift ratio, ductility demand of the ground
floor, rotational ductility demand in columns or the ductility demand of the added structural elements [16,23,38]. Other studies [28]
proposed a Displacement-Based Design procedure to design hysteretic dampers for structures that require to be retrofitted, modelling
the concrete elements with a hysteretic model. In this study, the author analyzed three structures that must attain performance levels of
the Italian code and showed the importance of properly consider the hysteretic behavior of the structural elements in the methodology.
Since the main objective of many passive control systems is to increase the energy dissipation capacity of the structure, it would be
desirable to design the retrofitted structures with energy dissipation systems using an energy balance based seismic design.
The September 19, 2017 earthquake in Mexico caused the collapse of dozens of structures with a weak ground floor. An important
number of seismic stations were situated very close to the damaged and collapsed buildings. The seismic vulnerability assessment of
structures requires to select accelerograms recorded from earthquakes with important magnitude. Frequently, the number of seismic
records is limited and therefore accelerograms from seismic events of moderate magnitude that do not necessarily lead to seismic
demands of the collapse limit state must be included. In these cases, seismic records are scaled in amplitude to quantify the building
seismic vulnerability. The 2017 earthquake caused partial and total collapses in buildings of Mexico City and there were seismic
stations that recorded accelerograms very close to the collapsed structures. These set of seismic records allow evaluating the structural
reliability of this type of structures without the need to scale the recorded signals. The reliability indices serve as the basis for
quantifying the effect of adding energy dissipators in reducing the probability of failure of the retrofitted buildings. This study de
termines the reliability index as a negative value of the inverse cumulative distribution function of the standardized normal variable of
the failure probability of the buildings.
This study describes building damages and some statistics of the September 19, 2017 earthquake in Mexico City, making particular
emphasis on first soft-story buildings. Nonlinear dynamic analyses of typical three- five- and eight-story buildings, subjected to a suite
of 45 seismic records, were conducted to evaluate their failure probability and reliability index. To reduce the expected failure
probability in existing buildings subjected to seismic ground motions in the future, friction devices, viscous dampers and hysteretic
energy dissipation systems were proposed as retrofit alternatives to increase the reliability index. The design of the passive control
devices uses an energy-based methodology.
3. Seismic records
On September 19, 2017, an earthquake with epicenter in the center of the Mexican Republic occurred and caused serious damages
2
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
3
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
in Mexico City. The seismic movement of magnitude M = 7.1, originated in a normal type intraplate seismic source at a depth of 57 km
was located 120 km away from Mexico City. Dozens of buildings collapsed and hundreds of structures were severely damaged. The
severity of the earthquake and the proximity of the seismic stations to the location of the damaged and collapsed buildings allowed to
conduct nonlinear analyses in a group of structures to assess reliability indices without the need to perform any process of scaling of
accelerograms.
The earthquake was recorded in 67 seismic stations in Mexico City. 61 stations belong to the Center for Instrumentation and Seismic
Recording (CIRES) and six to the Institute of Engineering of the National Autonomous University of Mexico (UNAM). 15% of them are
located on hard soil (type I), 15% on transition soil (type II) and the remaining 70% on flexible soil (type III), named lake area of the
city. Fig. 5 shows the epicenter of the seismic event in the Mexican Republic located on the border of the states of Morelos and Puebla in
the south-central zone of Mexico.
Fig. 6 shows the CIRES seismic network, the location of collapsed buildings and the seismic zonation of Mexico City. As shown in
Figs. 2 and 6, there are several seismic stations close to the sites of collapsed buildings. Since most of the building collapses occurred in
zones IIIa and IIIb, the accelerograms recorded in these two seismic zones were collected.
Fig. 7a shows the response spectra for 5% damping of the two horizontal components recorded in zones IIIa and IIIb, and it displays
the dominant periods and spectral amplitudes of the seismic records. Fig. 7b presents the response spectra of the root mean square of
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
the two horizontal components (quadratic mean) calculated as Sa = 0.5*(Sa21 + Sa22 ); Sa1 = pseudoacceleration in one horizontal
direction and Sa2 = pseudoacceleration in the other horizontal direction. Fig. 7b also shows the mean value of the response spectra
(solid line), the mean value±one standard deviation (dotted line and dashed line) and the elastic response spectrum of the 1976 code
(code in force at the time the damaged buildings were built). The dominant period is located at T = 1.4 s. Some response spectra had
amplitudes greater than 600 cm/s2, although most seismic records had spectral amplitudes below that value.
One of the main reasons that explains the damage concentration in buildings in the range of 2–8 stories as compared with taller
buildings, is related to the dynamic characteristics of the ground movement that presented dominant periods for values less than 1.4 s.
The most damaged and collapsed buildings were built in the late 1970s and early 1980s, when the current seismic code established a
seismic coefficient of 0.24 g at the plateau of the elastic design spectrum. For this reason, the buildings designed in those decades have
relatively high fundamental periods, which means that buildings between four and eight stories are situated to the left of the highest
spectral amplitude of the September 19, 2017 earthquake. When buildings enter the range of inelastic behavior, the period grows,
increasing the seismic spectral demands as well, and eventually causing partial or total collapses. Fundamental periods of 10-story
buildings or taller were located in areas close to the maximum spectral amplitude. Any increase of the period moves away the
structures from the area of maximum amplifications, and decreases rapidly the acceleration amplitudes.
4. Numerical models
4.1. Original buildings
The most frequent use in buildings with a weak ground floor is for residential apartments using a parking area on the ground floor.
According to the statistics of severe damaged and collapsed buildings during the 2017 earthquake, rectangular and square shaped
structures with three, five and eight stories were selected. Despite not having specific plan dimensions of the damaged buildings, most
of them were of rectangular shape. Many years ago, most of the damaged zones were housing areas and it is very common to find
building widths in the range of 10–20 m. For this reason, a width of 15 m was selected in the numerical models of rectangular plan. On
the other hand, for a better use of both facades in corner buildings, it is quite common to find square buildings.
Table 1 shows the four buildings selected for the analyses; 3S is a rectangular three-story building, 5S-RP is a rectangular five-story
building, 8S is a rectangular eight-story building and 5S- SP is a square five-story building. Two five-story buildings (rectangular and
square shaped floors) were analyzed due to the typologies of the collapsed structures and because the buildings with this number of
floors were the most damaged structures during the September 19, 2017 earthquake.
Fig. 5. September 19, 2017 earthquake’s epicenter (star symbol) located at 120 km from Mexico City (adapted from http://www.ssn.unam.mx/sismicidad/reportes-
especiales/2017/SSNMX_rep_esp_20170919_Puebla-Morelos_M71.pdf).
4
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Fig. 6. Seismic stations, collapsed buildings and seismic zonation of Mexico City.
Fig. 7. Response spectra of the accelerograms recorded in zones IIIa and IIIb. (a) Response spectra of the two horizontal components and (b) root mean square of
spectral acceleration, mean value±one standard deviation and design spectrum.
Table 1
Building models.
Model Number of floors Height of first floor (m) Interstory height (m) Plan dimensions (m) Seismic zone
5
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Fig. 8a shows the floor plan of models 3S, 5S-RP and 8S. Fig. 8b displays the elevation view of the 3-story building façade, and
Fig. 8c and d shows the floor plan and elevation view of model 5S-SP, respectively. The ground floor of all buildings is supported on
reinforced concrete beams and columns, and the upper levels are supported on confined masonry walls. The longitudinal direction of
the buildings is labeled with numbers and the transverse direction is labeled with letters.
Depending on the location of stairs, light wells and infill walls in interior bays, they can be a source of irregularity and eventually
produce torsion in buildings during an earthquake occurrence. The location of these elements varies in each architectural project. The
reports of the visual inspections after the earthquake did not provide specific information in this regard. Many images of partial and
total collapses of weak-story buildings suggest low torsion contribution to the building failure. However, some collapsed buildings
presented simultaneously more than one structural deficiency (torsion, weak story, among others). This study focused in weak-story
buildings disregarding any element that could produce torsion in the structure.
The buildings were analyzed and designed with the seismic code in force when they were built. The code of that period was the
Fig. 8. Floor plan and elevation of the building models. (a) Plan floor of the 3S, 5S-RP and 8S buildings; (b) elevation view of the 3S model; (c) plan floor of the 5S-SP
models and (d) elevation view of the 5S-SP model.
6
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
1976 Federal District Construction Regulations [33], together with the Technical Standards. In that code, Mexico City was divided into
three seismic zones: zone I (hard soil), zone II (transition soil) and zone III (soft soil). Due to the location of the damaged buildings, the
structural design was performed considering that they were located on zone III.
Initially, a spectral modal analysis was carried out to later design the buildings with the SAP2000 program [9]. The Federal District
Construction Regulations of 1976 [33] established three design spectra for the seismic zones of the city (Table 2). a0 is the peak ground
acceleration (PGA); c is the pseudo-acceleration at the plateau; T1 is the initial period of the plateau and T2 is the period where the
plateau ends. The design of the buildings was carried out considering all of them in seismic zone III.
According to the inspection reports of the collapsed buildings during the 2017 earthquake, on the ground floor the buildings had
masonry infill walls in the longitudinal direction of the two perimeter axes (bordering with neighboring properties). On the upper
floors there were no columns and the slabs were supported on confined masonry walls. Visual inspections also showed that the
buildings presented negligible damage in the longitudinal direction and upper stories, displaying partial and total collapses in the
transverse direction of the ground floor, where the weak floor mechanism was presented.
The compressive strength of concrete and the yield strength of the longitudinal and transverse reinforcement in beams and columns
were f’c = 25 MPa and 412 MPa, respectively. The Young modulus for steel was 200,000 MPa and for concrete was 3150 (f’c)1/2 (MPa).
Poisson ratio for concrete was 0.2, and 0.3 for steel. The same Poisson ratio for concrete was used for masonry. The shear modulus, G,
of masonry was 0.4 Em, where Em is the masonry Young modulus. The 1976 code specified that Em could be calculated as Em = 600
f*m for short duration loads and f*m = 0.0015 GPa for typical masonry built with handmade solid bricks in Mexico. Then, masonry
walls were modeled with Em = 0.9 GPa.
The compressive strength of concrete increases with time, reaching up to 25% more than the strength at 28 days after 10 years [14].
Conversely, aging effects include the possibility of corrosion attack. After evaluating the seismic damages in old RC buildings, many of
them showed poor seismic details, including large stirrups spacing and small concrete cover thickness, parameters that could induce
steel corrosion. When corrosion is of importance, the gain in concrete strength with age can be reduced or even the strength could be
eventually lower than the design strength [14]. Both parameters could be incorporated into a vulnerability study by assigning a
probability density to these variables. This study did not consider any of the two variables that could be important in the seismic
vulnerability of buildings and, therefore, they should be addressed in future works.
The columns and beams of the ground level were modeled with frame elements with six degrees of freedom at each node. These
three-dimensional elements include the effect of biaxial bending, torsion, axial deformation and biaxial shear deformations. The
confined masonry walls of the upper floors were created with elastic shell elements. Confined masonry includes tie RC columns and
beams (same width as wall thickness) that produce an integrated wall behavior. 140 mm thick walls are composed of brick units
connected with mortar. They were modeled with in-plane behavior as elastic elements using four-joint shell type finite element (FE)
with membrane behavior. The mesh of FE was created using the span distance divided by ten. The walls are connected to floor di
aphragms to prevent out-of-plane behavior. Observed damages and collapses of soft-story buildings have showed that upper stories
supported on brick masonry walls remain practically undamaged. In this study, the high flexibility of the ground level, as compared
with lateral stiffness of upper levels, produced low drift ratios in upper levels (around 0.0014). Consequently, elastic behavior of the
confined walls can be expected. Infill walls of the ground floor were modeled with equivalent diagonal compression strut using the
recommendations of the document FEMA 356 [15].
Table 3 shows the dimensions of columns and beams cross sections. In all cases, the first floor was supported by square shape
columns, whose dimensions varied in the interval from 0.40mx0.40 m to 0.55mx0.55 m, depending on the number of floors of the
building, with percentages of longitudinal steel in the interval from 0.79% to 1.13%. The cross section of the beams was 0.25 × 0.40,
with percentages of steel in the range of 0.28%–0.76%. The axial load ratio P/(Ag*f’c) of the load combination (dead load, live load
and earthquake loads) were in the range of 0.28–0.38 depending on the number of floors of the building. Ag is the gross area of the
column.
The soft-story mechanism can be observed by drawing the mode shapes of the buildings’ first mode of vibration. Fig. 9 shows the
modal configuration of the fundamental modes of three-, five- and eight-story buildings. It also shows the interstory drift of the modal
shapes, where the soft-story mechanism is more evident.
Table 2
Design spectra of the 1976 seismic code for Mexico City.
7
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Table 3
Characteristics of beams and columns of the buildings.
Cross section (m) Steel ratio (%) Cross section (m) Steel ratio (%)
3S 0.40 × 0.40 0.82–0.99 0.25 × 0.40 0.28–0.41
5S-RP 0.45 × 0.45 0.79–0.96 0.25 × 0.40 0.38–0.57
8S 0.55 × 0.55 0.90–1.13 0.25 × 0.40 0.38–0.76
5S-SP 0.50 × 0.50 0.77–0.91 0.25 × 0.40 0.38–0.57
Fig. 9. Fundamental mode shapes and interstory drift of modal shapes. (a) and (b) 3S building; (c) and (d) 5S-RP buildings and (e) and (f) 8 S building.
Fig. 10. 3-story retrofitted buildings. (a) Plan view; (b) Elastoplastic dissipators and (c) friction and viscous devices.
8
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
with elastoplastic dissipators and Fig. 10c presents the 3-story building retrofitted with viscous and friction devices.
The methodologies used to design an energy dissipation system are frequently based on forces or displacements. Some procedures
establish a target displacement, some define a ductility demand and others are based on controlling drift demands. It is widely
recognized that the expected damages in a structure subjected to seismic movements depend not only on the displacement demands,
but also on the hysteric energy demand. Uang and Bertero [42] showed the energy balance equations of a structure and establish the
difference between the use of absolute or relative displacements in energy balance components. In another study [30], the authors
showed that the input energy and hysteretic energy of systems with several degrees of freedom with energy dissipators of hysteretic
behavior, can be predicted adequately with single degree of freedom (SDOF). Approaches based on the energy balance as design
alternatives for buildings have also been proposed [25].
In this study, an energy-based methodology for the design of energy dissipation devices was used [19]. The method applies an
iterative process based on interstory energy demand proposed by Chou and Uang [7]. As a measure of energy demand, the equivalent
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
velocity defined as Va = 2Ea /m is used. Ea is the sum of the elastic and hysteretic strain energy and m is the mass. Subsequently,
constant ductility strength spectra defined as Cy = Fy /mg are constructed, and finally the displacement spectra Ds for several damping
ratios are used to assess the seismic demand. The energy demand is calculated by selecting the first two vibration modes with the
highest participation factor and afterwards performing pushover analysis to characterize equivalent single-degree-of-freedom systems.
The structure with energy dissipation devices was modeled based on estimating an equivalent damping and stiffness. The process
started assuming a damping (normally 5%) and following with iterations considering the energy that needs to be dissipated. At the end
of each iteration the equivalent damping was updated. Finally, when converging, the energy demand was distributed at all levels of the
structure. The process is described in detail in Ref. [19]. Fig. 10 shows the strength response spectra (Cy ), the equivalent velocity
spectra (Va), and the displacement spectra (Ds ), used in the design process. The response spectra in Fig. 11 are the average values of the
response spectra of the 45 seismic records mentioned in section 3 and they include five curves of uniform ductility spectra in the range
of 1–8.
Each iteration requires an effective damping ratio, as a function of the ductility demand, determined according to Ref. [26], which
is based on equating the energy dissipated per cycle of a periodic excitation with the corresponding linear viscous damping. Table 4
shows the parameters of the devices obtained after the convergence of the energy-based process. The nonlinear analyses of the
buildings were conducted using these parameters. F0 is the friction force of the sliding device; K0 is the device stiffness, defined as K0 =
F0 /u0 ; u0 is the relative axial displacement of the friction device; c is the damping coefficient of the viscous dissipator; Fy is the yield
strength of the elastoplastic device, KD is the stiffness of the elastoplastic dissipator and a hardening ratio of 2% was assumed.
Table 5 shows the fundamental periods of the building models, including the non-retrofitted structure. The periods of the original
buildings were in the range of 0.55–1.06 s. When including friction energy dissipators the periods were reduced to the interval of
0.37–1.03 s and the structures retrofitted with hysteric dissipators had fundamental periods in the range of 0.39–1.02 s. Viscous
dampers did not modify the fundamental periods of the original buildings.
Fig. 11. Uniform ductility spectra (μ = 1–8) used to design the retrofitted buildings with energy dissipation systems. (a) Strength response spectra Cy ; (b) equivalent
velocity spectra Va and (c) displacement spectra.Ds
9
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Table 4
Parameters of the energy dissipation devices.
F0 K0 c Fy KD
(kN) (kN/m) (kNs/m) (kN) (kN/m)
Table 5
Fundamental periods of the building models.
selection was based on a comparison of experimental tests of RC columns [37] with the numerical response obtained using the values
shown in Fig. 12. These parameters define the material inelastic behavior with a backbone curve. Y identifies the yield strength, U the
ultimate strength, L the ductile limit, R the residual strength and X the failure.
Initially, the non-retrofitted buildings were subjected to the 45 accelerograms recorded in Mexico City during the September 19,
2017 earthquake. Fig. 13a shows drift ratio demands of the four numerical models and the admissible drift ratio (0.008) of the code in
force at the time of construction of the buildings. The largest drift ratio demand occurs in the five-story building with a rectangular
floor plan, exceeding the allowable drift ratio of the code in 31 of the 45 accelerograms used. Some of them with large drift demands
that explain the observed damages.
The rotation demands of the ground floor columns were comparable when subjected the buildings to a seismic record. Fig. 14a
shows the rotation demands in a corner column of the ground floor of the original buildings. Again, the highest rotation demand
corresponds to the rectangular five-story buildings. Three-story buildings presented the lowest rotation demands. Fig. 14b shows the
displacement ductility demands of the ground floor (peak lateral displacement demand divided by the yield displacement of the
ground floor). Several seismic records produced large ductility demands on the five-story buildings that also explain the severe damage
and collapses observed after the occurrence of the seismic event.
The energy dissipation systems designed to retrofit the buildings were incorporated into the numerical models and subjected to the
suite of 45 accelerograms. Friction devices, with rigid-plastic behavior, were designed for a slip displacement slightly smaller than the
lateral yield displacement of the ground floor. The slip force, showed in Table 4, was obtained with the procedure presented in section
4.2. Seismic isolator rubber type element was used in Perform 3D to model the elastoplastic devices with the parameters presented in
Table 4. This element does not have stiffness degradation or strength loss. Again, using the parameters displayed in Table 4, and with a
Fluid Damper type element in PERFORM 3D, viscous dampers were modeled. Viscous elements have not elastic stiffness and support
only axial forces.
Fig. 15a shows the rotation demands in a column of the ground floor of the original and retrofitted three-story buildings. The
horizontal lines display damage limit states of immediate occupancy (IO), life safety (LS) and collapse prevention (CP), in accordance
with the values suggested in the ASCE Standard [3]. The numerical values that define each limit state depend on the failure condition,
on the axial load ratio P/Ag*f’c and on the ratio ρ = Av/bw*s [3]. P is the axial load on the column, Ag is the cross-sectional area, f’c is
10
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Fig. 13. Seismic demands in the original buildings. (a) Maximum first-story drift ratios and (b) base shear demand/building weight ratios.
Fig. 14. Rotation and displacement ductility demands in ground floor of the original buildings. (a) Rotation demands and (b) displacement ductility demands.
Fig. 15. Seismic demands of the original and retrofitted three-story buildings (3S). (a) Rotation demands and (b) Base shear ratios.
the compressive strength of the concrete, Av is the area of the transverse reinforcement, bw is the web width, and s is the stirrup
spacing. The nonlinear analysis showed that the buildings reached flexural failure by forming plastic hinges. P/Ag*f’c ratios of the
buildings were in the range of 0.28–0.38 and in all cases ρ was close to 0.002. The acceptance criteria for each limit state (IO, LS nd CP)
showed in the following figures were obtained by interpolation of the values indicated in table 10.8 of the ASCE Standard [3].
Fig. 15b shows the ratio of the base shear demands of the retrofitted models to the base shear demands of the original model. A ratio
greater than unity implies that the retrofitted model increased the base shear demands of the non-retrofitted building.
The inclusion of energy dissipators appreciably reduced the rotation demands in columns. The seismic response of the three-story
Fig. 16. Seismic demands of the original and retrofitted five-story buildings with square floor (5S-SP). (a) Rotation demands and (b) base shear ratios.
11
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
retrofitted buildings placed them below the limit state of life protection and, in most of the cases, below the limit state of immediate
occupation. With the exception of the first three records that had the highest seismic demand, the seismic response of the retrofitted
models was relatively close regardless of the energy dissipator used. The base shear demands of the retrofitted models were, in general
terms, very similar to that of the original models (many cases close to unity). However, buildings with friction and hysteretic devices
presented in most cases lower demands than those of the original model, mainly due to the energy dissipation of the devices.
Fig. 16 shows the seismic demands of the square five-story buildings. Buildings with this number of floors were the most damaged
during the seismic event. The inclusion of hysteretic and friction type energy dissipators significantly reduced the seismic response and
the building with viscous dampers presented slightly higher demands than the other two dissipation systems (Fig. 16a). However, most
seismic records produced demands that placed the retrofitted buildings below the limit state of immediate occupancy.
The base shear demand of buildings with viscous devices was again greater than that obtained with other retrofitted structures
(Fig. 16b). However, the demand was very close to that of the original building. The decrease in base shear demands of the models with
hysteric and friction devices, with respect to the original building, is slightly less than that of the three-story buildings, mainly due to
higher seismic demands of the 2017 earthquake in five-story buildings.
Fig. 17a shows the rotation demands of the rectangular five-story building. Despite high rotation demands of the original building,
the inclusion of energy dissipation systems reduced the seismic response to levels below the limit state of life safety, and in most cases
even below the immediate occupancy limit state. The column rotation demands were quite similar in the three retrofitted models.
Fig. 17b shows the ratio of base shear demand of the retrofitted models to the original buildings. Most of the retrofitted buildings
reduced the base shear demands of the original buildings. Again, the building with hysteric devices presented the lowest base shear
demands.
Fig. 18a shows the rotation demands in a column of the eight-story building. Although the maximum demand of the original model
decreased with respect to the rotation demands of lower-rise buildings, the efficiency of the inclusion of energy dissipation devices is
notably reduced. The CH84 seismic record produced rotation demands very close to the collapse prevention limit state in the three
retrofitted buildings. Although some accelerograms placed the response of retrofitted buildings above the limit state for life safety, a
greater number was located below this limit state, and most of them were again below the limit state of immediate occupancy.
The quite different behavior of the eight-story buildings, as compared with previous cases, showed the greater contribution of the
higher modes in the seismic response and the larger base shear demand in the eight-story structures. Fig. 18b shows the base shear
force ratios. It is also noticeable the increase in base shear demands of the retrofitted models, reaching the retrofitted model with
viscous dampers subjected to the JC54 accelerogram a base shear 60% greater than that of the original model. Buildings retrofitted
with hysteretic dissipators kept, even in this case, base shear demands very close to those of the original building.
The inclusion of passive control systems sometimes increases building floor accelerations. Fig. 19 shows the ratio of the maximum
floor acceleration of the retrofitted models to that of the non-retrofitted models. Values greater than one mean that the retrofitted
structure increased the floor acceleration of the original buildings.
Three- and five-story retrofitted buildings maintained less floor accelerations than the demands of the non-retrofitted structures.
However, the eight-story building retrofitted with friction dampers and hysteric devices presented higher acceleration demands than
the original building from the second floor to the roof. This building retrofitted with viscous devices, maintained the floor acceleration
demands at values less than or very close to those of the non-retrofitted model. Generally speaking, the building with viscous dampers
presented the lowest floor accelerations, followed by buildings with friction dampers and finally, the highest floor acceleration de
mands were generated in the building with elastoplastic devices. Clearly, the results showed the difference in the seismic response of
the eight-story building compared to the lower-rise buildings.
5. Reliability index
The reliability index is a parameter that provides a measure of the structural safety. In the Load and Resistance Factor Design
methodology, it has been used for the calibration of load and resistance factors [17]. The reliability index in the useful life of a structure
(T), is defined as,
( )
βf = Φ− 1 Pf ,T (1)
Pf ,T = 1 − e− λf T
(2)
Equation (2) assumes that the earthquake occurrence is a Poisson Process. λf is the annual failure rate of the structure (Eq (3)).
⃒
∫ ⃒
⃒
λf = P[f |im]⃒⃒dλim (3)
⃒
im
P[f|im] is the probability of reaching or exceeding a specified limit state, in this case the collapse, given a seismic intensity (im). λim is
the exceedance rate of the seismic intensity selected at the study site (result of a seismic hazard assessment) and dλim is the negative
derivative of λim with respect to the seismic intensity. As a measure of seismic intensity, the peak ground acceleration (PGA) was used.
The exceedance rate λim was determined based on the seismic hazard study reported in Ref. [35] for the hard soil area of Mexico City.
12
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Fig. 17. Seismic demands of the original and retrofitted five-story buildings with rectangular floor (5S-RP). (a) Rotation demands and (b) base shear ratios.
Fig. 18. Seismic demands of the original and retrofitted eight-story buildings (8S). (a) Rotation demands and (b) base shear ratios.
Fig. 19. Floor acceleration ratios of the retrofitted to the original buildings. (a) Three-story building; (b) square five-story building; (c) rectangular five-story building
and (d) eight-story building.
According to the amplification of the PGA from the area of firm ground to the location area of the damaged structures, the exceedance
rate was obtained and used to calculate the reliability indices. PGA amplification from hard to soft soils in Mexico City was obtained
with the seismic records of the September 19, 2017 earthquake. The seismic station on hard soil was University City (CU), which
13
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
recorded a PGA of 117 gals. An average amplification of 2.8 was estimated for seismic stations on soft soils. This value was consistent
with that obtained in the study by Román-de la Sancha [34]. It was assumed that the capacity and the seismic demand are lognormally
distributed, so the failure probability is also a lognormal density function. Studies of RC structures have found coefficients of variation
of the capacity in the range of 0.33–0.52 [6] and 0.53–0.59 [32]. The coefficients of variation of the seismic demand in the interval of
the building periods were in the range of 0.30–0.43. The variation coefficient of the lognormal density function is similar to the value of
the standard deviation of the natural logarithm (dispersion parameter of the probability density function). According to the previous
values, if the average value of the highest dispersion in seismic capacity is used, 0.5*(0.53 + 0.59) = 0.56 and using the average value
of dispersion in seismic demand, 0.5*(0.30 + 0.43) = 0.37, the dispersion parameter of the joint probability density function (failure
probability) is: sqrt(0.56^2 + 0.37^2) = 0.67, which was rounded to 0.7.
The failure probability P[f|im] was determined based on nonlinear dynamic analyses of the original and retrofitted buildings
subjected to a suite of accelerograms selected from those seismic stations located close to the area of the collapsed buildings. The
seismic intensity that causes the failure of the structure was considered as a random variable lognormally distributed. The process
started by selecting the parameters of the failure probability density function. The building failure was assumed when the average
rotation demands in columns exceeded the collapse limit state, when subjected the building to the suite of seismic records. The seismic
intensity (im) of this limit state is the mean PGA value of the group of accelerograms that it was used to obtain the second parameter of
the failure probability density function (the other parameter was described in the previous paragraph). Because of the weak story
mechanism, several columns reached, at the same time step, the collapse limit state conducting to the building failure. Fig. 20 shows
the location of the set of 16 seismic stations (black triangle) selected to carry out these analyses. The seismic stations are inside or on
the edge of the polygon drawn in the figure and all of them are close to the collapsed structures (black circle). Fig. 20 also shows a table
to identify the seismic station and the PGA in each direction of the seismic records used in the analysis. In the original five-story
buildings, without any retrofit system, the real accelerograms (without any scaling process) led to an average column rotation de
mands that reached or exceeded the collapse limit state. Through the convolution of the failure probability (fragility curve) with the
exceedance rate of PGA of the site, the annual failure rate of the structure was obtained, which served as the basis for obtaining the
reliability indices. Conversely, to assess the failure probability density function of the other non-retrofitted and retrofitted buildings, it
was required to scale the seismic records to reach the failure limit states and thereby to obtain the median value of PGA that deter
mined the failure probability density function of these buildings.
The reliability indices were evaluated for a useful life T = 50 years, a value considered in the design spectra of the code [33] for the
seismic analysis and design of buildings. The analysis assumes that during the useful life of the buildings, the structures did not
accumulate any damage that would reduce their seismic capacity. Table 6 shows the failure probability and the reliability indices for T
= 50 years. As mentioned before, the suite of real seismic records produced rotation demands in columns of the five-story buildings
above the collapse limit state. Hence, the failure probability and reliability indices were assessed without the need of scaling the
accelerograms.
The minimum reliability indices corresponded to the five-story buildings, explaining the damages produced by the seismic event.
Five-story buildings with rectangular floor plan and few bays in the transverse direction of the structure were particularly vulnerable
Fig. 20. Seismic stations used to compute reliability indices using the accelerograms recorded during the September 19, 2017 earthquake.
14
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
Table 6
Reliability index of the original and retrofitted buildings.
(β = 1.37). The square five-story building that increased the number of bays in the transverse direction, raised the reliability index to β
= 1.71. The analyses showed that the three- and eight-story buildings increased the reliability indices that satisfactorily correlates with
the statistics of the damaged structures (smaller number of buildings severely damaged and collapsed).
The reliability indices of the retrofitted buildings were substantially increased. Three-story buildings (T = 0.371–0.391 s) displayed
reliability indices in the range of 3.28–3.62, five-story buildings (T = 0.591–0.689 s) in the range of 3.05–3.66 and eight-story
buildings (T = 1.022–1.026 s) in the range of 3.15–3.53. The reliability indices were dependent on the number of the structure’s
floors. However, the inclusion of any of the energy dissipators studied, greatly reduced the failure probability of the buildings with a
weak ground floor.
To quantify the impact of the addition of energy dissipation systems in the failure probability of the buildings, the ratio of the
failure probability of the original building to that of the retrofitted structure with the highest reliability index of each building model
was determined. This ratio was 0.015812/0.000147 = 107 for the three-story building (3S), 0.085363/0.000376 = 227 for the 5S-RP
building, 0.030054/0.000209 = 144 for the eight-story building (8S) and 0.043634/0.000126 = 346 for the 5S-SP building. All
retrofitted models, but particularly five-story buildings, substantially reduced the failure probability of the original buildings.
Other aspect that should be addressed in future studies, is to analyze the optimal retrofit system based on the selection of an
objective function that allows choosing the optimum solution. A methodology based on minimizing the expected annual loss to select
the optimal retrofit system is described in Ref. [18]. Additionally, before selecting the retrofit scheme, studies can be conducted to
determine the optimal seismic level to retrofit the buildings based on establishing an objective reliability index, which is not neces
sarily the same for all types of structures. These analyses could include damage control and cost of maintenance during the service life
of the buildings [5].
6. Conclusions
This study determined reliability indices of buildings with a weak ground floor creating numerical models with the geometric
characteristics of the severely damaged structures and collapsed buildings in Mexico City during the September 19, 2017 earthquake.
The seismic response of three-, five- and eight-story buildings subjected to a set of 45 real accelerograms recorded during the 2017
seismic movement was determined. Reliability indices for retrofitted buildings with energy dissipation devices were also obtained to
quantify the impact of these systems in decreasing the failure probability of the structures. The conclusions of the study are as follows:
The seismic response of the original buildings subjected to the recorded accelerograms showed large displacement and ductility
demands. Despite this information allowed to infer the expected damages generated by the earthquake, it does not allow directly to
quantify the failure probability of the structures. For this reason, reliability indices were calculated, which are directly related to the
failure probability of the buildings.
The reliability indices of the original structures (without any retrofit system) were quite low, which explained the observed
building damages during the September 19, 2017 earthquake. Five-story buildings had the smallest reliability indices that correlate
well with the technical reports of the post-earthquake visual inspections. Five-story buildings were the most damaged structures and
most of the collapsed buildings had this number of floors. The low reliability indices of existing weak first-story buildings on soft soils
in Mexico City showed the need to carry out structural interventions to mitigate the seismic risk in future earthquakes.
The inclusion of friction type, hysteretic and viscous energy dissipators significantly reduced the seismic demand of the buildings.
The increment of reliability indices in the retrofitted buildings showed that they are an efficient way to reduce the seismic risk of
existing weak first-story buildings. The impact of using energy dissipation systems to retrofit the original buildings was quantified with
the ratio of the failure probability of the original buildings to that of the retrofitted structures. This ratio was found in the range of
107–346, depending on the building height and retrofit system (fundamental periods in the range of 0.391–0.692 s). Although the
probability of failure provides information to quantify expected losses, it is more common to indicate reliability indices in the regu
lations. In general, the reliability indices related to building collapse depend on the code, the impact of the failure on people’s lives and
15
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
the speed of propagation of the damage (not sudden or sudden). According to Steenbergen [40], the American Society of Civil En
gineers (ASCE, 2010) recommends values in the range of 2.5–4.5 and the Eurocode (1990) in the range of 3.3–4.3. The use of the
buildings in this study (apartments), suggests that an acceptable value could be around 3.5. All original buildings had lower reliability
indices and some of the retrofitted buildings presented reliability indices slightly below this value.
Sometimes, by including a retrofit system in a building, the floor acceleration demands increase. This parameter is important when
evaluating the contents of a structure subjected to earthquakes. This study also determined the change in floor acceleration demands
when including energy dissipation systems. With the exception of some cases and in some floors of the retrofitted eight-story model, all
other retrofitted buildings reduced the floor acceleration demands of the non-retrofitted buildings.
This study was based on the collected information of damaged buildings during the September 19, 2017 earthquake. The failure
probability of the structures was assessed with a suite of seismic accelerograms recorded during this earthquake. Despite the proximity
of the seismic stations to the collapsed buildings, it should be noted that the study was limited to using the seismic records of this
earthquake. Additionally, the results and conclusions are applicable to the type of structures studied and designed with a regulation in
force in Mexico from 1976 to 1985. In the quantification of the reliability indices, this study used exceedance rates of PGA from a
seismic hazard assessment carried out specifically for Mexico City. The values reported here could be different according to con
struction practices, regulatory requirements and seismicity of other sites.
Author statement
José Jara: Responsible for writing the manuscript, formal analysis, data curation, supervision and reviewing the results.
Carlos García: Creation of numerical models, data curation, validation and creation of figures and tables.
Bertha Olmos: Formal analysis, supervision, validation and manuscript review.
Guillermo Martínez: Formal analysis, supervision and manuscript review.
References
[1] K. Adalier, O. Aydingun, Structural engineering aspects of the June 27, 1998 Adana-Ceyhan (Turkey) earthquake, Eng. Struct. 23 (2001) 343–355.
[2] M.H. Arslan, H.H. Korkmaz, What is to Be learned from damage and failure of reinforced concrete structures during recent earthquakes in Turkey, J. Eng. Fail.
Analys.. 14 (2007) 1–22.
[3] ASCE, Seismic Evaluation and Retrofit of Existing Buildings, American Society of Civil Engineers, Virginia, USA, 2014. ASCE Standard ASCE/SEI 41-13.
[4] A. Benavent-Climent, S. Mota-Páez, Earthquake retrofitting of R/C frames with soft first story using hysteretic dampers: energy-based design method and
evaluation, Eng. Struct. 137 (2017) 19–32.
[5] D. Biskinis, N. Fardis, Flexure-controlled ultimate deformations of members with continuous or lap-spliced bars, Struct. Concr. 11 (2010) 94–109, https://doi.
org/10.1680/stco.2010.11.2.93.
[6] C. Chien-Kuo, J. Wen-Yu, Evaluating procedure of optimal seismic retrofit level for a low-rise reinforced concrete building, J. Adv. Concr. Technol. 9 (2) (2011)
287–299.
[7] C.C. Chou, C.M. Uang, A procedure for evaluating seismic energy demand of framed structures, Earthq. Eng. Struct. Dynam. 32 (2003) 229–244.
[8] CICM, Preliminary Report of the Inspected Buildings after the September 19, 2017 Earthquake, College of Civil Engineers of Mexico, México, 2017 (In Spanish).
[9] S.A.P.2000 CSIa, Computer and Structures Inc., Berkeley, CA, 2015.
[10] CSIb PERFORM 3D, Computer and Structures Inc., Berkeley, CA, 2015.
[11] T.N. Do, F.C. Filippou, A damage model for structures with degrading response, Earthq. Eng. Struct. Dynam. 47 (2017) 311–332.
[12] A. Dogangün, Performance of reinforced concrete buildings during the may 1, 2003 Bingöl earthquake in Turkey, Eng. Struct. 26 (2004) 841–856.
[13] A.F.C. Dya, A.W.C. Oretaa, Seismic vulnerability assessment of soft story irregular buildings using pushover analysis, Procedia Eng. 125 (2015) 925–932.
[14] M.A. Ei-Reedy, Reinforced Concrete Structural Reliability, CRC Press, Boca Raton, FL, USA, 2012.
[15] FEMA 356, Prestandard and Commentary for the Seismic Rehabilitation of Buildings, Federal Emergency Management Agency. American Society of Civil
Engineers, Washinghton D.C, 2000.
[16] A. Furtado, H. Rodrigues, H. Varum, A. Costa, Assessment and strengthening strategies of existing RC buildings with potential soft-storey response, in: 9th
International Masonry Conference, Guimaraes, Portugal, 2014, pp. 1–9.
[17] T.V. Galambos, Load and resistance factor design, Eng. J.. Am. Inst. Steel Constr. 18 (1981) 74–82.
[18] R. Gentile, C. Galasso, Simplified seismic loss assessment for optimal structural retrofit of RC buildings, Earthq. Spectra 37 (1) (2021) 346–365.
[19] A. Habibi, F. Albermani, Energy-based design method for seismic retrofitting with passive energy passive energy dissipations systems, Austr. Earthq. Eng. Soc.
46 (2013) 77–86.
[20] F. Hejazi, S. Jilani, J. Noorzaei, C.Y. Chieng, M.S. Jaafar, A.A. Abang Ali, Effects of soft story on structural response of high rise buildings, IOP Conf. Ser. Mater.
Sci. Eng. (2011) 1–13, 012034.
[21] J.M. Jara, E.J. Hernández, B.A. Olmos, Effect of epicentral distance on the applicability of base isolation and energy dissipation systems to improve seismic
behavior of RC buildings, Eng. Struct. 230 (2021) 1–15.
[22] J.M. Jara, E.J. Hernández, B.A. Olmos, G. Martínez, Building damages during the September 19, 2017 earthquake in Mexico City and seismic retrofitting of
existing first soft-story buildings, Eng. Struct. 209 (2020) 1–15.
[23] D. Khan, A. Rawat, Nonlinear seismic analysis of masonry infill RC buildings with eccentric bracings at soft storey level, Procedia Eng. 161 (2016) 9–17.
[24] N. Kirac, M. Dogan, H. Ozbasaran, Failure of weak-storey during earthquakes, Eng. Fail. Anal. 18 (2011) 572–581.
[25] S. Leelataviwat, S.C. Goel, B. Stojadinovic, Energy-based seismic design of structures using yield mechanism and target drift, J. Struct. Eng. 128 (8) (2002)
1046–1054.
[26] Y.Y. Lin, M.H. Tsai, J.S. Hwang, K.S. Chang, Direct displacement-based design for building with passive energy dissipation systems, Eng. Struct. 25 (2003)
25–37.
[27] J.B. Mander, M.J. Priestley, R. Park, Theoretical stress–strain model for confined concrete, J. Struct. Div. ASCE 144 (8) (1998) 1804–1826.
[28] F. Mazza, A simplified retrofitting method based on seismic damage of a SDOF system equivalent to a damped braced building, Eng. Struct. 200 (2019) 1–19.
[29] F. Mazza, A plastic-damage hysteretic model to reproduce strength stiffness degradation, Bull. Earthq. Eng. 17 (2019) 3517–3544.
16
J.M. Jara et al. Journal of Building Engineering 50 (2022) 104199
[30] M. Nakashima, K. Saburi, B. Tsuji, Energy input and dissipation behaviour of structures with hysteretic dampers, Earthq. Eng. Struct. Dynam. 25 (1996)
483–496.
[31] R. Park, T. Paulay, in: Reinforced Concrete Structures. John Wiley & Sons, first ed., United States of America, 1975, pp. 36–45.
[32] K. Ramanathan, R. DesRoches, J.E. Padgett, Analytical fragility curves for multispan continuous steel girder bridges in moderate seismic zones, J. Transport.
Res. Board 2202 (2010) 173–182, https://doi.org/10.3141/2202-21.
[33] RCDF, District Federal Department. Technical Regulation for Seismic Design. Diario Oficial de la Federación, 14 de diciembre de 1976. México: Distrito Federal
1976, 1976 (in Spanish).
[34] C. Reyes, E. Miranda, M. Ordaz, R. Meli, Acceleration spectra assessment for different return periods in seismic zones of Mexico City, Rev. Ingen. Sísmica 66
(2002) 95–121 (in Spanish).
[35] A. Román-de la Sancha, R. Silva, Multivariable analysis of transport network seismic performance: Mexico city, Sustainability 12 (22) (2020) 9726.
[36] M. Saatcioglu, A. Ghobarah, I. Nistor, Performance of structures in Indonesia during the December 2004 Great Sumatra earthquake and Indian ocean tsunami,
Earthq. Spectra 22 (2006) 295–319.
[37] M. Saatcioglu, G. Ozcebe, Response of reinforced concrete columns to simulated seismic loading, ACI Struct. 86 (1) (1989) 3–12.
[38] D.R. Sahoo, D.C. Rai, Design and evaluation of seismic strengthening techniques for reinforced concrete frames with soft ground story, Eng. Struct. 56 (2013)
1933–1944.
[39] H. Sezen, A.S. Whittaker, K.J. Elwood, K.M. Mosalam, Performance of reinforced concrete buildings during the August 17, 1999 Kocaeli, Turkey earthquake, and
seismic design and construction practice in Turkey, Eng. Struct. 25 (2003) 103–114.
[40] R.D.J.M. Steenbergen, A. Rózsás, A.C.W.M. Vrouwenvelder, Target reliability of new and existing structures - a general framework for code making, Heron 63
(3) (2018) 219–242.
[41] S.S. Swain, S.K. Patro, Seismic protection of soft storey buildings using energy dissipation device, in: Vasant Matsagar (Ed.), Advances in Structural Engineering.
Dynamics, vol. 2, 2015, https://doi.org/10.1007/978-81-322-2193-7_102:1311-1338.
[42] C.M. Uang, V.V. Bertero, Evaluation of seismic energy in structures, Earthq. Eng. Struct. Dynam. 19 (1) (1990) 77–90.
17