Estimating Direct and Indirect Losses Due To Earthquake Damage in Residential RC Buildings
Estimating Direct and Indirect Losses Due To Earthquake Damage in Residential RC Buildings
Estimating Direct and Indirect Losses Due To Earthquake Damage in Residential RC Buildings
A R T I C LE I N FO A B S T R A C T
Keywords: The downtime due to earthquake damage in RC buildings is examined. The rational component of downtime is
Seismic performance assessment evaluated based on the results of probabilistic seismic loss assessment analyses, using suitable time estimating
RC frame buildings manuals for civil construction works. The irrational component of downtime is derived from empirical data
Downtime relevant to post-earthquake reconstruction in Italy. Suitable criteria are then adopted to monetize the downtime
Indirect losses
for indirect loss estimation. Finally, a simplified model, expressing indirect monetary loss as a function of the
Expected annual loss
earthquake intensity level, is tentatively proposed, to be incorporated in practice-oriented procedures for the
estimation of the expected annual loss of residential RC buildings.
∗
Corresponding author. School of Engineering, University of Basilicata, Viale dello Ateneo Lucano 10, 85100, Potenza, Italy.
E-mail addresses: [email protected] (D. Cardone), amedeo.fl[email protected] (A. Flora), [email protected] (M. De Luca Picione),
[email protected] (A. Martoccia).
https://doi.org/10.1016/j.soildyn.2019.105801
Received 6 March 2019; Received in revised form 27 June 2019; Accepted 9 August 2019
0267-7261/ © 2019 Elsevier Ltd. All rights reserved.
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
2
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
(ZL) limit state, Sa,ZL, and the corresponding initial loss threshold, single walls of 100 mm thickness each, separated by a cavity of
q = MLZL/RepC, where MLZL is the monetary loss at the zero loss limit 100 mm. Internal partitions are realized with a single layer of hollow
state [11] and RepC is the replacement cost of the building. clay bricks with 100 mm thickness. The layout of the internal partitions
The ZL limit state corresponds to the onset of damage in the ma- is shown in Fig. 4.
sonry infills of a critical storey of the building. In first approximation, All the case-study buildings are supposed to be located in the city of
the ZL limit state can be deemed to be attained when a maximum in- L’Aquila (central Italy) on stiff soil (soil type A), which is characterized
terstorey drift of the order of 0.075–0.1% is reached. The aforesaid by the highest levels of seismic hazard for Italy (0.452 g PGA with 2475
threshold limit represents the median value of interstory drift derived years return period on stiff soil). This choice was purposely done in
from fragility curves relevant to light damage (i.e. detachment of infill, order to exploit the experimental data on downtime available, as shown
light diagonal cracking) of masonry infills [12]. The slope md is derived later, for a number of real buildings damaged by the M6.3 earthquake
by fitting the performance points associated with a number of limit that occurred in 2009 with epicenter near L’Aquila.
states (e.g. Zero Loss, Operational and Damage Control), which are It is worth noting that the choice of locating all the buildings in the
determined through simple structural analysis (e.g. Response Spectrum same, high seismicity, region is not detrimental to the general validity
Analysis). of the downtime model proposed in this study. Indeed, in first ap-
The simplified expression for the estimation of EALR is as follows: proximation, the structural response associated with a given earthquake
intensity level (hence the corresponding damage scenario, repair efforts
k 0 ⎡⎛ m ⎛ 1−q 1 −k ⎤
EALR = K1 ⎢⎜
q+ ⎜⎛1 + ⎞ − 1⎟⎞⎟⎞⎥, and downtime) can be assumed to be substantially the same, regardless
Sa, ZL ⎝ 1−k ⎝⎝ m ⎠ ⎠ ⎠⎦ (2) the building site and neglecting the return period of the seismic event.
⎣
Minor differences are expected due to different criteria in ground mo-
where k0 and k represent the intercept and the slope, respectively, of a tion selection.
line fit to the site hazard curve [11], that can be determined as follow: The structural response and direct loss assessment of the buildings
under consideration were carried out in previous studies [15]. In the
k=
Log ( )
λULS
λ SLS
, next paragraph, the main results of such studies are summarized since
Log ( )
Sa,ULS
Sa,SLS (3)
they are exploited in the heuristic approach for the estimation of
downtime presented in Section 4.
k
k0 = λ SLSSa,SLS , (4)
3.1. Summary of previous studies
where λULS and λSLS are the mean annual frequency of exceeding any
serviceability (SLS) and ultimate (ULS) limit state imposed by the re- A time-based performance assessment of the case-study buildings
ference seismic code, and Sa,SLS and Sa,ULS are the corresponding spec- examined in this paper was performed with PACT [12] to estimate the
tral accelerations at the effective fundamental period of vibration of the expected direct losses due to building repair in the event of an earth-
building (T*). quake. Nine different earthquake intensity levels, with return period TR
Equation (2) can be considered as a practice-oriented tool for a ranging from 30 to 2475 years were considered. For each intensity
large-scale estimation of EALR, which may be very useful for a rapid level, ten ground motion pairs were selected to be compatible (on
seismic performance assessment of a building stock with similar char- average) with given Conditional Mean Spectra [16], considering a
acteristics (e.g. all the residential RC buildings realized in Italy in the suitable M-R-ε (Magnitude-Distance-Epsilon) disaggregation and at-
1950s through to the early 1970s). tenuation relationship deemed appropriate for the city of L’Aquila.
In the present form, the simplified procedure by Cardone et al. [9] Extensive NRHA on accurate lumped-plasticity models [12,14,15] were
provides an estimate of the EAL associated with building repair (EALR) carried out to evaluate the structural response of the building models at
only. The contribution due to downtime (EALD) is neglected. One of the increasing seismic intensity levels. The seismic demand to the building
main goals of this paper is to further develop the proposed simplified was derived considering 500 realizations (for each earthquake intensity
procedure (see Eq. (2)) to incorporate indirect losses related to down- level) generated in PACT through Monte Carlo simulations on the basis
time. of the engineering demand parameters (maximum interstorey drifts and
peak floor accelerations) obtained from NRHA.
3. Case studies Five drift-sensitive Fragility Groups (FGs) were defined in PACT to
account for damageable structural components, which include: external
A set of archetype RC frame buildings with number of storeys ran- beam-column joints, internal beam-column joints, ductile weak col-
ging from four to eight (labeled with 4A, 6A and 8A in Fig. 4) has been umns and brittle weak columns liable to shear failure. For each FG,
examined. The archetype buildings feature 27.0 m × 15.0 m plan di- three repairable damage states were identified, differing in the exten-
mensions and 3.0 m inter-storey height. All the archetype buildings sion and gravity of damage, hence repair actions required to restore the
have the same structural configuration (with internal frames in the long structural component to its pre-earthquake (essentially undamaged)
direction only), with RC columns tapering along the height of the state. Similarly, for damageable non-structural components, six drift-
building. All the buildings feature a dog-leg stair with cantilever steps sensitive and one acceleration-sensitive FGs were defined (exterior
sustained by two stiff ‘flight’ beams. The structural characteristics of the masonry infill walls with and without openings, interior partitions with
archetype buildings (geometric dimensions, reinforcement ratios, and without openings, partition-like non-structural components and
structural details, etc.) have been derived from a simulated design, generic acceleration-sensitive non structural components). For each FG,
considering gravity loads only, according to the technical standard and a set of specific fragility functions was developed by the same authors
state of practice enforced in Italy before ’70s [13]. [17] and then implemented in PACT to relate the levels of drift (or
Cross section dimensions and reinforcement ratios of beams and acceleration) the component will experience during an earthquake to
columns are reported in detail in [14] for each building under con- the probability of that component will attain a certain Damage State
sideration. Steel reinforcement is realized with smooth steel rebars with (DS). Finally, a set of specific consequence functions (loss functions),
end-hooks in the exterior beam-column joints and at the base of the relating the quantity of elements to be repaired with the associated
columns. As far as the strength of materials is concerned, an average repair cost, was used in PACT to estimate expected direct losses.
compression strength of 25 MPa and a yield strength of 325 MPa have Building repair was deemed to be economically and practically not
been assumed for concrete and steel, respectively. Finally, the buildings feasible when residual drifts exceeded 1% [2]. This was taken into
feature external infills made of hollow clay bricks arranged in two account assuming a lognormal residual drift fragility function with
3
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
Fig. 4. Typical layout (plan view, front and lateral views) of the archetype buildings (4A, 6A and 8A).
median value of 1% and dispersion of 0.3 [12]. A repair loss thresholds 4.1. Damage scenarios
equal to the replacement cost of the building (RepC) was assumed, to
impose a cap to the building repair efforts. The Replacement Cost of The starting point for the definition of the damage scenarios re-
each building was estimated based on the current average construction quired for downtime estimation were the results of seismic performance
cost per square meter for new residential buildings with similar volume assessment obtained from previous studies (see paragraph 3.1). In
and characteristics (equal to 730 €/m2 according to CIAMI [18]). In particular, in order to account for the uncertainties associated with the
addition, a cost per cubic meter for demolition and waste disposal of 44 seismic demand imposed to the building, three inter-storey drift and
€/m3 was considered [19]. The building collapse was taken into ac- floor acceleration profiles have been selected for each earthquake in-
count by means of specific collapse fragility functions derived based on tensity level, corresponding to the 16th, 50th and 84th percentiles of
the results of pushover analysis, using the SPO2IDA (Static PushOver to the 500 realizations carried out in PACT.
Incremental Dynamic Analysis) tool provided in PACT, and assuming a The selected drift and acceleration profiles have been used to de-
total dispersion of 0.6. termine the number (percentage) of components of any fragility group
Results from rigorous probabilistic loss assessment with PACT [11] reaching a given DS in any direction of each storey of the building, by
pointed out values of the EAL associated with repair/replacement costs entering the associated fragility functions, as shown in Fig. 5(a). In the
(EALR) ranging from 0.98% RepC (for the building 4A) to 0.72% (for example of Fig. 5(a), valid for masonry infill walls without openings, a
the building 8R). drift value of the order of 1.2% produces a percentage of components
experiencing a DS2 (extensive diagonal cracking) of the order of 28%, a
percentage of components featuring a DS3 (corner crushing) of the
4. Heuristic approach for downtime estimation order of 55% and a percentage of components experiencing a DS4
(collapse) of the order of 12%. There are no elements damaged ac-
In this section, the downtime of the selected case study buildings cording to the DS1 (light cracking).
(4A, 6A and 8A) is evaluated following a heuristic approach, based on For each earthquake intensity level, the collapse fragility curve of
the use of suitable time estimating manuals for civil construction works. the building has been used to derive the probability of collapse and
The heuristic approach under consideration is based on the results of non-collapse, as shown in the example of Fig. 5(b), valid for the 4A
the probabilistic seismic assessment described in Section 3.1. An ap- building model, where a probability of collapse of 44% and a prob-
proximate relationship relating downtime with earthquake intensity ability of non-collapse of 56% is registered for a spectral acceleration of
level is then derived. 0.6 g.
It is anticipated that, strictly speaking, the proposed relationship is
valid only for pre-70 RC frame residential buildings, with character- 4.2. Downtime estimation
istics representative of the construction practice adopted in Italy in the
1950s through to the early 1970s. However, it can be easily extended to Downtime is influenced by many factors, e.g.: building inspection,
other building types following the same heuristic approach. damage assessment, finance planning, professional consultancies, pos-
sible competitive bidding process and, finally, the difficulty and
4
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
Fig. 5. (a) Use of fragility curves for the definition of the damage scenarios (b) probability of collapse and non-collapse for a given earthquake intensity level.
extension of the repair activities needed to return the building to its First of all, the expected replacement time (E[R'T|C,IM]) has been
undamaged state. estimated in terms of man-days unit and then converted in labour days
Usually, two components of downtime are identified, namely: the considering 8 hrs per day working crew, and a number of workers equal
rational component and the irrational component. The former is the to 2 per 50 m2.
period of time needed to complete the building repair activities, from Next, the expected total repair time conditioned on a specific in-
the day in which the activities actually starts until the day when they tensity measure (E[RT|NC,IM]) has been computed with the following
are actually over. The latter (often referred to as mobilization time expression valid for the slow-track repair scheme:
[20]) includes a series of preliminary activities, such as: damage as-
sessment, occupants/activities relocation, bureaucratic issues, requests E [RT |NC , IM ]slow − track = ∑ E [RU⁎ (m)|NC, IM ] + E [CS |NC, IM ],
m (6)
for financing, technical consultation, design of interventions, etc..
Considering the high level of uncertainties, the evaluation of the irra- In Eq. (6), the building repair time (E[RU∗(m)|NC,IM]), measured in
tional component of downtime is considered tricky and sensitive. calendar days from the date on which repair work begun, is calculated
According to Mitrani-Raiser [20], the expected building downtime, by summing up the expected repair times of each operational unit:
for a given hazard level, can be calculated using downtime estimates for
collapse and non-collapse cases. As a consequence, the expected value E [RU* (m)|NC , IM ] = ∑ tr (FGi, DSj)⋅Qr (FGi, DSj),
i, j (7)
of downtime, conditioned on a given intensity measure (IM), can be
expressed as follows: where Qr(FGi,DSj) is the number of components included in the i-th
Fragility Group (FG), located in the m-th operational unit, and featuring
E [Downtime IM ] = E [R′T |C , IM ]⋅P [C|IM ] + E [RT |NC , IM ] the j-th Damage State (DS). Qr(FGi,DSj) has been derived as described in
⋅P [NC|IM ] + E [RT 0 |IM ] + NT ⋅E [RCOT ] (5) the paragraph 4.1, using the EDPs associated with given realizations
(16th, 50th and 84th percentile) and suitable fragility functions for
where: each fragility group. tr(FGi,DSj) represents the total repair time required
to restore the damaged components of the i-th FG to their undamaged
• E[R' |C,IM] represents the expected replacement time needed to
T state, starting from the j-th Damage State.
demolish and rebuild the building in case of severe or extensive The interval of time needed to complete each repair activity has
damage, conditioned on a given IM. Such value can be calculated by been derived from time estimating manuals for civil construction works
summing the demolish time and the reconstruction time; [21]. The total repair time tr(FGi,DSj) has been obtained by combining
• E[RT|NC,IM] represents the expected total repair time conditioned the repair times associated with each activity, also considering the
on the structure not collapsing (NC), conditioned on a given IM; possibility of performing two or more sub-activities in parallel. An ex-
• E[RT0|IM] is the mobilization time conditioned on a certain IM; ample of repair activities and corresponding repair times for different
• NT and E[RCOT] represent the number of change-of-trade and the damage states of External RC beam/column joints and External ma-
average time associated with the change-of-trade delay, respectively sonry infill walls are reported in Tables 1 and 2, respectively.
• P[C|IM] and P[NC|IM] represent the probability of collapse and The second term in Eq. (6) represents the construction site time, i.e.
non-collapse conditioned on a given IM, respectively, obtained en- the time needed to assemble and dismantle the construction site. It has
tering the collapse fragility function of the building with the value of been calculated as follows:
IM (like in example see Fig. 5(b)).
E [CS |NC , IM ] = tcs + tes, (8)
For what concerns the expected total repair time (E[RT|NC,IM]),
where tcs represents the time required for material/equipment hand-
considering each floor as an independent operational unit (m), a slow-
ling, installation/removal of access protection, application/removal of
track scheme, in which the operational units are supposed to be re-
curtains against dust, installation/disassembly of work platforms, waste
paired in series, has been assumed as repair strategy in this study.
disposal, etc.; tes represents the time needed to assembly/disassembly
external scaffolding. Both tcs and tes have been derived from Beck et al.
4.2.1. Rational component of downtime [22], and expressed in terms of man-days per floor.
The first two terms of Equation (5) represents the rational compo- Obviously, the repair time depends on the number of workers in-
nent of downtime. volved in the repair activities. In line with the current practice in Italy,
5
D. Cardone, et al.
Table 1
Damage States, Repair Activities and Repair times for external RC beam/column joints.
Damage state Damage state description Repair activities Unit of Unit of time (min/unit Area/volume Time per element Total time per
measure measure) involved (min) element (hours)
DS1 Light cracking at beam/column-joint interfaces (< 1–1.5 mm), yielding of Floor and screed demolishing m2 14.4 2.25 32.4 7.2
beam rebars. Possible first inclined cracks in EWJ. Plaster removing m2 9 3.7 33.3
Concrete cover removing and m2 12 1.44 17.28
cleaning
Epoxy injection m2 192 0.7 134.4
(cracks < 1.2 mm)
Screed restoring m2 9.6 2.25 21.6
Floor restoring m2 48 2.25 108
Plaster restoring m2 15.6 3.7 57.72
Painting m2 7.3 3.7 27.01
DS2 Severe cracking (≥3–5 mm), often localized at the beam/column-joint Floor and screed demolishing m2 14.4 3 43.2 13.67
interface in a single wide crack. Possible spalling of cover concrete. Plaster removing m2 9 5.1 45.9
Epoxy injection m2 216 1.4 302.4
6
(cracks > 1.2 mm)
Concrete cover removing and m2 12 2 24
cleaning
Concrete patching with mortar m2 57.6 2 115.2
mix
Screed restoring m2 9.6 3 28.8
Floor restoring m2 48 3 144
Plaster restoring m2 15.6 5.1 79.56
Painting m2 7.3 5.1 37.23
DS3 Spalling of cover concrete, possible crushing of concrete at beam/column- Floor and screed demolishing m2 14.4 4 57.6 28.16
joint interface. Possible buckling of column rebars. Plaster removing m2 9 6.7 60.3
Complete element restoringa m2 396 3 1188
Screed restoring m2 9.6 4 38.4
Floor restoring m2 48 4 192
Plaster restoring m2 15.6 6.7 104.52
Painting m2 7.3 6.7 48.91
a
includes concrete restoring with mortar mix, steel jacketing and epoxy injections.
Soil Dynamics and Earthquake Engineering 126 (2019) 105801
D. Cardone, et al.
Table 2
Damage States, Repair Activities and Repair times for external masonry infills.
Damage state Damage state description Repair activities Unit of Unit of time (min/unit Area/volume Time per element Total time per element
measure measure) involved (min) (hours)
DS1 Separation of the infill from the frame. Light diagonal cracking Ceramic tiles removing m2 13.5 1.9 25.65 6.56
(width < 1 mm) Plaster removing m2 9 3.8 34.2
Fibre-glass reinforcing mesh m2 9.6 3.8 36.48
installation
Ceramic tiles installation m2 27.6 1.9 52.44
Plaster restoring m2 15.6 3.8 59.28
Painting m2 7.3 25.4 185.42
DS2 Extensive diagonal cracking Ceramic tiles removing m2 13.5 3.2 43.2 9.23
(1 mm < width < 2 mm). Plaster removing m2 9 6.3 56.7
Possible failure of brick units. Loosen panel area demolishing m2 16.8 1.27 21.34
Fibre-glass reinforcing mesh m2 9.6 6.3 60.48
installation
Plaster restoring m2 15.6 6.3 98.28
7
Painting m2 7.3 25.4 185.42
Ceramic tiles installation m2 27.6 3.2 88.32
DS3 Corner crushing, Brick spalling, Detachment of large plaster area. Complete panel demolition m2 16.8 12.7 213.36 38.17
Sliding in the mortar joints Panel reconstruction m2 96 12.7 1219.2
Electrical service restoring ml 7.2 4.7 33.84
Hydraulic service restoring ml 13.14 4.7 61.76
Plaster restoring m2 15.6 25.4 396.24
Painting m2 7.3 25.4 185.42
Mopboard installation ml 9 4.7 42.3
Ceramic tiles installation m2 27.6 5 138
DS4 In-plane or out-of-plane collapse Complete panel demolition m2 16.8 12.7 213.36 38.17
Panel reconstruction m2 96 12.7 1219.2
Electrical service restoring ml 7.2 4.7 33.84
Hydraulic service restoring ml 13.14 4.7 61.76
Plaster restoring m2 15.6 25.4 396.24
Painting m2 7.3 25.4 185.42
Mopboard installation ml 9 4.7 42.3
Ceramic tiles installation m2 27.6 5 138
Soil Dynamics and Earthquake Engineering 126 (2019) 105801
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
two workers per crew and a floor area of 50 m2/crew has been assumed
Repair time + Construction site time
for all the repair activities, except for assembly/disassembly operations
concerning external scaffoldings, for which a ratio of 80 m2/crew/day
the 6th of April 2009 differing in terms of number of storeys and da-
mage level (hence post-earthquake repair activities).
The main characteristics of the selected real buildings are sum-
(days)
10)
10)
10)
09)
11)
ings under consideration has been estimated with the simplified ex-
Empirical data collected from building contractors involved in repair activities of pre-70 RC frame buildings damaged during the 2009 L'Aquila earthquake.
FRP
FRP
FRP
FRP
obtained values with the Square Root of the Sum of Squares (SRSS) rule.
In Table 3, information relevant to the actual repair time for each
Cracking of beam/column joints; diagonal cracks in external joints. Detachment of
external joints. Extensive diagonal cracking and failure of brick units in masonry
building are reported, including: (i) the interval of time needed for
Severe cracking of beam/column joints. Spalling of cover concrete. Extensive
Development of plastic hinges in beams and columns, extensive damage in
cratic issues ecc. (i.e. the so-called mobilization time) and (ii) the time
required for construction site and repair activities. A brief description
on the damage scenario observed and main retrofit strategy adopted for
each building is also provided.
In Fig. 6, the results in terms of rational component of downtime
obtained following the proposed heuristic approach are compared with
floor. Extensive damage of masonry infills. (B–C)
infills (B–C)
0.68
0.68
1.02
1.02
1.36
1.36
T*
1962–1971
1962–1971
1962–1971
1962–1971
1962–1971
time can be roughly estimated based on the global damage state of the
building, defining three tags (green, yellow and red) related to the
4
damage level (from low to severe) of the building. The ATC-20 guide-
Long (°)
13.37
13.37
13.41
13.41
13.41
13.40
42.36
42.36
42.36
42.35
42.34
ID
8
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
Fig. 6. Comparison between estimates obtained from the proposed heuristic approach and empirical data provided by building contractors for (a) 4-, (b) 6- and (c) 8-
storey buildings (see Table 3).
9
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
10
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
Fig. 7. 16th, 50th and 84th percentile values of downtime vs earthquake intensity level.
Table 5
Expected indirect losses conditioned on seismic intensity for each case-study
building, Lower Bound and Upper Bound conditions.
Case IM Replacement Replacement cost E[Indirect E[Indirect
Study time (days) (€) losses|IM] losses|IM]
Lower Upper Bound
Bound (€)
(€)
11
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
Table 6 estimation presented in Cardone et al. [11] (see Fig. 3) to account for
Expected Annual Loss due to repair/replacement costs (EALR), downtime monetary loss due to downtime.
(EALD) and total Expected Annual Loss (EALT) for each case-study building, This could represent an important refinement of the simplified
Lower bound and Upper bound conditions. procedure under consideration. However, further studies are needed to
Building EALD, EALD, EALR EALT, EALT, validate and further substantiate the proposed relationship based on the
Lower Upper bound (%RepC) Lower bound Upper bound results of comprehensive parametric analyses, involving more case
bound (%RepC) (%RepC) (%RepC) studies, buildings sites and soil conditions, as well as a significant
(%RepC)
amount of experimental data derived in the post-earthquake re-
4A 0.17% 0.29% 0.98% 1.15% 1.27% construction of past seismic events.
6A 0.22% 0.36% 0.81% 1.03% 1.17%
8A 0.24% 0.38% 0.68% 0.92% 1.06% 6. Conclusions and further developments
12
D. Cardone, et al. Soil Dynamics and Earthquake Engineering 126 (2019) 105801
13