Conference PaperPDF Available

OPEN MODELS AND SOFTWARE FOR ASSESSING THE VULNERABILITY OF THE EUROPEAN BUILDING STOCK

Authors:
OPEN MODELS AND SOFTWARE FOR ASSESSING THE
VULNERABILITY OF THE EUROPEAN BUILDING STOCK
Helen Crowley1, Vitor Silva2, Luis Martins2, Xavier Romão3 and Nuno Pereira3
1 EUCENTRE
Via Ferrata 1, Pavia 27100, Italy
e-mail: helen.crowley@eucentre.it
2 GEM Foundation
Via Ferrata 1, Pavia 27100, Italy
vitor.silva@globalquakemodel.org, luis.martins@globalquakemodel.org
3 University of Porto
Rua Dr. Roberto Fria, 4200-465 Porto, Portugal
xnr@fe.up.pt, nmsp@fe.up.pt
Abstract
Reproducibility and transparency have become standard practice in earthquake loss modelling
following a growing demand from both the public and private sectors for understandable and
accessible risk model input data, software and results. Since the 1990’s the European scientific
community has worked together on many aspects of seismic hazard and risk modelling. These
advances have led to the 2020 European Seismic Risk Model (ESRM20), which has been com-
puted with open source software and is now being openly released to the wider scientific com-
munity. This paper presents the recently released input models and software to assess the
vulnerability of the European building stock within ESRM20.
Keywords: Vulnerability, Fragility, European Exposure, Open-Source, Open Data.

"&-#(' +'('
(&).--#('%-"(,#'-+.-.+%/'&#,'+-"*.$'!#'+#'!
)+$$#,+!#$#,,
-+& +(&-"',+.'
2678
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
1INTRODUCTION
Reproducibility and transparency have become standard practice in earthquake loss model-
ling following a growing demand from both the public and private sectors for understandable
and accessible risk model input data, software and results. Since the 1990’s the European sci-
entific community has worked together on many aspects of seismic hazard and risk modelling.
These advances have led to the 2020 European Seismic Risk Model (ESRM20), which has been
computed with open source software and is now being openly released to the wider scientific
community (Crowley et al., 2019 [1]). This paper summarises the recently released input mod-
els and software to assess the vulnerability of the European building stock within ESRM20.
2CAPACITY CURVES FOR EUROPEAN BUILDING STOCK
Capacity curves provide a description of the lateral strength and deformation capacity of
buildings or building classes, and are often transformed to the ADRS (acceleration displace-
ment response spectrum) format for the purposes of developing fragility functions.
For ESRM20, capacity curves for a large range of building classes are needed to cover the
varying construction types in Europe (as included within the European exposure model: Crow-
ley et al., 2020a [2]; Crowley et al., 2020b [3]). The GED4ALL Building Taxonomy has been
used to classify the vulnerability of European buildings (Brzev et al., 2021 [4]) with the attrib-
utes summarised below:
Materials. CR: reinforced concrete, MR: reinforced masonry, MCF: confined ma-
sonry, MUR: unreinforced masonry, MUR-ADO: adobe, MUR-CB99: concrete
block masonry, MUR-CL99: clay brick masonry, MUR-STDRE: dressed stone ma-
sonry, MUR-STRUB: rubble stone masonry, S: steel, W: wood/timber.
Lateral load resisting systems. LDUAL: dual frame-wall system, LFINF: infilled
frame, LWAL: load bearing wall, LFM: moment frame, LFBR: braced frame.
Code Level or Ductility. CDN: absence of seismic design, CDL: low code level (de-
signed for lateral resistance using allowable stress design), CDM: moderate code
level (designed for lateral resistance with modern limit state design), CDH: high code
level (designed for lateral resistance coupled with target ductility requirements and
capacity design), DNO: non-ductile, DUL: low ductility, DUM: moderate ductility,
DUH: high ductility.
Height. H: number of storeys.
Lateral Force Coefficient. The value of the lateral force coefficient, i.e. the fraction
of the weight that was specified as the design lateral force in the seismic design code
(see Code Level), expressed in % (applied to reinforced concrete moment and infilled
frames only).
Following their use in the calculation of their Global Seismic Risk Map (GEM, 2018), the
GEM Foundation has released a global database of capacity curves (Martins and Silva, 2020
[5]) which have been derived through the compilation of data coming from research studies and
experimental campaigns. These capacity curves have been used to represent the European
CR_LDUAL, CR_LWAL, MCF, MR, MUR, S and W typologies with different heights and
ductility, for a total of 217 building classes.
As part of the European SERA project (www.sera-eu.org), a detailed set of capacity curves
for European reinforced concrete infilled frames (CR_LFINF) and moment frames (CR_LFM)
has been recently developed (Romão et al., 2019 [6]). A total of 264 reinforced concrete classes
have been identified by combining different numbers of storeys (1 to 6), seismic design code
levels (no code: CDN, low code: CDL, moderate code: CDM, high code: CDH) and lateral
force coefficient levels (0, 5, 10, 15, 20, 25, 30 % of the weight of the structure). Buildings of
2679
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
design class CDN were typically designed to older codes (from before the 1960’s) that used
allowable stresses and very low material strength values and considered predominantly the
gravity loads. Buildings of design class CDL were designed considering the seismic action by
enforcing values of the seismic coefficient, β (referred to herein as lateral force coefficient).
Structural design for these codes was typically based on material-specific standards that used
allowable stress design or a stress-block approach. Seismic design including modern concepts
of ultimate capacity and partial safety factors (limit state design) was the basis of the CDM
category of codes. The seismic action was also accounted for in the design by enforcing values
for the lateral force coefficient, β. Finally, the CDH class refers to modern seismic design prin-
ciples that account for capacity design and local ductility measures, similar to those available
in Eurocode 8 (CEN, 2004 [7]). Seismic zonation maps associated with the seismic design codes
employed in Europe over the last century can be used to identify the lateral force coefficient
(i.e. the fraction of the weight that was specified as the design lateral force in the seismic design
code, expressed in %) (see Crowley et al., 2021 [2] and Crowley et al., 2020c [9]).
The capacity curves for the 264 building classes were developed through simulated design
of prototype frames (see e.g. Borzi et al., 2008 [10]; Verderame et al., 2010 [11]) and then
nonlinear analysis has been undertaken to obtain the backbone capacity curves of these frames.
Up to 300 capacity curves have been simulated per class by modifying the geometrical and
material properties of the prototype frames, and thus accounting for the building-to-building
variability in the simulated design. Figure 1 shows the median capacity curves for reinforced
concrete frames with masonry infills not designed with seismic loads (CDN).
Figure 1: Median capacity curves from Romão et al. (2020) [12] for reinforced concrete frames with masonry
infills (CR/LFINF) not designed with seismic loads (CDN)
All of the median capacity curves for the 481 building classes are publicly available on a
GitLab repository (Romão et al. 2020 [12]).
3VULNERABILITY MODELLING
3.1Vulnerability Modellers’ Toolkit (VMTK)
The fragility functions of these European building classes have been computed using the
Vulnerability Modeller’s Toolkit, a resource that has been developed and released by the GEM
Foundation in collaboration with members of the European risk community (Martins et al.,
2021 [13]). This toolkit is a set of Python scripts that read the capacity curves, produce SDOF
2680
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
hysteretic models, launch OpenSeesPy1 to run nonlinear dynamic analysis, apply linear cen-
sored regression to the cloud of nonlinear responses, and compute fragility functions for differ-
ent damage states, based on the user-defined damage state thresholds. A graphical user interface
is included with the toolkit and includes a set of default assumptions, allowing less experienced
users to interact with the VMTK (Figure 2). Experienced users are instead encouraged to make
use of Python’s scripting capabilities to explore all the features of the VMTK source code and
to contribute to future releases of the toolkit. The complete toolkit, including source code and
GUI, is currently hosted in a publicly available GitHub repository2.
Figure 2: Graphical user interface of the fragility module of the VMTK (Martins et al., 2021 [13])
3.2Fragility Functions
The median capacity curves in Romão et al. 2020 [12] have been read by the VMTK’s
‘nlth_on_sdof.py’ script which has been used to produce SDOF models using the Pinching4
hyseresis curve of OpenSeesPy. Pinghing4 is a uniaxial material that represents a ‘pinched’
load-deformation response and exhibits degradation under cyclic loading. In addition to the
response envelope (taken from the capacity curves), the hysteretic properties given in Table 1
have been assumed for the SDOF models for the different material types. Cyclic degradation
of the strength and stiffness is modelled for the CR, MUR, MCF and MR typologies through
unloading stiffness degradation, reloading stiffness degradation, and strength degradation.
Mass proportional damping is used with different damping ratios for each typology. Typically,
masonry is assigned 10% damping, reinforced concrete and wood is 5%, and steel is 3%. The
nlth_on_sdof.py’ script runs nonlinear dynamic analysis using each SDOF and a set of records.
A database of recordings has been compiled for the nonlinear dynamic analyses using records
with PGA above 0.05g in the Engineering Strong Motion (ESM, Luzi et al 2016 [14]; Luzi et
al. 2020 [15]) and NGA (Chiou et al. 2008 [16]) databases. Records have then been selected
from this database to match a range of intensity measure bins with a maximum scaling factor
of 2.
1 https://openseespydoc.readthedocs.io/en/latest/index.html
2 https://github.com/GEMScienceTools/VMTK-Vulnerability-Modellers-ToolKit
2681
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
The SDOF properties reported above are currently being further tested and modifications
might still be incorporated in the final European fragility models. The final scripts used to de-
velop the models will be made available on the aforementioned GitLab repository (Romão et
al. 2020 [12]) so that are final assumptions are transparent and it will be possible to easily
reproduce the models.
Table 1: Adopted parameters of the Pinching4 hyseresis model
Parameters of Pinching4* CR, MUR, MCF, MR W, S
rdispP/N, fForceP/N, uForceP/N 0.5, 0.25, 0.05 0.5, 0.25, 0.5
gK1 gK2 gK3 gK4 gKLim 0, 0.1, 0, 0, 0.2 0,0,0,0,0
gD1 gD2 gD3 gD4 gDLim 0, 0.1, 0, 0, 0.2 0,0,0,0,0
gF1 gF2 gF3 gF4 gFLim 0, 0.4, 0, 0.4, 0.9 0,0,0,0,0
gE 10 10
dmgType energy energy
*See https://openseespydoc.readthedocs.io/en/latest/src/Pinching4.html?highlight=pinching4 for definition of the parameters
The ‘fragility_censored_cloud_analysis.py’ script in the VMTK uses the nonlinear response
outputs from each dynamic analysis together with damage thresholds to apply linear censored
regression, and this is then used to compute fragility functions (lognormal distributions) for
different damage states, based on the user-defined damage state thresholds. For the European
fragility functions, the damage thresholds presented in Figure 3 have been assumed.
Figure 3: Damage thresholds assumed in the development of the fragility functions (Martins and Silva, 2020 [8])
Four different intensity measure types have been used in the censored regression: PGA, Sa(0.3),
Sa(0.6) and Sa(1.0). An average spectral acceleration (see e.g. Baker and Cornell (2006) [17];
Eads et al., 2015 [18]) with 20 intervals between 0.05 and 3 seconds is also currently being
investigated. The final intensity measure for each building typology is taken as that with the
lowest lognormal dispersion in the fragility function given that this is related to efficiency (i.e.
low dispersion in the nonlinear response, given the intensity measure), and it has been found
by checking the sufficiency (i.e. conditional independence of the distribution of nonlinear re-
sponse, given IM, on other parameters of the ground motion) of the different intensity measures
that the most efficient is typically also sufficient. Some recent studies have shown that the
higher the efficiency, the higher the sufficiency of the intensity measure (e.g. Bradley et al.
2010 [19]). Others have cautioned that the typical checks for sufficiency (e.g. Luco and Cornell,
2007 [20]) only provide evidence rather than proof of sufficiency and that it should be ensured
2682
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
that the intensity measure is also efficient to ensure that the additional parameters really are
having a significant influence on the response (Kazantzi and Vamvatsikos, 2015 [21]). For il-
lustration purposes, Figure 4 shows the fragility functions in terms of Sa(0.3s) that have been
obtained with the capacity curves presented in Figure 1.
Figure 4: Fragility functions calculated using the capacity curves presented in Figure 1, where DS1 = slight dam-
age, DS2 = moderate damage, DS3 = extensive damage and DS4 = complete damage
3.3Vulnerability functions for cost of repair/replacement
The fragility functions are converted into vulnerability models using damage-loss models
which provide damage ratios for each damage state (slight, moderate, extensive and complete).
For losses due to the repair of damage, the damage ratios (which represent the ratio of cost of
repair to cost of replacement) from HAZUS (FEMA, 2004 [22]) are being adopted, i.e. 0.02
(slight damage), 0.1 (moderate damage), 0.5 (extensive damage), 1.0 (complete damage). These
values are being further tested and calibrated using the results of the tests described in Section
4 and may be adapted before the final release of ESRM20. For a range of intensity measure
levels, the probability of occurrence of each damage state is obtained from the fragility func-
tions, multiplied by the damage ratios and summed leading to a mean loss ratio. A lognormal
cumulative distribution function has then been fit to the vulnerability data to obtain the median
and lognormal standard deviation (dispersion).
In order to account for the building-to-building variability, which is not accounted for given
the use of the median capacity curves, a separate study to produce vulnerability functions using
all available capacity curves for the CR/LFINF and CR/LFM typologies has been undertaken.
This study showed that the mean vulnerability function based on all of the capacity curves of a
given typology had a very similar median to the vulnerability function based on the median
capacity curves. However, the dispersion was, as expected, slightly higher. It was found that
the additional dispersion required to account for building-to-building variability was around 0.3,
2683
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
and this is therefore combined with the dispersion obtained with the median capacity curves
through the square-root-sum-of-squares to produce the final vulnerability functions.
3.4Vulnerability functions for loss of life
For loss of life, the model uses a number of factors obtained from both past observations and
expert judgment, including the likelihood that a completely damaged building will collapse to
the extent that it could cause loss of life (currently taken as 1.5% based on the data from recent
earthquakes: Antonios Pomonis, personal communication), the probability of entrapment given
collapse (Reinoso et al. 2017 [23]), and the probability of loss of life given entrapment (Reinoso
et al. 2017 [23]). Different entrapment ratio for day and night are assumed, with higher values
for the latter given the increased time required for people to wake up and escape from the col-
lapsing building. Also, an increase in the entrapment ratio with number of storeys has been
implemented for the day-time entrapment rates. The loss of life given entrapment has been
obtained from the data in Table 2 of Reinoso et al. (2017 [23]) with a clear distinction between
buildings with less than and more than 5 storeys being observed and thus applied. Table 2 pre-
sents all of the assumed values.
Table 2: Adopted fatality model input parameters
Number of storeys P-entrapment (day) P-entrapment (night) P-loss-life|entrapment
1 0.25 0.95 0.4
2 0.5 0.95 0.4
3-4 0.75 0.95 0.4
>5 0.95 0.95 0.7
The fatality vulnerability models are obtained by multiplying the complete damage fragility
functions by 1.5% and the factors presented in Table 2. A lognormal function is then fit to the
data and the dispersion is increased to account for building-to-building variability (as presented
in the previous section).
3.5Vulnerability Viewer
A Dash App is currently under development to allow users to view and compare the fragility
and vulnerability functions for all of building classes and for all the available intensity measure
types. A prototype that presents the vulnerability functions in terms of Sa(0.3) is currently avail-
able at http://vulncurves.eu-risk.eucentre.it/ (see Figure 5).
2684
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
Figure 5: Screenshot of the vulnerability viewer currently under development
4VULNERABILITY TESTING
Extensive testing of the resulting fragility and vulnerability models has been undertaken by
comparing the modelled damage/losses with empirical models as well as with observations
from a large number of past European earthquake scenarios (Crowley et al. 2020d [24]).
Simple sanity checks, so-called ‘unit tests’ are undertaken to ensure the median and disper-
sion values are within sensible ranges, and to compare with existing functions from the litera-
ture. The VMTK includes a module that allows users to test their model’s against GEM’s global
vulnerability models, which have been calibrated or tested using past earthquake damage and
loss data. Useful, and openly available, data for this purpose includes the empirical vulnerability
models developed by PAGER (Jaiswal et al., 2009 [25]; Jaiswal and Wald, 2013 [26]), as well
as fatality, economic loss and damage data from various databases including the Centre for
Research on the Epidemiology of Disasters (CRED)’s EMDAT data-base (EMDAT, 2019 [27]),
the Italian Department of Civil Protection’s Da.D.O. data-base (Dolce et al., 2019 [28]),
NOAA’s Significant Earthquake Database (NGDC/WGS), and the Cambridge Earthquake Im-
pact Database (www.ceqid.org).
Figure 6 shows two examples of tests that have been undertaken with the European vulner-
ability model. In the first example, a mean vulnerability function calculated through an expo-
sure-weighted combination of all the building classes in the country has been produced and
compared with the empirical models developed by PAGER, following conversion of the spec-
tral ordinates to macroseismic intensity (with the associated uncertainty in the conversion
shown by the mean and +/1 standard deviation vulnerability curves). In the second example,
ground motion fields (modelled using scenario rupture models and the latest European ground
motion and site amplification models) have been developed for over 30 damaging historical
events above magnitude 5 in Europe since the 1980s. These ground motions are then combined
with the current exposure and vulnerability models to estimate direct economic losses. These
are then compared in a statistical sense with the reports on economic losses (that are corrected
to today’s value). Similar tests can also be undertaken for fatalities These tests provide a check
to ensure that probabilistic losses predicted by the risk model are not biased.
2685
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
Figure 6: Example tests of the European vulnerability model: (left) comparison with the PAGER vulnerabil-
ity model, where the three blue dashed lines represent the mean and +/- sigma, (right) comparison of estimated
and observed losses for damaging European events since 1980
These tests will continue to be undertaken until the finalization and release of ESRM20
which is expected in October 2021. As mentioned previously, all of the scripts and input data
described herein will then be released on the dedicated vulnerability GitLab repository (Romão
et al. 2020 [12]).
REFERENCES
[1]H. Crowley, D. Rodrigues, V. Silva, V. Despotaki, L. Martins, X. Romão, J.M. Castro,
N. Pereira, A. Pomonis, A. Lemoine, A. Roullé, B. Tourlière, G. Weatherill, K. Pitilakis,
L. Danciu, A.A. Correira, S. Akkar, U. Hancilar, P. Covi, The European Seismic Risk
model 2020 (ESRM20). 2nd International Conference on Natural Hazards and Infra-
structure, ICONHIC, Chania, Crete, Greece, 2019.
[2]H. Crowley, V. Despotaki, D. Rodrigues, V. Silva, D. Toma-Danila, E. Riga, A.
Karatzetzou, S. Fotopoulou, Z. Zugic, L. Sousa, S. Ozcebe, P. Gamba, Exposure model
for European seismic risk assessment, Earthquake Spectra, DOI:
https://doi.org/10.1177/8755293020919429, 2020a.
[3]H. Crowley, V. Despotaki, D. Rodrigues, V. Silva, C. Costa, D. Toma-Danila, E. Riga,
A. Karatzetzou, S. Fotopoulou, L. Sousa, S. Ozcebe, P. Gamba, J. Dabbeek, X. Romão,
N. Pereira, J. M. Castro, J. Daniell, E. Veliu, H. Bilgin, C. Adam, M. Deyanova, N. Ad-
emović, J. Atalic, B. Bessason, V. Shendova, A. Tiganescu, D. Toma-Danila, Z. Zugic,
S. Akkar, U. Hancilar, Exposure Contributors, European Exposure Model Data Reposi-
tory (Version 0.9) [Data set]. Zenodo. DOI: http://doi.org/10.5281/zenodo.4062044,
2020b.
[4]S. Brzev, V. Silva, L. Allen, C. Scawthorn, C. Yepes, J. Dabbeek, H. Crowley, A building
classification scheme for multi-hazard risk assessment. Submitted to Natural Hazards and
Earthquake System Sciences, 2021.
[5]GEM, Global earthquake maps. Available at: www.globalquakemodel.org/gem (accessed
21 March 2021), 2018.
[6]X. Romão, N. Pereira, J. M. Castro, F. De Maio, H. Crowley, V. Silva, L. Martins, Euro-
pean physical vulnerability models. SERA Deliverable D26.5, Available from URL:
http://static.seismo.ethz.ch/SERA/JRA/SERA_D26.5_Physical_Vulnerability.pdf, 2019.





    


2686
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
[7]CEN, Eurocode 8: Design of structures for earthquake resistance - Part 1: General rules,
seismic actions and rules for buildings. Comité Européen de normalization, Brussels,
Belgium, 2004.
[8]L. Martins, V. Silva, Development of a fragility and vulnerability model for global seis-
mic risk analyses. Bulletin of Earthquake Engineering, https://doi.org/10.1007/s10518-
020-00885-1, 2020.
[9]H. Crowley, V. Despotaki, V. Silva, J. Dabbeek, X. Romão, N. Pereira, J.M. Castro, J.
Daniell, E. Veliu, H. Bilgin, C. Adam, M. Deyanova, N. Ademović, J. Atalic, E. Riga, A.
Karatzetzou, B. Bessason, V. Sendova, D. Toma-Danila, Z. Zugic, S. Akkar, U. Hancilar,
Model of Seismic Design Lateral Force Levels for the Existing Reinforced Concrete Eu-
ropean Building Stock. Bulletin of Earthquake Engineering, in press, 2020d.
[10]B. Borzi, H. Crowley, R. Pinho, Simplified pushover-based vulnerability analysis for
large-scale assessment of RC buildings. Engineering Structures, 30, 804-820, 2008.
[11]G.M. Verderame, M. Polese, C. Marinello, G. Manfredi, A simulated design procedure
for the assessment of seismic capacity of existing reinforced concrete buildings. Advances
in Engineering Software, 41(2), 323- 335, 2010.
[12]X. Romão, N. Pereira, J. M. Castro, F. De Maio, H. Crowley, V. Silva, L. Martins, Euro-
pean Building Vulnerability Data Repository (Version v1.1) [Data set]. Zenodo.
http://doi.org/10.5281/zenodo.4087810, 2020.
[13]L. Martins, V. Silva, H. Crowley, F. Cavalieri, Vulnerability Modeller’s Toolkit, an
Open-Source Platform for Vulnerability Analysis. Computers & Structures, submitted.
[14]L. Luzi, R. Puglia, E. Russo, M. D'Amico, C. Felicetta, F. Pacor, G. Lanzano, U. Çeken,
J. Clinton, G. Costa, L. Duni, E. Farzanegan, P. Gueguen, C. Ionescu, I. Kalogeras, H.
Özener, D. Pesaresi, R. Sleeman, A. Strollo, M. Zare, The Engineering StrongMotion
Database: A Platform to Access PanEuropean Accelerometric Data. Seismological Re-
search Letters, 87(4): p. 987-997. DOI: 10.1785/0220150278, 2016.
[15]L. Luzi, G. Lanzano, C. Felicetta, M.C. D’Amico, E. Russo, S. Sgobba, F. Pacor,
ORFEUS Working Group 5, Engineering Strong Motion Database (ESM) (Version 2.0).
Istituto Nazionale di Geofisica e Vulcanologia (INGV) DOI:
https://doi.org/10.13127/ESM, 2020.
[16]B. Chiou, R. Darragh, N. Gregor, W. Silva, NGA project strong-motion database. Earth-
quake Spectra, 24, 23–44, 2008.
[17]J.W. Baker, C.A. Cornell, Spectral shape, epsilon and record selection. Earthquake En-
gineering & Structural Dynamics, 35(9), 1077-1095, DOI:
https://doi.org/10.1002/eqe.571, 2006.
[18]L. Eads, E. Miranda, D.G. Lignos, Average spectral acceleration as an intensity measure
for collapse risk assessment. Earthquake Engineering & Structural Dynamics, 44(12),
2057-2073, DOI: doi:10.1002/eqe.2575, 2015.
[19]B.A. Bradley, R.P. Dhakal, G.A. MacRae, M. Cubrinovski, Prediction of spatially dis-
tributed seismic demands in specific structures: Ground motion and structural response.
Earthquake Engineering and Structural Dynamics, 39, 501-520, 2010.
[20]N. Luco, A. Cornell, Structure-Specific Scalar Intensity Measures for Near-Source and
Ordinary Earthquake Ground Motions. Earthquake Spectra, 23(2), 357–392, 2007.
2687
Helen Crowley, Vitor Silva, Luis Martins, Xavier Romão and Nuno Pereira
[21]A.K. Kazantzi, D. Vamvatsikos, Intensity measure selection for vulnerability studies of
building classes. Earthquake Engineering and Structural Dynamics, 44, 2677-2694, 2015.
[22]FEMA, HAZUS-MH Technical Manual, Federal Emergency Management Agency,
Washington DC, 2004.
[23]E. Reinoso, M.A. Jaimes, L. Esteva, Estimation of life vulnerability inside buildings dur-
ing earthquakes. Structure and Infrastructure Engineering, DOI:
10.1080/15732479.2017.1401097, 2017.
[24]H. Crowley, V. Silva, P. Kalakonas, L. Martins, G. Weatherill, K. Pitilakis, E. Riga, B.
Borzi, M. Faravelli, Verification of the European Seismic Risk Model (ESRM20). Pro-
ceedings of 17th World Conference on Earthquake Engineering, Sendai, Japan, 2020d.
[25]K. Jaiswal, D. Wald, M. Hearne, Estimating casualties for large worldwide earthquakes
using an empirical approach. US Geological Survey Open-File Report 1136, 2009.
[26]K. Jaiswal, D. Wald, Estimating Economic Losses from Earthquakes Using an Empirical
Approach. Earthquake Spectra, 29(1), 309-324, 2013.
[27]EMDAT, International Disasters Database of the Centre for Research on the Epidemiol-
ogy of Disasters. Available at https://www.emdat.be/ 2019.
[28]M. Dolce, E. Speranza, F. Giordano, B. Borzi, F. Bocchi, C. Conte, A. Di Meo, M.
Faravelli, V. Pascale, Observed damage database of past Italian earthquakes: the Da.D.O
WebGIS. Bollettino di Geofisica Teorica ed Applicata, 60(2), 141-164, DOI:
10.4430/bgta0254, 2019.
2688
... This building taxonomy was simplified for the GED4All Global Exposure Database for Multi-Hazard Risk Analysis [14] by selecting a lower number of attributes for the classification. This classification system was adapted for the ESRM20 to classify the vulnerability of European buildings [4,15]. ...
Article
Full-text available
For seismic risk analysis, reliable predictions and estimations of earthquake damage and seismic behaviour of buildings are essential. A common method is the use of fragility curves. In this paper, fragility functions are developed based on various numerical damage criteria for five defined damage grades, from slight to destruction. The proposed new multiscale approach establishes a correlation between observed damage patterns due to foreign earthquakes and the seismic response of the building using thresholds for material-specific and global characteristics. This approach takes into account various possible damage patterns on different scales more comprehensively than the well-known approach of displacement criteria. Moreover, the approach is universal and adaptable for building classes as well as region-specific material and system characteristics. Several damage criteria with their defined limit values are assigned to the five proposed damage grades, whereby quantity and distribution of the exceeded criteria are relevant, since the first occurrence does not always lead to damage. With the new approach, damages that are not evident in the pushover curve in terms of strength degradation can be detected and taken into account for the damage thresholds. The derived displacement values associated with the damage levels are the basis for developing fragility functions. The results—damage criteria, pushover curves with damage grades, capacity curves as well as fragility functions and parameters—are presented for a four-storey reinforced concrete frame building. These results are discussed and validated with data from the literature. Comparisons to existing fragility functions in the literature show that our developed fragility functions are mostly located in the middle range, graphically as well as for the curve parameters. This specific example was chosen to present our multiscale approach, but for general building classes, numerous simulations with varying characteristics are essential and result in a higher standard deviation of the final fragility curves.
... We simulate thirty thousand stochastic catalogues, each with one-year of seismicity, as well as ground motions at the exposure locations for each simulated event, using the ESHM13 (Woessner et al. 2015) source model with the ESHM20 ground-motion models . The losses for each event are then calculated by combining the simulated ground motions with the exposure model, together with a set of vulnerability functions that represent the probability of loss conditional on the level of ground shaking for each building class in the exposure model (Martins and Silva 2020;Crowley et al. 2021). It is worth noting here that ground motion spatial correlation and buildingto-building damage correlation has not been considered in this study. ...
Article
Full-text available
The spatial resolution of exposure data has a substantial impact on the accuracy and reliability of seismic risk estimates. While several studies have investigated the influence of the geographical detail of urban exposure data in earthquake loss models, there is also a need to understand its implications at the regional scale. This study investigates the effects of exposure resolution on the European loss model and its influence on the resulting loss estimates by simulating dozens of exposure and site models (630 models) representing a wide range of assumptions related to the geo-resolution of the exposed asset locations and the associated site conditions. Losses are examined in terms of portfolio average annual loss (AAL) and return period losses at national and sub-national levels. The results indicate that neglecting the uncertainty related to asset locations and their associated site conditions within an exposure model can introduce significant bias to the risk results. The results also demonstrate that disaggregating exposure to a grid or weighting/relocating exposure locations and site properties using a density map of the built areas can improve the accuracy of the estimated losses.
Article
Full-text available
A uniform and comprehensive classification system, often referred to as taxonomy, is fundamental for the characterization of building portfolios for natural hazard risk assessment. A building taxonomy characterizes assets according to attributes that can influence the likelihood of damage due to the effects of natural hazards. Within the scope of the Global Earthquake Model (GEM) initiative, a building taxonomy (GEM Building Taxonomy V2.0) was developed with the goal of classifying buildings according to their seismic vulnerability. This taxonomy contained 13 building attributes, including the main material of construction, lateral load-resisting system, date of construction and number of stories. Since its release in 2012, the taxonomy has been used by hundreds of experts working on exposure and risk modeling efforts. These applications allowed the identification of several limitations , which led to the improvement and expansion of this taxonomy into a new classification system compatible with multi-hazard risk assessment. This expanded taxonomy (named GED4ALL) includes more attributes and several details relevant for buildings exposed to natural hazards beyond earthquakes. GED4ALL has been applied in several international initiatives, enabling the identification of the most common building classes in the world, and facilitating compatibility between exposure models and databases of vulnerability and damage databases.
Article
Full-text available
A uniform and comprehensive classification system, often referred to as taxonomy, is fundamental for the characterization of building portfolios for natural hazard risk assessment. A building taxonomy characterizes assets according to attributes that can influence the likelihood of damage due to the effects of natural hazards. Within the scope of the Global Earthquake Model (GEM) initiative, a building taxonomy (GEM Building Taxonomy V2.0) was developed with the goal of classifying buildings according to their seismic vulnerability. This taxonomy contained 13 building attributes, including the main material of construction, lateral load-resisting system, date of construction and number of stories. Since its release in 2012, the taxonomy has been used by hundreds of experts working on exposure and risk modeling efforts. These applications allowed the identification of several limitations , which led to the improvement and expansion of this taxonomy into a new classification system compatible with multi-hazard risk assessment. This expanded taxonomy (named GED4ALL) includes more attributes and several details relevant for buildings exposed to natural hazards beyond earthquakes. GED4ALL has been applied in several international initiatives, enabling the identification of the most common building classes in the world, and facilitating compatibility between exposure models and databases of vulnerability and damage databases.
Article
Full-text available
Vulnerability functions describe the expected loss for a given ground shaking intensity level and are an essential component in probabilistic seismic risk assessment. This manuscript presents a novel open-source platform for the derivation of analytical fragility and vulnerability models, covering state-of-the-art methodologies, and addressing critical issues in vulnerability modelling such as uncertainty propagation, validation/verification of results and sufficiency/efficiency of intensity measure types. This framework is divided into seven modules designed to guide users through the different stages of analytical vulnerability modelling from the selection of ground motion records to the validation and verification of the models. The platform was implemented in the Python programming language and it is freely accessible through a public GitHub repository. A graphical user interface is included with the toolkit and is intended to be a general-purpose method for modellers to interact with the vulnerability modellers toolkit (VMTK). Experienced users are encouraged to use Python’s scripting capabilities to explore all the features of the VMTK source code and to contribute to future releases of the toolkit.
Article
Full-text available
Seismic fragility and vulnerability assessment is an essential step in the evaluation of probabilistic seismic risk. Ideally, models developed and calibrated for the building portfolio of interest would be readily available. However, the lack of damage data and insufficient analytical studies lead to a paucity of fragility and vulnerability models, in particular in the developing world. This study describes the development of an analytical fragility and vulnerability model covering the most common building classes at the global scale. Nearly five hundred functions were developed to cover the majority of combinations of construction material, height, lateral load resisting system and seismic design level. The fragility and vulnerability were derived using nonlinear time-history analyses on equivalent single-degree-of-freedom oscillators and a large set of ground motion records representing several tectonic environments. The resulting fragility and vulnerability functions were validated through a series of tests which include the calculation of the average annual loss ratio for a number of locations, the comparison of probabilities of collapse across all building classes, and the repetition of past seismic events. The set of vulnerability functions was used for the assessment of economic losses due to earthquakes as part of the global seismic risk model supported by the Global Earthquake Model Foundation. (full-text viewer available at: https://rdcu.be/b4Kal)
Article
Full-text available
Following major seismic events occurred in Italy over the last 50 years, a notable amount of data relevant to post-earthquake damage of ordinary buildings was collected. These data today represent an inestimable scientific heritage, useful for prevision and prevention purposes, including calibration of vulnerability models for seismic risk assessment and formulation of damage scenarios. However, the data sets resulting from different inspection tools developed over the years are not immediately comparable with each other. The need for enhancing the reliability of prevention models and more effectively support strategic decision-making has moved the Italian Civil Protection Department to undertake, since 2014, a specific project with this ambitious goal. Developed with the technological support of Eucentre Foundation, the web-gis platform, named Da.D.O. (Observed Damage Database), is meant to store and to share data from large post-earthquake damage campaigns occurred in the past. Da.D.O. is addressed to Civil Protection Department users, members of the scientific community and Regions, though its access could be further extended in the future to other stakeholders. The paper describes goals, contents and capabilities of the IT platform, which, at present, includes data sets relevant to nine seismic events (or sequences) occurred from 1976 (Friuli earthquake) to 2012 (Emilia-Romagna earthquake).
Article
Full-text available
A methodology is presented for assessing the expected loss of human lives due to the structural damage, failure of contents and non-structural components inside buildings during earthquakes. This approach estimates human life losses considering whether the structure collapses (partial or totally) or not. Entrapment and entrapment fatality rates are calculated using available data from worldwide literature and newspapers of 50 collapsed buildings, which information is reasonably reliable. We apply this methodology to one example of a five-floor RC frame building of typical office occupancy. Vulnerability functions are presented relating the intensity of the earthquake to the loss of human lives inside buildings, considering different times of occurrence of the earthquakes: daytime, commuting time and nighttime. Uncertainties in all parts of the process are very large, but they are formally addressed and taken into account. This methodology may improve the risk assessment for insurances purposes, optimisation of the cycle of life, mitigation strategies, risk control and tolerable risk.
Article
Full-text available
This article describes the Engineering Strong-Motion Database (ESM), developed in the framework of the European project Network of European Research Infrastructures for Earthquake Risk Assessment and Mitigation (NERA, see Data and Resources). ESM is specifically designed to provide end users only with quality-checked, uniformly processed strong-motion data and relevant parameters and has done so since 1969 in the Euro-Mediterranean region. The database was designed for a large variety of stakeholders (expert seismologists, earthquake engineers , students, and professionals) with a user-friendly and straightforward web interface. Users can access earthquake and station information and download waveforms of events with magnitude ≥ 4:0 (unproc-essed and processed acceleration, velocity, and displacement, and acceleration and displacement response spectra at 5% damping). Specific tools are also available to users to process strong-motion data and select ground-motion suites for code-based seismic structural analyses.
Article
Full-text available
The selection of a scalar Intensity Measure (IM) for performing analytical vulnerability (loss) assessment across a building class is addressed. We investigate the ability of several IM choices to downgrade the effect of seismological parameters (sufficiency) as well as reduce the record-to-record variability (efficiency) for both highrise and lowrise sets of ‘index’ buildings. These characteristics are explored in unprecedented detail, employing comparisons and statistical significance testing at given levels of local engineering demand parameters (story drift ratios and peak floor accelerations) that relate to losses, instead of global variables such as the maximum interstory drift. Thus, a detailed limit-state-specific view is offered for the suitability of different scalar IMs for loss assessment. As expected, typical single-period spectral values are found to introduce unwanted bias at high levels of scaling, both for a single as well as a class of buildings. On the other hand, the geometric mean of the spectral acceleration values estimated at several periods between the class-average second-mode and an elongated class-average first-mode period offers a practical choice that significantly reduces the spectral-shape bias without requiring the development of new ground motion prediction equations. Given that record selection remains a site- and building-specific process, such an improved IM can help achieve reliable estimates for building portfolios, as well as single structures, at no additional cost.
Article
Building exposure and vulnerability models for seismic risk assessment have been the focus of a number of European projects in recent years, but there has never been a concerted effort among the research community to produce a uniform European risk model. The European Commission’s Horizon 2020 SERA project has a work package that is dedicated to that objective, through the development of an exposure model, an associated set of fragility/vulnerability models, and a database of socioeconomic indicators in order to calculate probabilistic integrated seismic risk at a European scale. This article provides details of the development of the first versions of the European exposure model that describe the distribution of the main residential, industrial and commercial building classes across all countries in Europe, as well as their occupants and replacement costs. The v0.1 of the European exposure model has been integrated within the Global Earthquake Model’s global exposure and risk maps. Preliminary analyses using the model show that almost 35% of the residential population in Europe is exposed to a 475-year return period peak ground acceleration (PGA) hazard of at least 0.1 g, thus highlighting the importance of European seismic risk modeling and mitigation.
Chapter
We studied the earthquake mortality rates for more than 4,500 worldwide earthquakes since 1973 and developed an empirical country- and region-specific earthquake vulnerability model to be used as a candidate for post-earthquake fatality estimation by the U.S. Geological Survey’s Prompt Assessment of Global Earthquakes for Response (PAGER) system. Earthquake fatality rate is defined as the ratio of the total number of shaking-related fatalities to the total population exposed at a given shaking intensity (in terms of Modified-Mercalli (MM) shaking intensity scale). An atlas of global Shakemaps developed for PAGER project (Allen and others, 2008) and the Landscan 2006 population database developed by Oak Ridge National Laboratory (Dobson and others, 2000; Bhaduri and others, 2002) provides global hazard and population exposure information which are necessary for the development of fatality rate. Earthquake fatality rate function is expressed in terms of a two-parameter lognormal cumulative distribution function. The objective function (norm) is defined in such a way that we minimize the residual error in hindcasting past earthquake fatalities. The earthquake fatality rate is based on past fatal earthquakes (earthquakes causing one or more deaths) in individual countries where at least four fatal earthquakes occurred during the catalog period. All earthquakes that have occurred since 1973 (fatal or non-fatal) were included in order to constrain the fatality rates for future estimations. Only a few dozen countries have experienced four or more fatal earthquakes since 1973; hence, we needed a procedure to derive regional fatality rates for countries that had not had enough fatal earthquakes during the catalog period. We propose a new global regionalization scheme based on idealization of countries that are expected to have similar susceptibility to future earthquake losses given the existing building stock, its vulnerability, and other socioeconomic characteristics. The fatality estimates obtained using an empirical country- or region-specific model will be used along with other selected engineering risk-based loss models (semi-empirical and analytical) in the U.S. Geological Survey’s Prompt Assessment of Global Earthquakes for Response (PAGER) system for generation of automated earthquake alerts. These alerts could potentially benefit the rapid earthquake response agencies and governments for better response to reduce earthquake fatalities. Fatality estimates are also useful to stimulate earthquake preparedness planning and disaster mitigation. The proposed model has several advantages as compared with other candidate methods, and the country- or region-specific fatality rates can be readily updated when new data become available.
Article
This paper investigates the performance of spectral acceleration averaged over a period range (Saavg) as an intensity measure (IM) for estimating the collapse risk of structures subjected to earthquake loading. The performance of Saavg is evaluated using the following criteria: efficiency, sufficiency, the availability or ease of developing probabilistic seismic hazard information in terms of the IM and the variability of collapse risk estimates produced by the IM. Comparisons are also made between Saavg and the more traditional IM: spectral acceleration at the first-mode period of the structure (Sa(T1)). Though most previous studies have evaluated IMs using a relatively limited set of structures, this paper considers nearly 700 moment-resisting frame and shear wall structures of various heights to compare the efficiency and sufficiency of the IMs. The collapse risk estimates produced by Saavg and Sa(T1) are also compared, and the variability of the risk estimates is evaluated when different ground motion sets are used to assess the structural response. The results of this paper suggest that Saavg, when computed using an appropriate period range, is generally more efficient, more likely to be sufficient and provides more stable collapse risk estimates than Sa(T1). Copyright © 2015 John Wiley & Sons, Ltd.