Post-earthquake Damage Data Research Articles

Machine learning and soft voting ensemble classification for earthquake induced damage to bridges

Seismic vulnerability assessment of bridges is essential for both pre-earthquake planning and post-earthquake rapid loss assessment to facilitate maintenance planning and recovery initiatives. Given the impracticality of creating numerical models for every conceivable attribute variation within bridge classes, data-driven approaches offer a versatile solution to address such variations. While data-driven techniques are widely adopted in bridge engineering, many existing models rely on numerical or experimental datasets. However, such models may not adequately capture the dynamic and complex nature of seismic response of bridges. This presents an opportunity to evaluate the effectiveness of data-driven approaches using post-earthquake damage data. This paper assesses the efficacy of machine learning models and develops a robust framework for the classification of earthquake-induced damage incurred to some bridge classes. We deployed damage data collected following the 2015 Gorkha earthquake, incorporating taxonomical, stiffness, and excitation variables as predictor variables. Considering three distinct damage classes—minor, major, and critical—in a heterogeneous bridge class environment, we evaluated multiple machine learning classifiers. To account for the influence of heterogeneous bridge classes in the dataset, we conducted separate assessments for the reinforced concrete (RCC) bridge class. The Random Forest classifier achieved a substantial accuracy improvement, reaching 83% when compared to the mixed bridge class accuracy of 79% reflected by the ANN classifier. Similarly, a minor improvement was observed when deploying soft voting classifiers for RCC only bridge data. The sum of observations suggest that tree-based classifiers consistently outperform other classifiers when considering single bridge classes. This study identifies stiffness variables as the most critical factor in bridge damage classification.

Towards a multi-parametric fragility model for Italian masonry buildings based on the informative level

This study provides a multi-parametric fragility model for Italian masonry buildings, which varies based on available information about buildings characteristics, sources from data of varying accuracy level. Three increasing levels are considered: limited, extensive, and exhaustive. These levels correspond to different degree of knowledge about the building stock, ranging from detailed in situ inspections, where a comprehensive understanding of nearly all the building features is achievable, to the case where it must necessarily rely upon a limited data (census data). This multilevel classification is established based on the number of explanatory parameters employed in defining building taxonomies. The parameters considered in this study are derived from post-earthquake damage data collected after the 2009 L'Aquila earthquake. They have undergone a thorough review and a refinement process and have been supplemented with 2001 census data to overcome the critical issue of data completeness. In addition to providing information about the geometric and typological characteristics of residential masonry buildings, the same data source also yields insights into the extent of damage sustained by these buildings. This, in turn, enables the derivation of a Damage Probability Matrix (DPM), obtained as a function on ground motion intensity values evaluated from ShakeMap of the event. The outcomes of a nonparametric statistical test, applied on DPMs, are used to drive an agglomerative clustering procedure, to establishing a hierarchical ranking of the parameters in term of their discriminatory power. The absolute (number of cases where the null testing hypothesis is rejected) and the relative (examining the test’s strength) comparison of the outcomes, allows to identify the parameters with the greatest discriminatory power. These are the one that define the branches of the taxonomies under consideration. Conversely, the less influential parameters with regard to seismic vulnerability are eligible to be grouped together, allowing the definition of a more compact taxonomy. This approach yields distinct sets of fragility curves, contingent upon the information level gained for the specific building stock being investigated, ranging from census data to comprehensive in-situ surveys. A last focus on the impact of the enhancement in the level of knowledge on seismic vulnerability is also done.

Influence of structural-geometric features on seismic vulnerability of masonry buildings based on post-earthquake damage data

In this study, seismic vulnerability of Italian masonry buildings is analyzed taking advantage of the post-earthquake damage data recently released by the Italian Department of Civil Protection through the online platform Da.D.O. (Dolce et al. 2019).The main purpose of the study is to describe the damage attitude of several building's typologies, univocally defined by means of structural and geometrical features. To this aim, a time-consuming data processing has been performed to obtain a more effective taxonomy, based on masonry quality layout, type of horizontal structure, presence/absence of retrofit interventions, construction age and number of storeys.Moreover, to avoid bias in vulnerability analysis, the most reliable datasets has been detected through a comprehensive data analysis, also integrating the original database with census data. Then, damage analysis has been done, converting the damage observed on vertical structures in 5+1 damage levels of the whole building according to the European Macroseismic Scale classification. ShakeMaps in terms of peak parameters released by National Institute of Geophysics and Volcanology are also considered to characterize ground motion shaking at each building's location. Lastly, a lognormal statistical model has been used to derive vulnerability curves for considered typologies, fully investigating the influence of each building's feature on damage attitude.

Empirically based approaches for the derivation of fragility curves of Italian RC building typologies

In this study, fragility curves for Italian RC building typologies are derived, by taking advantage of post-earthquake damage data gathered after the Irpinia (1980) and L'Aquila (2009) seismic events. The severity of the ground shaking is characterized by extrapolating peak ground acceleration values from updated INGV ShakeMaps and a global level of damage, consistently defined with the EMS-98 and accounting for both structural and non-structural damage observed on preselected building components, is associated with each inspected building. RC buildings are allocated to twenty-four building typologies, accounting for construction age, design level and number of storeys. Statistical processing of observational damage data showed the general tendency of seismic vulnerability to increase with the number of storeys and to reduce with design level and construction age. In some cases, the reliability of the obtained results was however affected by the limited amount of damage data and/or by scarcely populated building typologies. Two alternative empirically based approaches, relying on different data processing techniques and fitting strategies, are hence developed, overcoming issues in empirical fragility assessment driven by data paucity and allowing for a robust seismic fragility model.

Empirical seismic vulnerability of Italian URM churches hit by the 2016-17 Central Italy earthquake sequence

This study provides some insights on the empirical seismic vulnerability of unreinforced masonry (URM) churches, by statistically processing post-earthquake damage data gathered in the aftermath of the Central Italy (2016-2017) damage and usability assessment campaign. The available damage database, collecting 3356 post-earthquake survey forms, is first critically examined. The peak ground acceleration (PGA), estimated from updated INGV ShakeMaps and suitably accounting for the several main shocks of the earthquake sequence, is adopted for locally characterizing the ground motion severity. A global level of damage is associated with each inspected church, starting from damage information on individual collapse mechanisms available from the post-earthquake survey form. Empirical damage probability matrices (DPMs) are derived by statistically processing the entire dataset of masonry churches, then refined based on further attributes (i.e. plan area and construction age) consistent with exposure data. Parametric representation of the resulting observational damage frequencies is also provided by taking advantage of the binomial distribution, which allows for fully describing the entire damage repartition in the different levels by a single parameter.

Empirical fragility curves for Italian residential RC buildings

In this paper, empirical fragility curves for reinforced concrete buildings are derived, based on post-earthquake damage data collected in the aftermath of earthquakes occurred in Italy in the period 1976–2012. These data, made available through an online platform called Da.D.O., provide information on building position, building characteristics and damage detected on different structural components. A critical review of this huge amount of data is carried out to guarantee the consistency among all the considered databases. Then, an in-depth analysis of the degree of completeness of the survey campaign is made, aiming at the identification of the Municipalities subjected to a partial survey campaign, which are discarded from fragility analysis. At the end of this stage, only the Irpinia 1980 and L’Aquila 2009 databases are considered for further elaborations, as fully complying with these criteria. The resulting database is then integrated with non-inspected buildings sited in less affected areas (assumed undamaged), to account for the negative evidence of damage. The PGA evaluated from the shakemaps of the Italian National Institute of Geophysics and Volcanology (INGV) and a metric based on six damage levels according to EMS-98 are used for fragility analysis. The damage levels are obtained from observed damage collected during post-earthquake inspections through existing conversion rules, considering damage to vertical structures and infills/partitions. The maximum damage level observed on vertical structures and infills/partitions is then associated to the whole building. Fragility curves for two vulnerability classes, C2 and D, further subdivided into three classes of building height, are obtained from those derived for specific structural typologies (identified based on building height and type of design), using their frequency of occurrence at national level as weights.

Empirical seismic fragility models for Nepalese school buildings

Empirical vulnerability models are fundamental tools to assess the impact of future earthquakes on urban settlements and communities. Generally, they consist of sets of fragility curves that are derived from georeferenced post-earthquake damage data. Following the 2015 Nepal earthquake sequence, the World Bank, through the Global Program for Safer Schools, conducted a Structural Integrity and Damage Assessment (SIDA) of about 18,000 school buildings in the earthquake-affected area. In this work, the database is utilized to identify the main structural characteristics of the Nepalese school building stock. For the first time, extended SIDA school damage data is processed to derive fragility curves for the main structural typologies. Data sets for each structural typology are used for a Bayesian updating of existing fragilities to obtain regional models for Nepalese schools. These fragility estimates can be adopted to assess potential seismic losses of the school infrastructure in Nepal. Additionally, they can be used for calibrating loss assessment studies in the wider Himalayan region where the structural typologies are similar.

ANALYSIS OF DAMAGE DATA COLLECTED FOR WINE STORAGE TANKS FOLLOWING THE 2013 AND 2016 NEW ZEALAND EARTHQUAKES

The 2013 Seddon earthquake (Mw 6.5), the 2013 Lake Grassmere earthquake (Mw 6.6), and the 2016 Kaikōura earthquake (Mw 7.8) provided an opportunity to assemble the most extensive damage database to wine storage tanks ever compiled worldwide. An overview of this damage database is presented herein based on the in-field post-earthquake damage data collected for 2058 wine storage tanks (1512 legged tanks and 546 flat-based tanks) following the 2013 earthquakes and 1401 wine storage tanks (599 legged tanks and 802 flat-based tanks) following the 2016 earthquake. Critique of the earthquake damage database revealed that in 2013, 39% and 47% of the flat-based wine tanks sustained damage to their base shells and anchors respectively, while due to resilience measures implemented following the 2013 earthquakes, in the 2016 earthquake the damage to tank base shells and tank anchors of flat-based wine tanks was reduced to 32% and 23% respectively and instead damage to tank barrels (54%) and tank cones (43%) was identified as the two most frequently occurring damage modes for this type of tank. Analysis of damage data for legged wine tanks revealed that the frame-legs of legged wine tanks sustained the greatest damage percentage among different parts of legged tanks in both the 2013 earthquakes (40%) and in the 2016 earthquake (44%). Analysis of damage data and socio-economic findings highlight the need for industry-wide standards, which may have socio-economic implications for wineries.

Empirical fragility curves for Italian URM buildings

This paper illustrates the derivation of an empirical fragility model for residential unreinforced masonry (URM) buildings, calibrated on Italian post-earthquake damage data and compatible with the key features of the Italian national seismic risk platform. Seismic vulnerability is described by fragility functions for three vulnerability classes, then refined based on the building height. To this aim, a clustering strategy is implemented to merge predefined building typologies into vulnerability classes, based on the similarity of the observed seismic fragility. On the other side, a specific procedure is built up to determine the vulnerability composition of the exposed URM building stock, starting from national census data. The empirically-derived model was implemented into the national seismic risk platform and used, together with other vulnerability models, for assessing seismic risk in Italy. The results presented in this paper, consisting of refined typological fragility curves and fragility curves for vulnerability classes, can be also exploited for estimating both expected seismic damage and risk in sites with similar seismic hazard and building inventory.

Use of post-earthquake damage data to calibrate, validate and compare two seismic vulnerability assessment methods for vernacular architecture

The paper presents and discusses the application of two large scale seismic vulnerability assessment methods on the island of Faial in Azores (Portugal). The two methods are specifically conceived to assess the seismic vulnerability of vernacular architecture. The first method follows a classical seismic vulnerability index approach and is referred as SVIVA (Seismic Vulnerability Index for Vernacular Architecture). The second method is referred as SAVVAS (Seismic Assessment of the Vulnerability of Vernacular Architecture Structures) and it is a numerical tool intended to estimate the seismic capacity of vernacular buildings in terms of seismic load factors associated with different structural damage limit states. The main reason behind the selection of Faial Island as a case study was the availability of post-earthquake reports of the building stock after the 1998 Azores earthquake, which allowed comparing the damage scenarios obtained using both methods with the post-earthquake damage data and thus helped for the calibration and validation of the two methods. The application of both methods led to a good fit between estimated versus observed damage grades, which validated their applicability as large-scale first level approaches. Moreover, as the main outcome, the paper presents the novelties of the SAVVAS method, which had not been applied before, and discusses its main advantages, namely: no need for calibration with previous post-earthquake damage data, an enhancement of the prediction capabilities, a more individualized evaluation of the buildings and the possibility to assess the seismic performance of the building in different loading directions.

Preliminary prediction of damage to residential buildings following the 21st August 2017 Ischia earthquake

On August 21st, 2017, an earthquake with duration magnitude Md = 4.0 and epicentre in Casamicciola Terme hit Ischia island, in the South of Italy. This event caused two fatalities and dozens of injured people. Moreover, despite the low magnitude, the earthquake produced significant damages to masonry and reinforced concrete (RC) buildings, with some partial or complete collapse of structures, in a very limited area close to the epicentre, while even at small distance from the most damaged zone the earthquake was just felt by local people and tourists. In the days after the event, discussions concerning the destructive effects of such an earthquake arose in the scientific community—as also reported by local and national media. In this paper, the seismic history of Ischia island is recalled to show and explain the peculiarity of the August 21st earthquake, which is also described in terms of ground motion and response spectra characteristics. The results of the first surveys carried out in Casamicciola Terme are reported, together with appropriate pictures, to introduce and explain the observed damage state of masonry and RC buildings in the epicentral zone. Then, data from the 15th general census of the population and dwellings (ISTAT) is used to define vulnerability classes according to the classification of the European Macrosismic Scale (EMS-98) (Grunthal, 1998). Seismic damage scenarios are then evaluated combining macro-seismic intensity values obtained using an interpolation method starting from QUEST macro-seismic survey data (Azzaro et al., 2017 ) and fragility curves for A-to-D vulnerability classes and for five damage states, from DS0 (no damage) to DS5 (collapse) trough a Monte Carlo simulation technique. The distributions of Usable, Temporarily or Partially Unusable, and Unusable buildings, which are obtained by using relationships between damage and usability judgments obtained through post-earthquake damage data collected after past seismic events, result in very good accordance with those published in September 1st, 2017 by the Department of Civil Protection, regarding a dataset of about 600 buildings.

A class-oriented mechanical approach for seismic damage assessment of RC buildings subjected to the 2009 L’Aquila earthquake

The simplified mechanical method POST (PushOver on Shear Type models) for seismic vulnerability assessment of RC buildings is used in this study to derive damage scenarios for a database of 7597 RC buildings subjected to the 2009 L’Aquila earthquake. POST allows the evaluation of fragility curves through the determination of the non-linear static response of RC buildings in closed form, assuming the hypothesis of shear type behaviour, and considering the influence of infill panels both in the derivation of structural response and in assessment of building damage, which is defined according to the European Macroseismic Scale EMS-98. The aim of the present study is to provide a much more significant and reliable validation of the methodology, thanks to a much wider database compared to previous studies, and based on a different application of the methodology, i.e. at building class-level instead of single building-level. To this aim, the main geometrical-typological characteristics of the analysed buildings (number of storeys, age of construction, building area) have been statistically characterized based on data collected from post-earthquake AeDES survey forms, considering both the variability of each single parameter and the correlation that exists between one parameter and the other. This also allows to analyse the effectiveness of the adopted analytical procedure in predicting the general trends of observed damage with these parameters, showing a good agreement between observed and predicted trends. The overall predicted damage scenarios are compared with the corresponding observed ones, collected from AeDES survey forms, highlighting, again, a good agreement. Finally, the assumed mechanical interpretation of damage classification of EMS-98 is validated through the comparison between the distributions of damage to vertical structures and infill panels and the corresponding observed post-earthquake damage data.

Damage classification and derivation of damage probability matrices from L’Aquila (2009) post-earthquake survey data

Post-earthquake damage data represent an invaluable source of information for the seismic vulnerability assessment of the exposed building stock, as they are a direct evidence of the actual buildings’ performance under real seismic events. This paper exploits a robust and homogeneous database of damage data collected after the 2009 L’Aquila (Italy) earthquake, to derive damage probability matrices for several building typologies representative of the Italian building stock. To this aim, the first part of the work investigates several issues related to the definition of damage to be associated with each inspected building. Different approaches and damage conversion rules are applied, pointing out advantages and weaknesses of each one. Considering the widespread seismic damage observed on masonry infill panels and partitions of reinforced concrete constructions, the impact of this type of non-structural damage on empirical damage and functional loss distributions is explored. The second part of the study proposes different possible interpretations of the repartition of the observed damage in the different damage levels, showing in some cases a bimodal trend. Two novel hybrid procedures are outlined and compared with the classical binomial approach for predicting the subdivision of damage in the different levels. The application of the proposed methodologies to the different building typologies allows the selection, for each one, of the method providing the best fit to empirical results. The parameters required for the application of the optimal approach are reported in the paper, so that results can be used for forecasting the expected seismic damage in sites with similar seismic hazard and exposed buildings.

Seismic vulnerability assessment of stone masonry façade walls: Calibration using fragility-based results and observed damage

This paper is focused on the calibration of a simplified seismic vulnerability index method for masonry façade walls, through an innovative two-step calibration process based on two complementary approaches. The first one is based on a set of fragility curves constructed from out-of-plane damage limit states obtained from experimental data, which are used to calibrate the weights associated to the parameters of the vulnerability index method that rule the out-of-plane response of the masonry façade walls. The second approach, subsequently used to calibrate the weights of the remaining vulnerability parameters, is based on post-earthquake damage data collected after the seismic event that struck the Azores Archipelago in July 1998. The results obtained from such calibration are then presented and critically discussed taking into account not only the calibration itself, but also their reliability from the methodological point of view. Finally, the addition of three new evaluation parameters is further proposed and analysed. This two-step calibration process represents a valuable contribution for the current state-of-art of the simplified seismic vulnerability assessment methodologies which, up to the present, have been developed and calibrated only on the basis of empirical data.

Seismic Risk Assessment of Nonengineered Residential Buildings: State of the Practice

AbstractThe poor socioeconomic conditions in developing countries often lead to poorly constructed residential buildings that are particularly vulnerable to damage during an earthquake. A review of available literature carried out as part of a larger research program highlights the scarcity of existing fragility curves for the wide typology of nonengineered residential buildings around the world. Furthermore, fragility curves derived using empirical data are almost nonexistent due to the lack of postearthquake damage data and insufficient ground motion recordings in developing countries. The diversity in construction techniques and material quality in developing countries, particularly for nonengineered residential buildings, cannot be sufficiently represented through simplified or idealized analytical models. Therefore, the use of empirical based fragility curves is considered to be a well-suited approach for assessing the seismic risk levels for nonengineered residential buildings in developing countrie...

Planning helicopter logistics in disaster relief

This paper describes an efficient planning system for coordinating helicopter operations in disaster relief. This system can be used as a simulation tool in contingency panning for better disaster preparedness and helps to generate plans with estimated data. The proposed system consists of a mathematical model and a Route Management Procedure (RMP) that post-processes the outputs of the model. The system is concerned with helicopter operations that involve last mile distribution and pickups for post-disaster medical care and injured evacuation. Delivering items such as medicine, vaccines, blood, i.v., etc. to affected locations, and evacuating injured persons from these locations comprise the transportation tasks to be performed by helicopters. The proposed modeling system accommodates the special aviation constraints of helicopters and it can handle large scale helicopter missions. The goal of the system is to minimize the total mission time required to complete the transportation task. The RMP enables the Decision Maker (DM) to specify either the mission completion time or the number of vehicles available for the mission. Respecting the limitations imposed by the DM, the RMP generates fuel and capacity feasible helicopter itineraries that complete within the specified mission completion time. A scenario that is based on the post-earthquake damage data provided by the Disaster Coordination Center of Istanbul is used for testing the method.

Post-earthquake loss assessment based on structural component damage inspection for residential RC buildings

There is substantial evidence showing that the seismic performance of many existing buildings may be inadequate to resist another strong earthquake. The losses from a devastating earthquake are always huge. In order to prevent damage extension and to restore the damaged community as quickly as possible, immediate post-earthquake damage assessment is always conducted through site inspection on structural components within a restricted short period of time to screen buildings that are damaged to certain levels or in danger of collapse. Without detail financial loss estimation, engineers have to face the challenge to decide whether a badly damaged building is worth retrofitting for sustainability, or needs to be demolished because existing loss estimation models are not based on the post-earthquake damage rating system. Based on some post-earthquake damage data of RC residential buildings, this paper aims to link inspected component damage level, building damage state and direct financial loss in terms of repair to replacement cost ratio. Damage of structural components are quantified by a set of damage factors and finally integrated as a building damage indicator. Building repair to replacement cost ratio and storey repair to replacement cost ratio corresponding to various damage levels of RC residential buildings have been estimated. With these statistical data, relationships of building damage indicator to repair to replacement cost ratio has been built from regression analysis.