Towards an integrated vulnerability-based approach for evaluating, managing and mitigating earthquake risk in urban areas

Universidade do Minho Escola de Engenharia Rafael Ramírez Eudave Towards an integrated vulnerability-based approach for evaluating, managing and mitigating earthquake risk in urban areas. maio de 2023 UMinho | 2023 Rafael Ramírez Eudave Towards an integrated vulnerability-based approach for evaluating, managing and mitigating earthquake risk in urban areas. Universidade do Minho Escola de Engenharia Rafael Ramirez Eudave Towards an integrated vulnerability-based approach for evaluating, managing and mitigating earthquake risk in urban areas. Doctoral Thesis Civil Engineering Work conducted under supervision of Tiago Miguel dos Santos Ferreira Paulo José Brandão Barbosa Lourenço may 2023 Copyright and Terms of Use for Third Party Work. This dissertation reports on academic work that can be used by third parties as long as the internationally accepted standards and good practices are respected concerning copyright and related rights. This work can thereafter be used under the terms established in the license below. Readers needing authorization conditions not provided for in the indicated licensing should contact the author through the RepositóriUM of the University of Minho. License granted to users of this work: CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ i Acknowledgements. I want to express my sincere gratitude to Dr Tiago Miguel Ferreira for his strong commitment during this research. His generosity while giving advice and sharing his experiences has been a very relevant support throughout this work. I would also like to thank Dr Paulo B. Lourenço for co-orienting this work and providing valuable support and guidance. Special thanks to Héctor Adrián Ramírez Eudave for his significant help during the field works reported in Chapter 5. The datasets used in Chapter 5 were partly obtained from the Master Thesis of Aurelia Flores Rodríguez, Francisco Ruíz Herrera, Pedro Martínez Bret, and Eduardo Tototzintle Huitle. Dr Carlos Montero Pantoja facilitated the contact during the COVID-19 lockdown. The feedback and involvement of Dr Romeu Vicente enriched the works reported in chapters 5, 6 and 7. Works reported in Chapter 7 were supported by the Institute of Engineering of the National Autonomous University of Mexico (II-UNAM) between January and August 2022, under the supervision of Dr Fernando Peña Mondragón. Dr Marcos Mauricio Chávez Cano and M. C. Víctor Hugo Torres Romero supported the laboratory experiments, while the field campaigns had the valuable help of M. I. Cyprien Lubin, Ing. Alexis Brito Sánchez and Tech. Arturo Fuerte. These works were financed by the II-UNAM through project R562. Special thanks to Prof. Natalia García Gómez (Autonomous University of the State of Morelos) for her support during the first stages of these campaigns. This work was partly financed by FCT/MCTES through national funds (PIDDAC) under the R&D Unit Institute for Sustainability and Innovation in Structural Engineering (ISISE), reference UIDB/04029/2020. This research had financial support provided by the Portuguese Foundation of Science and Technology (FCT) through the Analysis and Mitigation of Risks in Infrastructures (InfraRisk) program under the PhD grant PD/BD/150385/2019. ii Dedicatoria. María, Fermín, Verónica, Héctor e Isabel: hogar y puerto; Bech (†), Ricardo, Elisa, Areli y Marlène: puente y refugio; Isis (y familia), Ósca, Carlos, Matea y quienes hicieron de Portugal una casa. Gracias. Ihcuac tlahtolli ye miqui [...] Quinihcuac motzacua nohuian altepepan in tlanexillotl, in quixohuayan. In ye tlamahuizolo occetica in teoyotl, in tlacayotl, in machi mani muan yoli in tlalticpac. [...] Ihcuac tlahtolli ye miqui, occequintin ye omiqueh ihuan miec huel miquizqueh. Tezcatl maniz puztecqui, netzatzililiztli icehuallo cemihcac necahualoh: totlacayo motolinia. Miguel León Portilla. Cuando muere una lengua [...] Entonces se cierra a todos los pueblos del mundo una ventana, una puerta. Un asomarse de modo distinto a las cosas divinas y humanas, a cuanto es ser y vida en la tierra. [...] Cuando muere una lengua, ya muchas han muerto y muchas pueden morir. Espejos para siempre quebrados, sombra de voces para siempre acalladas: la humanidad se empobrece. Miguel León Portilla. iii Statement of Integrity. I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. University of Minho, Guimarães, may 2023 Rafael Ramirez Eudave iv Resumo. Sismos de grande intensidade, como aqueles que ocorreram na Turquía-Síria (2023) ou México (2017) deviam chamar a atenção para o projeto e implementação de ações proativas que conduzam à identificação de bens vulneráveis. A presente tese propõe um fluxo de trabalho relativamente simples para efetuar avaliações da vulnerabilidade sísmica à escala urbana mediante ferramentas digitais. Um modelo de vulnerabilidade baseado em parâmetros é adotado devido à afinidade que possui com o Catálogo Nacional de Monumentos Históricos mexicano. Uma primeira implementação do método (a grande escala) foi efetuada na cidade histórica de Atlixco (Puebla, México), demonstrando a sua aplicabilidade e algumas limitações, o que permitiu o desenvolvimento de uma estratégia para quantificar e considerar as incertezas epistémicas encontradas nos processos de aquisição de dados. Devido ao volume de dados tratado, foi preciso desenvolver meios robustos para obter, armazenar e gerir informações. O uso de Sistemas de Informação Geográfica, com programas à medida baseados em linguagem Python e a distribuição de ficheiros na ”nuvem”, facilitou a criação de bases de dados de escala urbana para facilitar a aquisição de dados em campo, os cálculos de vulnerabilidade e dano e, finalmente, a representação dos resultados. Este desenvolvimento foi a base para um segundo conjunto de trabalhos em municípios do estado de Morelos (México). A caracterização da vulnerabilidade sísmica de mais de 160 construções permitiu a avaliação da representatividade do método paramétrico pela comparação entre os níveis de dano teórico e os danos observados depois do terramoto de Puebla-Morelos (2017). Esta comparação foi a base para efetuar processos de calibração e ajuste assistidos por algoritmos de aprendizagem de máquina (Machine Learning), fornecendo bases para o desenvolvimento de modelos de vulnerabilidade à medida (mediante o uso de Inteligência Artificial), apoiados nas evidências de eventos sísmicos prévios. Palavras-chave: Aprendizagem de Máquina, Avaliação da vulnerabilidade sísmica; Avaliações de escala urbana; Avaliação do risco sísmico; Estruturas de alvenaria; Sistemas de Informação Geográfica. v Abstract. Strong seismic events like the ones of Türkiye-Syria (2023) or Mexico (2017) should guide our attention to the design and implementation of proactive actions aimed to identify vulnerable assets. This work is aimed to propose a suitable and easy-to-implement workflow for performing large-scale seismic vulnerability assessments in historic environments by means of digital tools. A vulnerability-oriented model based on parameters is adopted given its affinity with the Mexican Catalogue of Historical Monuments. A first large-scale implementation of this method in the historical city of Atlixco (Puebla, Mexico) demonstrated its suitability and some limitations, which lead to develop a strategy for quantifying and involving the epistemic uncertainties found during the data acquisition process. Given the volume of data that these analyses involve, it was necessary to develop robust data acquisition, storing and management strategies. The use of Geographical Information System environments together with customised Python-based programs and cloud-based distribution permitted to assemble urban databases for facilitating field data acquisition, performing vulnerability and damage calculations, and representing outcomes. This development was the base for performing a second large-scale assessment in selected municipalities of the state of Morelos (Mexico). The characterisation of the seismic vulnerability of more than 160 buildings permitted to assess the representativeness of the parametric vulnerability approach by comparing the theoretical damage estimations against the damages observed after the Puebla-Morelos 2017 Earthquakes. Such comparison is the base for performing a Machine Learning assisted process of calibration and adjustment, representing a feasible strategy for calibrating these vulnerability models by using Machine-Learning algorithms and the empirical evidence of damage in post-seismic scenarios. Keywords: Machine Learning; Seismic Vulnerability Assessment; Urban-Scale Assessment; Seismic Risk Assessment; Masonry Structures; Geographical Information Systems. vi 5.7 Summary of results (3/5). The interval between µ− Dand µ+ Dis represented as the white box. .........................................124 5.8 Summary of results (4/5). The interval between µ− Dand µ+ Dis represented as the white box. .........................................124 5.9 Summary of results (5/5). The interval between µ− Dand µ+ Dis represented as the white box. .........................................125 5.10 Variability of the Mean damage grades for the 83 constructions by using different values of ductility Q(The buildings were arranged from the lowest up to the highest mean damagegradevalue)..................................126 5.11 Distribution of the sample according to the calculated mean damage grade against the uncertaintyindex....................................128 5.12 Example of a graphical representation for the building SECII-27. . . . . . . . . . . . . 130 6.1 Screenshot of QGIS interface. The active layer includes a series of features associated to the Parametric Seismic Vulnerability Index for a specific set of constructions. . . . . . 140 6.2 Integrated data flow on GIS databases for VIM surveys. . . . . . . . . . . . . . . . . . 140 6.3 The main window of the seismic vulnerability calculator and the display modules (A–D). . 142 6.4 Example of the Input interface and its utilisation during on-site data acquisition. . . . . . 145 6.5 Damage curve based on the values obtained from the seismic vulnerability calculator (top) and visual representation of the levels of vulnerability Vand mean damage grade µDby using the GIS environment (bottom). . . . . . . . . . . . . . . . . . . . . . . 147 7.1 USGS Shake maps for the Pijijiapan (left) and Puebla-Morelos (right) Earthquakes[Geodetic Facility for the Advancement of Geoscience, 2023]. . . . . . . . . . . . . . . . . . . 153 7.2 Screenshot of the python-based front-end. . . . . . . . . . . . . . . . . . . . . . . . 160 7.3 EMS-98 five damage grade scale [Centre Europèen de Géodynamique et de Séismologie, 1998] and examples of buildings with corresponding damage grades from 1 to 5 in the studiedmunicipalities. ................................163 7.4 Examples of urban (left) and rural (right) houses. . . . . . . . . . . . . . . . . . . . 164 7.5 Typologically representative constructions from the municipality of Tepoztlán. . . . . . . 164 7.6 Typical adobe configuration in a house in the municipality of Tlayacapan. . . . . . . . . 165 7.7 Comparison between the intervals for the permissible range of observed damage (bars) and analytical damage (black crosses) by sample. . . . . . . . . . . . . . . . . . . . 166 xiii 7.8 Analytical and real damage grade intervals for the municipality of Jojutla (1 of 2). . . . . 169 7.9 Analytical and real damage grade intervals for the municipality of Jojutla (1 of 2). . . . . 169 7.10 Analytical and real damage grade intervals for the municipality of Tlayacapan. . . . . . 170 7.11 Analytical and real damage grade intervals for the municipality of Yautepec (1 of 2). . . . 170 7.12 Analytical and real damage grade intervals for the municipality of Yautepec (2 of 2). . . . 171 7.13 Analytical and real damage grade intervals for the municipality of Tepoztlán (1 of 2). . . 171 7.14 Flow diagram of the training and prediction processes. . . . . . . . . . . . . . . . . . 174 7.15 Schematisation of the bagging process. . . . . . . . . . . . . . . . . . . . . . . . . 175 7.16 Schematisation of the evaluation of the model. . . . . . . . . . . . . . . . . . . . . . 177 7.17 Confusion matrix. True values correspond to observed levels of damage. . . . . . . . . 178 7.18 Vulnerability class distribution per parameter. . . . . . . . . . . . . . . . . . . . . . 178 7.19 Relative importance by feature according to the RFC model. . . . . . . . . . . . . . . 179 B.1 Examples of L,aand bfor multiple plan configurations. . . . . . . . . . . . . . . . . 223 xiv List of Tables 2.1 City structural breakdown proposed by Bambara et al. [2015]. . . . . . . . . . . . . . 14 2.2 Level of Valuation and their main features. . . . . . . . . . . . . . . . . . . . . . . . 15 2.3 Comparison between features of Aerial, close range and UAV photogrammetry [Hund etal.,2019]. ..................................... 21 2.4 Qualitative comparison between photogrammetric and laser-scan surveys, adapted from Gonizzi Barsanti et al. [2013]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.5 Examples of baseline data for several hazards [Department of Regional Development and Environment Executive Secretariat for Economic and Social Affairs Organization of AmericanStates,1991]................................. 25 2.6 Matrix for different 3D modelling techniques and representativeness, adapted from Wate etal.[2013]. ..................................... 30 3.1 Key indicators discussed in the 2007 convention of St. Petersburg as referred by Azpeitia Santanderetal.[2018]................................. 45 3.2 SWOT analysis for the historical city of Queretaro (México). Adapted from Guzman [2020]. 45 3.3 Sustainable urban revitalisation indicators. Adapted from Vehbi and Hoşkara [2009]. . . 48 3.4 Data obtained during the survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.1 Summary of inventorying systems by country. Adapted from H. Sykes [1984]. . . . . . 71 4.2 Vulnerability Index parameters, according to the calibration of Ferreira et al. [2017c]. . . 82 4.3 Parameters and vulnerability classes, according to Aguado et al. [2018]. . . . . . . . . 84 4.4 Correlation between discrete damage grades and ranges of mean damage grade. Adapted fromAguadoetal.[2018]. .............................. 90 4.5 Set of selected buildings, their corresponding Catalogue number and pre and post-event picture......................................... 92 4.6 Summary of attributes available in the public INAH datasheet. . . . . . . . . . . . . . 93 xv 4.7 Summary of the vulnerability results. . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.8 Summary of the vulnerability results. . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.9 Summary of the vulnerability results. . . . . . . . . . . . . . . . . . . . . . . . . . 97 5.1 Number of historical monuments and housing units that are considered historical monuments in the states of Oaxaca, Chiapas and Puebla, according to INAH [2019]. . . . . 103 5.2 Concepts as defined in the ISO 31000 and contextualised for assessing the risk of an earthquake.......................................105 5.3 Key concepts according to the Basic Guide for the Elaboration of Statal and Municipal ThreatsandRiskAtlas. ................................106 5.4 Components for the qualitative assessment of the vulnerability of housing units towards seismicactions.....................................110 5.5 Vulnerability Index parameters, according to the calibration of Ferreira et al. [2017c]. . . 113 5.6 Parameters and vulnerability classes, according to Aguado et al. Aguado et al. [2018]. . 114 5.7 Fields considered in the datasheets of Martínez Bret [2018], Ruíz Herrera [2018], Tototzintle Huitle [2018] and Flores Rodríguez [2018]. . . . . . . . . . . . . . . . . . . . . 118 5.8 Correlation between discrete damage grades and ranges of mean damage grade. . . . . 119 5.9 Example of conservative grades obtained after combining different grades and QC levels. 121 5.10 Example of conservative grades obtained after combining different grades and QC levels. 129 6.1 Parameters for the VIM approach based on Ramírez Eudave and Ferreira [2021d]. . . . 138 7.1 Affected housing units per State as result of the Earthquakes of the 7th and 19th September, 2017 [Senado de la Republica, 2017]. . . . . . . . . . . . . . . . . . . . . . . . 154 7.2 Vulnerability Index parameters, according to the calibration of Ferreira et al. [2017c]. . . 158 7.3 Correspondence between discrete damage grades and rages of mean damage grade. . . 159 7.4 General composition of the QGIS layer attribute table. . . . . . . . . . . . . . . . . . 160 7.5 General description of the housing environment for the studied municipalities according to the Census of 2014 [INEGI - Instituto Nacional de Estadística y Geografía, 2023]. . . . 161 7.6 Number of constructions considered per municipality. . . . . . . . . . . . . . . . . . 161 7.7 Number of structures assessed by municipality and level of damage. . . . . . . . . . . 165 7.8 Summary of successful predictions for the level of physical damage. . . . . . . . . . . 167 7.9 Summary of samples in which the observed and estimated damages are overlapped. . . 168 7.10 Hyperparameter setup for training the RFC algorithm. . . . . . . . . . . . . . . . . . 176 xvi 7.11 Classificationreport. .................................177 A.1 FP1. Geometry of façade. Based on the length (B) and height (H) ratio. . . . . . . . . . 210 A.2 FP2. Maximum slenderness based on the height (H) and thickness (s) of the wall. . . . 210 A.3 FP3. Area of openings. Ratio between the area of openings with respect to the entire façade.........................................210 A.4 FP4. Misalignment of openings. Vertical and horizontal alignments of the openings. . . . 210 A.5 FP5. Interaction between façades. To take into account «pounding effects» that may occur due to adjacent façades with different heights. . . . . . . . . . . . . . . . . . . 211 A.6 FP6. Quality of materials. It evaluates the type of masonry. . . . . . . . . . . . . . . 211 A.7 FP7. State of conservation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 A.8 FP8. Replacement of the original floor system. Percentage P of the original horizontal diaphragms that were replaced by heavy concrete-based structures. . . . . . . . . . . 211 A.9 FP9. Connection to orthogonal walls. . . . . . . . . . . . . . . . . . . . . . . . . . 212 A.10 FP10. Connection to horizontal diaphragms. Evaluates the efficiency of the interaction between the façade and the diaphragms, regarding the percentage eof diaphragms with anefficientconnection.................................212 A.11 FP11. Impulsive nature of the roofing system. Equivalent to the [P12] parameter of the Vulnerability Index Methodology, «Roofing system». . . . . . . . . . . . . . . . . . . 212 A.12 FP12. Elements connected to the façade. . . . . . . . . . . . . . . . . . . . . . . . 213 A.13 FP13 Improving elements. Elements that might help the structural performance of façades, mitigating out-of-plane mechanisms. . . . . . . . . . . . . . . . . . . . . . 213 B.1 Classes for the parameter BP1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 B.2 Classes for the parameter BP2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 B.3 Classes for the parameter BP3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 B.4 Correspondence among the uses stated in the datasheets for Atlixco against the uses and live loads found in the Mexican Code and Eurocode 1. The value for pmwas selected as the most conservative one between both codes. . . . . . . . . . . . . . . . . . . . 219 B.5 Dead load associated to the materials. . . . . . . . . . . . . . . . . . . . . . . . . . 220 B.6 Dead load associated to the materials. . . . . . . . . . . . . . . . . . . . . . . . . . 220 B.7 Classes for the parameter BP4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 B.8 Classes for the parameter BP5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 xvii B.9 Classes for the parameter BP6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 B.10 Classes for the parameter BP7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 B.11 Classes for the parameter BP8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 B.12 Classes for the parameter BP9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 B.13 Classes for the parameter BP10. . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 B.14 Classes for the parameter BP11. . . . . . . . . . . . . . . . . . . . . . . . . . . . 225 B.15 Classes for the parameter BP12. . . . . . . . . . . . . . . . . . . . . . . . . . . . 226 B.16 Classes for the parameter BP13. . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 B.17 Classes for the parameter BP14. . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 B.18 Classes for the parameter FP1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 B.19 Classes for the parameter FP2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 B.20 Classes for the parameter FP3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 B.21 Classes for the parameter FP4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 B.22 Classes for the parameter FP5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 B.23 Classes for the parameter FP8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 B.24 Classes for the parameter FP9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 B.25 Classes for the parameter FP10. . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 B.26 Classes for the parameter FP12. . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 B.27 Classes for the parameter FP13. . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 xviii Chapter 1 Introduction. «Hic locus est ubi mors gaudet succurrere vitae». This Latin phrase can be freely translated as ”This is the place where death delights in helping life” and is still found at the entrance of the Anatomical Theatre of Padua (Italy) and some other similar facilities. The sense of this phrase is not only a reminder of the inevitability of death, but also recalls the due respect regarding the knowledge obtained from the mortal rests of human beings. There should be a strong ethical commitment when learning from losses, having a clear duty of giving better opportunities for the present and future generations. In this sense, is convenient to always keep in mind that our current knowledge related to seismic risk mitigation is based on processes of forensic engineering that have been built from material and human losses. This implies an ethical positioning towards the role of architecture and engineering in the context of human settlements. Therefore, the knowledge and observations obtained from past destructive events must be respectfully framed in the context of the human dimension of disasters. *** 1 2 CHAPTER 1. INTRODUCTION. Seismic events represent a very important challenge towards the protection and conservation of historical cultural assets. Some recent seismic events, such as the one that occurred in Türkiye-Syria (2023) and Mexico (2017), exposed the lack of preparedness and emergency planning that was reflected in the loss of human lives and large sets of constructions, including numerous historical constructions. The typical consideration of seismic risk as the convergent product of hazard (the seismic event, in terms of intensity), exposure (the elements prone to be damaged or lost as a consequence of the event) and vulnerability (the capacity that a certain element has to cope towards the event) reflects the critical relevance that proactive vulnerability assessments may have for mitigating damages and losses in post-event scenarios. To undertake structural vulnerability assessments is not easy given a series of circumstances associated to the scale and specificity of historical cities. As an example, a relevant part of historical constructions in Mexico are catalogued and identified, but there is a generalised lack of the knowledge related to the structural safety of these buildings. Among the difficulties for performing structural assessments, it is convenient to remark the heterogeneity and variability of the structures, and the territorial-scale distribution of the samples. Therefore, to perform straightforward and generalised assessments is still object of discussion and development. Despite the fact that some relevant advances have been reached for assessing punctual monumental-scale constructions, relatively small (often vernacular housing) assets are commonly out of the scope of these approaches. This situation was identified as a relevant gap in literature, closely related to a series of current situations that compromise the sustainability of historical urban environments and their inhabitants. The early and proactive identification of vulnerable elements throughout the historical constructions is a key step for mitigating adverse scenarios, primarily preventing human losses and also the occurrence of irreversible damages to cultural heritage. This objective is part of the sustainability and resilience of communities, as stated in the UNESCO Historic Urban Landscape approach. Even if the discussion related to the conservation and protection of historical buildings is commonly focused on the cultural value of this assets, it becomes relevant to put attention on their role for satisfying human needs (namely housing, but also basic services) commonly adapted for local environmental conditions. Historical vernacular housing constructions typically reflect a construction tradition that encompasses a valuable source of popular knowledge for climatic passive technologies, structural solutions and habitat strategies. The loss of these material documents is also a loss of genius loci , that is often followed by a replacement process that implies overwriting the collective memory and identity. 1.1. OBJECTIVES. 3 1.1 Objectives. This thesis addresses the insufficient knowledge in large-scale seismic vulnerability assessment methods for historical cities by presenting and demonstrating the suitability, robustness and potential of involving parameter-based descriptive models, digital data acquisition and management tools, geographical vulnerability-oriented databases and machine-learning based calibrations. Even if the practical examples have been implemented in Mexican cities, the experiences herein demonstrated can be extrapolated to other Historic Urban Landscape environments. A set of four blocks of specific objectives was outlined in order to structure the order and sequence of the topics herein explored. 1. Descriptive models. Configuration of a theoretical framework for holistic, accumulative, coherent and comprehensive semantic models for describing the Historic Urban Landscape. Design of suitable data models for urban-scale information, exploration of the state-of-art for urban-scale data acquisition, including the semantic codification (selection of descriptors) of vulnerability and risk. 2. Data acquisition. Proposal of robust frameworks for obtaining vulnerability-related indicators from multiple sources (e.g., catalogues, in-situ surveys, remote sensing tools, etc.). Configuration of a vulnerability-oriented survey based on parametric methods. Design of large-scale implementation strategies by the means of digital tools and distributed databases. 3. Data framework and management. Implementation of a Geographical Information System database with a series of vulnerability surveying-oriented extended capabilities. Design of an adhoc software for managing the vulnerability indicators in order to perform analyses and calculations. Distribution and publishing of the databases for supporting large-scale in-situ surveying and remote works. 4. Seismic vulnerability assessment. Performance of large-scale implementations by the means of the tools developed for supporting data acquisition and management. Identification of the limitations of the parametric vulnerability index method and proposal of calibration by the means of Machine Learning algorithms. Selected Mexican case studies in the States of Puebla and Morelos will be the base for performing seismic vulnerability assessments in the context of the 2017 Puebla-Morelos and Chiapas Earthquakes. 4 CHAPTER 1. INTRODUCTION. Figure 1.1 illustrates the relations existing among the thesis’ objectives, the main subtopics and the content of each chapter. Each chapter is preceded by this same scheme in which the chapter’s topics are shown in black, while the topics previously treated are shown in grey and those that are still not discussed are presented in grey with dashed lines. Figure 1.1: Schematic final representation of all the topics herein discussed. 1.2 Thesis outline and organisation. This thesis has been organised as a sequence of chapters that have been conceived, consolidated and published as peer-reviewed journal articles, in accordance with the provisionat the School of Engineering of the University of Minho. The exceptions are Chapter 1 (Introduction), since it is aimed to present a general view of the motivations and objectives of this research work, and Chapter 8 (Final remarks and future works) that highlights the main findings and potential further explorations based on this work. The content of the remaining chapters is herein integrally reproduced from the published version with minor editorial changes. Each chapter was conceived as an autonomous unit and, in consequence, there are texts, tables and references that appear more than once in this thesis. For the sake of consistency with the original conception, it was decided to present them integrally. 2.1. INTRODUCTION. 11 this approach for generating models of existing structures and buildings, leading to the existence of the so-called Historical BIM or HBIM models, that can reflect specific singularities that are often found in historical constructions. Contrary to what happens for the new ones, the process of modelling existing entities does not occur in a ‘regular and well-behaved environment’. Historic buildings require sensitive approaches able to consider the active ageing of constructions, which are necessarily the result of multiple transformation, reparation, use and decay processes. These processes bring with them a series of physical prints that are meaningful for further interventions and diagnosis, such as deformations, cracks, strains or superficial decays. For this reason, when acquiring data, it is of paramount importance to adopt a comprehensive approach capable of incorporate the simplifications that are usually considered as acceptable when working with new constructions. When addressed to existing constructions, some data acquisition technologies (such as satellite images or photogrammetric point clouds) permit to translate the physical reality in representative models. Nevertheless, there are some gaps in the interoperability of GIS, BIM and HBIM approaches, mostly because of the different primary purpose of the platforms. BIM, HBIM and GIS models can separately contain valuable information for analysing vulnerability. Finding ways in which the codes and models of GIS, BIM and HBIM approaches are consistent and interoperable may be the base of a comprehensive strategy for urban vulnerability assessment. The existing challenges for this interoperability are, generically, four. The first challenge is to decide and establish what to code. Since the objects are real physical entities, they have virtually infinite attributes to be measured and characterised. The decision of what to code depends on which are the attributes considered as relevant for vulnerability analysis purposes. Secondly, it is necessary to decide appropriate data acquisition strategies. These strategies must be consistent with the information considered necessary, so the obtained data is reliable. The next challenge is related to the strategies for codifying the obtained information in models that can manage and reflect all the considered attributes namely concerning the interaction between different supports and software in which information has been coded and managed. Finally, it is essential to decide how to simulate the phenomena in order to assess the vulnerability of the represented object. The present work will analyse existing literature in order to explore how the challenges mentioned above may be solved. It will show some recent discussions about the attributes that are considered relevant in a historical centre, i.e., which information would be essential to code. Then, it will show some current strategies that permit to acquire information from large-scale entities, such as photogrammetry and laser 12 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). scan surveys. Once the data acquisition is complete, it becomes necessary to code it into adequate models, so a discussion about the potential use and limitations of GIS, BIM and HBIM models will be held. Finally, the suitability of a comprehensive and integrative framework will be discussed. The hypothesis is that an adequate selection of semantic descriptors supports the acquisition of the semantic descriptions in order to generate representative models for assessing the vulnerability of the represented objects (see Fig. 2.1). Figure 2.1: Summary of the main topics to be discussed. 2.2 Codification of physical entities. The first important decision is what to code. The delimitation of the information that will be coded is based on both the phenomena and the entity that will be modelled. Then, when developing models for assessing the vulnerability of historical centres, it becomes fundamental to understand which are the attributes that the concept of «historical centre» implies. Moreover, an adequate delimitation of the phenomena and its components is needed. There is a relevant interdependency between the phenomena and the modelled object, in the sense that the success of the simulation depends on the consistency of the logic operations with the coded data. As a departing point, a semantic description is a base for further coding. Descriptors are, then, the generic features that will be measured in order to characterise a specific entity. 2.2. CODIFICATION OF PHYSICAL ENTITIES. 13 2.2.1 Codification of the Historic Urban Landscape. There is no unique concept about what «urban» is. Despite this, there is a general agreement in considering that cities are shaped by a spatial context (with attributes such as size, density, distances or territories) as well as a social context, which comprises economic, social, political and historical processes [Boone and Fragkias, 2013]. The spirit of our present idea of heritage is strongly related to the relationship between intangible values and products concerning tangible entities in which these values are materialised [Bouchenaki, 2003]. Then, the concepts of «city» and «heritage» are often related. There is an increasing interest in cities through a holistic approach. One of the critical steps in this exploration is the concept of «cultural landscape», understood as the combined works of nature and of man [World Heritage Centre, 2010]. A relevant idea of the cultural nature of an anthropic landscape is found in ancient conceptions of the «genius loci», which refers to the uniqueness and authenticity of a city [Petzet, 2009]. One of the most recent efforts for providing international institutional frameworks for historical centres management and study is the development of the Historic Urban Landscape approach, supported by UNESCO. The HUL Guidebook [UNESCO, 2016], proposes that a Historic Urban Landscape is « . . . the result of the historical layering of cultural and natural values and attributes, extending beyond the notion of ‘historical centre’ or ‘ensemble’ to include the broader urban context and its geographical setting. This wider context includes notably the site’s topography, geomorphology, hydrology and natural features, its built environment, both historic and contemporary, its infrastructures above and below ground, its open spaces and gardens, its land-use patterns and spatial organisation, perceptions and visual relationships, as well as all other elements of the urban structure. It also includes social and cultural practices and values, economic processes and the tangible dimensions of heritage as related to diversity and identity ». In order to characterise the city as a system, it becomes necessary to establish a functional description of it. Bambara et al. [2015] propose a city structural breakdown (Table 2.1) in which the urban complex is approached from four general subsystems, each one of them divided into two inferior level categories. Semantic descriptions of the elements that are found in a historical centre (mostly as a network of buildings and transitive spaces, namely streets) should include not only geometric and material parameters, but also valuable information about users (including their potential vulnerabilities) and occupation patterns. There are precedents, such as Giuliani [2020], who proposed a spatial analysis based on a Space syntax representation. This approach considers two spatial properties, metric and topological, which permit to model physical space (namely urban space) as a network of flows that connect land uses. Then, this approach may enrich a set of semantic descriptors for streets and transitive spaces to be mathematically analysed. 14 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). Grade 1 category Grade 2 category Grade 3 category 1. Technical networks 1.1 Water 1.2 Power 1.3 Telecommunications 1.4 Transport 1.1.1 Drinking water supply 1.1.2 Sewage and stormwater 1.2.1 Electric network 1.2.2 Natural gas network 1.2.3 District heating system 1.3.1 Broadband 1.3.2 Landline telephone services 1.3.3 Mobile telephone services 1.4.1 Road 1.4.2 Railway 2. Housing 2.1 Mobile 2.2 Individual 2.3 Group 3. Business 3.1 Services 3.2 Manufacturing Industry 3.3 Stores 4. Public infrastructure 4.1 Decision-making crisis management infrastructure 4.2 Operational crisis management infrastructure 4.3 Infrastructure dedicated to vulnerable inhabitants 4.4 Other public services 4.1.1. Urban system administration 4.2.1 Police, emergency services 4.3.1 Health facilities (hospitals) 4.3.2 Educative facilities. 4.3.3 Prisons, retirement houses 4.4.1 Welfare benefits 4.4.2 Housing allowances 4.4.3 Business support 4.4.4 Record keeping 4.4.5 Legal and judicial facilities Table 2.1: City structural breakdown proposed by Bambara et al. [2015]. Even if we generically accept that buildings are one of the multiple agents that conform the urban entity, it is useful to study them separately, since that single constructions may experiment more changes in time than the city itself and because there is an important variability in terms of solutions, typologies and systems that demand to be approached according to their singularity. Then, the building objects are, for historical centre semantic description purposes, considered as a subset of the historical centre entity. 2.2.2 Codification of single constructions for structural analysis purposes. Structural assessment for historic structures is a challenging task, involving potentially significant uncertainties related to the mechanical properties of materials and the effect of different types of decay [Calvi et al., 2006]. Along with other aspects, these constructions are typically made of masonry, which brings multiple complexities to the process due to material heterogeneity, mechanical and geometrical non-linearities, weather/environmental related decay, among others. Nevertheless, there are several attempts for general 2.2. CODIFICATION OF PHYSICAL ENTITIES. 15 frameworks in which a set of assumptions is accepted, and it is possible, assuming some, to carry a simplified structural assessment. As an example, the Italian Directive [Consiglio dei Ministri della Reppublica Italiana, 2014] considers three different approaches for simplified seismic assessment by regarding the relation between the simplified assumptions that are taken against the detail and accuracy of the information available for a specific building. Each approach is considered as a Level of Valuation [Rondolini, 2018] and is associated with a certain kind of analysis (Table 2.2). Level of Valuation Description LV1 Qualitative analysis and assessment of seismic safety with simplified mechanical models, which allows estimating a global vulnerability index. LV2 Simple macroelement assessment (local mechanisms of collapse), based on the local response of macro-elements. LV3 Complex analysis of the seismic response, namely through Finite Elements Analysis. Table 2.2: Level of Valuation and their main features. The application of the ‘Level of Valuation’ approach permits to have a general framework for analysing several buildings with similar typologies and construction technologies. This is especially useful when analysing historical centres, such as the case of Matera [D’Amato et al., 2020]. In the case of nonreinforced masonry, there are numerous works addressed to the characterisation of the most common failure mechanisms in order to use them as a starting point for assessing the tendency that a specific building can have to the activation of those mechanisms. This approach is mostly based on geometric parameters having thus critical limitations when dealing with heterogeneous and discontinuous (resulting from different construction stages) masonries. The general idea behind this approach is to have a limited set of potential collapse mechanisms and a corresponding mathematical formulation in which a load factor may be found as a function of a set of geometrical parameters. It is relevant to remark that this approach has been developed specifically with the scope of non-reinforced masonry buildings. Some authors, such as D’Ayala and Speranza [2003] for example, propose abacus containing the most common failure mechanisms. A relevant work aimed at systematising a tool for applying the mechanism of collapse criteria can be found in the MEDEA (Manual for Earthquake Damage Evaluation and safety Assessment) project [Zuccaro et al., 2010], in which also reinforced concrete mechanisms of collapse are considered. When considering specific seismic characteristics of a determined region (more specifically, a historical centre), it becomes possible to have loss estimations based on past experiences [D’Ayala et al., 1997]. 16 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). Both approaches, Level of Valuation and Collapse Mechanisms, are basically supported by geometrical information with very general assumptions regarding the mechanical behaviour of materials. Nevertheless, there exist also simplified approaches for analysing or categorising determined structural systems, such as masonry works. A remarkable approach is the Masonry Quality Index [Borri and De Maria, 2015], which permits to develop a semi-quantitative assessment of masonry constructions by taking into account attributes that may be visually identified. This approach permits to obtain indicative values for some mechanical properties of masonry walls based of simple-view features, such as the kind of plaster, the size and shape of the masonry units or the regularity and path of joints. The definition of the attributes to be coded depends on the vulnerability assessment approach to be used. The above-mentioned ones – Level of Valuation, Collapse Mechanisms and Masonry Quality Index – permit simple assessments based on geometric and visual identification aspects. For such, it is necessary to establish a comprehensive list of parameters that can be used to describe the geometry of the main structural elements of a building (resisting walls, flooring and covering systems), as well as their state of conservation. Since the primary structural system of historical buildings is usually composed of masonry walls, most of the attributes should focus on the description of these elements. Some secondary attributes (typically obtainable through simple operations), such as slenderness or the relative area of the openings can be calculated directly in the databases. It should be noticed that there some recent exercises focused on the automatic generation of these core elements directly from the raw point cloud data, allowing the creation of 3D models more easily and quickly [Thomson and Boehm, 2015]. From exposed, it is plausible to assume the possibility of performing simplified structural assessments taking simple geometric parameters as a base. A general geometric description may permit to characterise a determined construction in terms of an abacus of common failure modes. Furthermore, the existence of detailed images (e.g., from a photogrammetric survey) may even help to characterise a determined masonry work and some mechanical aspects. 2.2.3 Codification of vulnerability and risk. The abundant concepts and conditions that are often related to the idea of risk may difficult to define the true meaning of this concept. A valuable description is that of the international standard ISO 31000:2009 [ISO - International Organization for Standardization, 2009]. In this standard, the risk is stated as the « effect of uncertainty on objectives ». Furthermore, uncertainty is « . . . the state, even partial, of deficiency of information related to, understanding or knowledge of an event, its consequence or likelihood ». Objectives may consider a full set of aspects that are addressed to specific systems or components. In terms 2.2. CODIFICATION OF PHYSICAL ENTITIES. 17 of historical centres, those objectives may be related to the urban functions. In terms of constructions, targets may be the performance states. Risk is, then, the effect of uncertainty of achieving objectives due to potential events (or hazards) with a certain likelihood, regarding their consequences in a specific system. Literature devoted to the multi-risk assessment of cultural assets offers numerous methodologies in order to describe hazards and vulnerabilities quantitatively. Most of these works are focused in specific contexts, frequently using a very local interpretation of risks. Since these exercises are generally developed for specifics typologies, there is a lack of universal or generic evaluations intended to classify historical constructions. Several experiences use semi-quantitative rankings in which physical properties are related to numerical scales between the most favourable and unfavourable extremes. Another approach is to rate by the mean of gradients that are informatically computed [Eastman, 2005]. When several features are taken into account, it is possible to provide more complex vulnerability scenarios. Some important cartographies that have been used in the bibliography as input information for vulnerability and risk assessment methodologies are: • Seismic macro zonation and Peak Ground Accelerations maps [Augusti and Ciampoli, 2000]; • Elevation, slope and orographic traits [Agapiou et al., 2016]; • Urban mapping, land use and urban expansion patterns [Hadjimitsis et al., 2013]; • Demographic statistics [Tate et al., 2010]; • Sea erosion, salinity and coast exposure [Agapiou et al., 2016, Hadjimitsis et al., 2013]; • Soil erosion and landslide events [Agapiou et al., 2016, Skilodimou et al., 2019]; • Fires historical occurrence [Agapiou et al., 2016]; • Air pollution [Hadjimitsis et al., 2013]; • Proximity to roads and land communica,tions [Hadjimitsis et al., 2013]; • Geological maps [Skilodimou et al., 2019]; • Precipitation data and flood events [Skilodimou et al., 2019]; • Cadastral maps and market prices mapping [van Westen et al., 2002]. The analysis of several hazard sources also aimed to develop multi-criteria approaches, in which natural phenomena are accounted and punctuated to establish prevention hierarchies and priorities. Some events are, however, conditioned for more than one layer of information. Tsunamis are a paradigmatic example of this. Also, the cartographies of specialised facilities and accessibility infrastructure are relevant for planning actions in case of emergency [Alzouby et al., 2019], aiming to update proper strategies for postdisaster stages. The diverse outcomes of risk assessment can complement and feedback existing data or new cartographies. 18 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). In the context of seismic vulnerability for historic buildings, Augusti and Ciampoli [2000] propose the following definitions: «...Vulnerability (of a building or other constructed facility): Sensitivity to actions in terms of damage. In rigour, given by the ensemble of the conditional probabilities of attaining of exceeding each level of damage (up to complete collapse), in function of the intensity of the relevant action; in practice, often measured by one or few parameters. Exposure: Existence of facilities subject to a hazard. Hazard: Probability of occurrence of an action (taking account of its intensity). Risk: Probability of damage/collapse (given by the convolution of hazard, exposure and vulnerability)» There is a broad spectrum of hazards to which cities are exposed to, even if they are unnoticed. Thanks to its generic nature, the framework presented in this chapter can be used to address most of these hazards, namely fire, floods or earthquake. However, due to their regional nature and high significant human, economic and social impact, special attention has been paid herein to the seismic vulnerability. According to the United Nations Office for Disaster Risk Reduction [2017], although earthquakes represented only 7.8% of disastrous events between 1998 and 2017, they are responsible for 56% of the deaths and about 23% of the economic losses resulting from natural disasters. 2.3 Data acquisition techniques. Once the descriptors are established, it becomes necessary to design and execute proper data acquisition techniques in order to characterise those descriptors with reliability. The most recent and promising techniques for urban and architectonic surveying are based on the acquisition of cloud points to be treated and manipulated as three-dimensional spatial descriptions. The usual acquisition of these points is made through processes such as photogrammetry and laser scan, sometimes assisted with external devices for covering broad areas. The representation of three-dimensional objects by means of point clouds is based on assuming that the physical surfaces can be represented through a sufficiently dense group of points that hypothetically tends to infinite. Then, a finite number of points is saved with information regarding relative or absolute spatial coordinates, which permits the spatial reconstruction of the object when assembling the sets of points. Those point clouds can be the base of further developments, such as generating surfaces and volumes through adequate processing. The most usual approaches for obtaining point clouds from physical objects are photogrammetry and laser scan [Lerma et al., 2013]. Two strategies top the list of the most used ones for urban analyses: the aerial vehicle assisted survey (which is an auxiliary tool for photogrammetric procedures) and the satellite images photogrammetry 2.3. DATA ACQUISITION TECHNIQUES. 19 [Manfreda et al., 2018]. A third powerful tool is the laser scan, which is still the most accurate surveying tool. However, it is yet unsuitability (or hardly applicable) for the acquisition of large-scale objects or extensive areas. Some research groups have applied technologies like Virtual Reality to anticipate more accurate and faster survey campaigns [Lee et al., 2018], showing that a preliminary urban 3D model can be the starting point for a more profitable on-site survey. 2.3.1 Acquisition of point clouds from photogrammetry and laser scan. Photogrammetry processes are based on the comparison of multiple photographic images, by using the camera as a «passive sensor». Even if the photographic image does not intrinsically contain information related with physical dimensions, the processing and comparison of a dense group of overlapped pictures permit to reconstruct the relative position of the points, given the optical features of the data acquisition device. This approach is relatively affordable but may need specialised personnel for post-processing the data. However, it is a very efficient strategy due to its relatively simple implementation. This technique is limited to the quantity and quality of the images for adequate processing. In contrast, laser scan or LiDAR (Laser Imaging Detection and Ranging) can take active measures of surfaces, so each point is associated with its relative distance from the emission source, allowing to obtain very accurate and dense clouds. This capacity overcomes most of the limitations that photogrammetry has (namely its dependency on light conditions and the quality of the images). However, the relatively high cost of the devices can still limit the use of this technique. The most recent laser scan devices are even capable of obtaining colour information from the surveyed surfaces. The suitability of a strategy must regard constraints such as the on-site limitations and the availability of time and technical resources [Ahmed et al., 2011]. Despite the existence of other techniques and tools for obtaining point clouds, photogrammetry and LiDAR are still the most convenient means for data acquisition at the urban context, namely because of their suitability and flexibility in terms of scale and resolution. However, all strategies implicitly have to deal with challenges related to accessibility, cost and urban constraints. For instance, local regulations may limit the use of unmanned aerial vehicles under determined conditions. The experiment carried out by Russo et al. [2019] provides interesting perspectives for the use of low-cost and small aerial devices with relatively simple and cheap cameras for obtaining complete urban façades. This experience permitted to discuss and solve some of the technical constraints associated with this technique, contributing thus to the development of more accessible and economical workflows for urban-scale data acquisition. 20 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). 2.3.2 Drone and satellite-assisted surveys. The acquisition of point clouds with photogrammetry or laser scans can be enhanced if supported by another sort of devices such as Unmanned Aerial Vehicles (UAV) and satellite. Since one of the most critical challenges is related to the areas that must be covered during a survey, these tools permit to sweep large areas in relatively short periods of time. The selection of an appropriate tool might take into account the scale of the surveyed entity as well as the accuracy requirements. In several exercises, the scale factor is determinant when deciding the strategies for surveying a determined area [Roca et al., 2014]. In the last few years, there has been increasing use of Unmanned Aerial Vehicles, namely drones, to assist photogrammetry data acquisition. UAV offer several advantages when compared with traditional ground-level approaches, such as an excellent cost/benefit ratio. There is a rising range of vehicles and cameras that may be combined, as well as different processes for data acquisition and processing [Siebert and Teizer, 2014]. The quality of UAV-assisted photogrammetric surveys is mostly dependent on weather conditions and the limitations of the data acquisition system (cameras or LIDAR systems). Meteorological conditions might compromise UAV use in terms of stability when following specific flight paths. The design of an adequate workflow for preparation, recording, post-processing and data generation is needed (Figure 2.2). However, the use of aerial surveys may be combined with satellite and on-site control measures in order to enhance the quality and reliability of the acquired data. Figure 2.2: General workflow for a UAV-photogrammetric survey, based on Siebert and Teizer [2014]. 2.4. STRATEGIES FOR DIGITAL MODELLING. 27 One of the most attractive advantages of BIM process is the continuous development of an ultimate model from which it is possible to extract several real-time updated products, namely visual information. This way, every change is real-time updated in the related outcomes, which represents a significant saving of time. Besides, models can be fed with new information, meaning that the time used in their development is always an investment. The work of several specialists in the same model is possible because of the capacity of semantic enrichment, in which a determinate entity (namely components in a model) can store different layers of information meant for different operators. This also makes possible the interoperability of the same model with different platforms. Most of this semantical enrichment may be found in collective libraries that are continuously updated [Quattrini et al., 2015]. Semantic work also permits to set conceptual models in which all the relations between the single elements are explained and established. BIM methodology is useful for modelling existing entities as well, such as historic buildings, permitting to document and analyse their current situation and to support future interventions. The adaptation of BIM methodology for representing the numerous singularities of the historic buildings led to the development of the so-called Historic Building Information Methodology (or HBIM), a term proposed by Murphy et al. [2009]. HBIM brings a series of additions to the traditional BIM approach aimed at overcoming some of the limitations inherent to the nature of the historic buildings. In general, an HBIM model is generated and organised following the same rules as a traditional BIM model. The main differences lie in the use of ad-hoc repositories enriched with historic architectonic and structural elements (due to their singular character against standardised and/or industrially produced ones) and the addition of data regarding deformations, decays and other sorts of damages, usually non-existent in new constructions. These decays and damages may be represented as parametric instances, which are meaningful and quantifiable in project processes, and programmed as adaptive components or additional properties of existing elements. These components are also allowed to retain metadata that can be useful in different processes during the intervention stage [Lo Turco et al., 2016]. Since HBIM models are expected to reflect the existing ‘as-built’ construction, conditions or pathologies (such as geometrical deformations and irregularities) should not be simplified or neglected. When modelling an existing building, it is necessary to consider that several layers of information have a cumulative nature (since historic buildings are the result of multiple changes along its timeline). The model becomes a kind of «inverse engineering» product (Figure 2.3), with information directly acquired from the building (Figure 2.4). An accurate model can be used as a container for previous, present and future information, permitting to evaluate changes, damage processes, carry structural tests and even planning interventions with all BIM modelling advantages [Biagini et al., 2016]. 28 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). Figure 2.3: Similarities and specificities of BIM models developed to a new building or an existing one, based on Volk et al. [2014]. Figure 2.4: Typical BIM model workflows for a new building and an existing one, adapted from Volk et al. [2014]. HBIM models represent buildings by using libraries of architectural elements mostly based on several academic books on architectural history. Most of these elements have been developed in Geometric Descriptive Language (GDL), which permits to create complex geometries based on simple geometric primitives. However, numerous entities that are usually present in historic buildings (as vaults, specific kind of walls, arches, etc.) are not available in standardised libraries [Maietti et al., 2018]. Even if the uniqueness of every single building is still challenging, most appropriate steps can be taken in this field for 2.4. STRATEGIES FOR DIGITAL MODELLING. 29 example, the adoption of standard terminologies for labelling and describing elements. Another feasible strategy is to use existing standard elements that may be later linked to details like pathologies, stratigraphic units or specific attributes that correspond to the uniqueness of cultural assets (for instance, by using the Industrial Foundation Classes or IFC format). This approach may be especially useful when working on open source platforms [Diara and Rinaudo, 2018]. The possibility of an automatic model generation directly from the surveyed information may imply a significant development in terms of wide-scale entities. Even if there are no consolidated tools for this purpose, several experiences have obtained promising results. Rutzinger et al. [2009] tested the identification of wall patterns in urban scale by using a semi-automatic recognition of surfaces. This approach permitted to model entire urban façades and to classify some exterior features, namely windows and doors, adequately. Thomson and Boehm [2015] tested a semiautomatic process to directly transform information from a point cloud to a BIM model in small examples successfully. However, these recognition algorithms are not yet implemented in commercial software, and the accuracy of academic experimentation is still unsuitable (too low) for the production of HBIM models [Dore and Murphy, 2017]. Automatic identification of walls and façades can be especially useful when it is possible to make a clear typological identification. In that circumstance, it is possible to extrapolate spatial and construction specificities through the information obtained in an external geometric survey, allowing to estimate some post-disaster scenarios related, for example, with the amount of debris after a collapse or to prevent fire propagation patterns. 2.4.3 Experiences and challenges on GIS-BIM-HBIM integration. At present, it is possible to use two-dimensional cartographies (namely GIS databases) in order to manage and store multiple layers of information and attributes. Many of the vulnerability-oriented attributes can be numerically stored in a simple 2D GIS database. However, even accepting that the data acquisition can be stored and managed under the basis of a 2D model, there is a great potential in the use of 3D cartographies to explore vulnerability-related phenomena determined by spatial conditions – for example, the study of air and water flow, the behaviour of volcanic ashes or solar incidences on façades. Although the use of 3D models for the extensive study of environmental vulnerabilities is still recent, it is a field in rapid expansion and with great potential for further development. An excellent example of the use of 3D cartography of the city of Sora can be found in Pelliccio et al. [2017]. Furthermore, the pursuit of developing 3D cartographies based on point clouds instead of the usual two-dimensional maps becomes useful for generating comprehensive documentation for the current state of an urban system. That three-dimensional 30 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). frame would permit to follow changes along the time and would even allow analysing disastrous events based on a well-documented precedent state. Even if at their origin GIS and BIM approaches were conceptually different, there is an increasing interest in their interoperability. This would permit a careful control of complex projects, even when regarding small components of it, as well as its further assessment when regulations change. One of the most important differences that must be taken into account when working between BIM and GIS platforms is the level of detail that each system can represent and handle. While BIM and HBIM models are intended to represent a building as accurate as possible, GIS copes with larger scales and, consequently, with lower levels of accuracy. There exist, however, a normalised index where BIM, HBIM and GIS (among others) tools may be framed to share information [Matrone et al., 2019]: the Level of Detail (LoD). Level of Detail (LoD) is the name of a discrete scale in which five different levels of representation are described by regarding the balance between reliability and simplification that may be assumed in a specific model [Biljecki et al., 2013] (Table 2.6). However, data acquisition is also a constraint that determines which LoD can be reached from the primary source. In general terms, the descriptors for these Level of details are those presented in [Biljecki et al., 2013]. LoD and Features Aerial Acquisition Terrestrial Acquisition LoD Abstract Feature Representation Distinct Features Identified Satellite Photogrammetry Spaceborne / Airborne Close-Range Photogrammetry Mobile Laser Scan 0Digital Terrain Model (DTM) Buildings footprints Yes Yes - - 1Digital Surface Model (DSM) Buildings as prismatic Yes Yes - - 2 Building Structures with prototypic features Prototypic roof shapes - Yes Yes Yes 3 Buildings with architectural features Openings and textured wall surfaces - - Yes Yes 4 Buildings with interior architectural features Rooms with interior furniture and texturing - - - Yes Table 2.6: Matrix for different 3D modelling techniques and representativeness, adapted from Wate et al. [2013]. 2.4. STRATEGIES FOR DIGITAL MODELLING. 31 Thanks to LoD definitions, it is possible to insert simplified BIM entities in GIS models. Nevertheless, those simplified BIM models can be linked with more complex representation (as external references), through a controlled decrease in their level of detail. Such a decrease can be done by following specific LoD definitions, as the CityGML ones [Deng et al., 2016]. Another feasible strategy for interoperability is the sharing of geometrical information between IFC and CityGML standards. An IFC solid object can be mapped to a set of surfaces that jointly permit to describe the solid semantically. However, this sharing may be problematical concerning the orientation of the surfaces. To extract the geometrical parameters of an IFC entity, it becomes necessary to filter the IFC classes in order to extract only mappable surfaces [Donkers et al., 2016]. At present, experiences such as the one of Cecchini [2019] demonstrate the suitability of downgrading a BIM model to simplify some geometric parameters of IFC instances, allowing CityGML standards-based entities reconstituting volumes from sets of surfaces. There are some examples in the literature – see for example Deng et al. [2016] –, where CityGML was chosen as a key schema in an instance-based method for generating mapping rules between IFC and CityGML. In this way, it is possible to generate a so-called Semantic City Model, which captures relevant information from BIM and GIS models, achieving automatic data mapping in different Levels of Detail. An in-depth exploration of the issues about the integration of BIM and GIS carried by Ohori et al. [2017] describes explicitly geometric and topological processing problems (mostly because of self-intersection inconsistencies) and incorrect georeferencing of IFC models, concluding that most of the errors are caused by source errors when developing BIM models. Since the conversion between data models usually leads to data loss, some authors have already proposed strategies to guarantee information coexistence by establishing references between the IFC and CityGML standards, see for example Cheng and Deng [2015]. It is pertinent to recall that the LoD is addressed to the granulometry of the information for representation purposes. Hence, it is possible to have an LoD 0 model with multiple data fields describing, for instance, three-dimensional attributes of the construction, even if they are not graphically represented. This means that such models (even low LoD models) can store the attributes required to feed simplified vulnerability assessment methodologies. In the other hand, a detailed model would be the base for more complex approaches. For example, a very refined LoD 4 model would precede a corresponding Finite Element Model, or an LoD 3 model would provide data for carrying a Masonry Quality Index [Borri et al., 2015] analysis. Hence, the limitations and needs of using a certain Level of Detail as a standard for vulnerability assessment purposes depend directly on the characteristics of the assessment approach to be used. When a BIM model already exists, it is possible to extract or filter input data to simplified models, preserving intact the original information. 32 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). The transition from 3D models to CityGML requires converting both geometric and semantic information. Despite the existence of some successful experiences [Dore and Murphy, 2012, Kokla et al., 2019], there is still a significant challenge associated with data sharing between BIM and GIS platforms. Some works, like the one of López et al. [2018], propose complete workflows for covering a wide variety of scales. Recent experiences have successfully demonstrated that HBIM-GIS integration is possible and useful – see, for example, the case of the ‘Gran Torre di Oristano’ [Vacca et al., 2018]. There are interesting works in the literature intended to automatically label urban elements with semantic concepts [Cordts et al., 2016] by using ground-level photographs. Furthermore, there is also recent literature addressing the possibilities of automatic and semi-automatic identification of objects directly from satellite images [Zhu et al., 2017]. A complete three-dimensional model can offer valuable information about materials and geometric aspects of the structural elements. This may constitute a first approach to estimating tensional states in the whole structure, permitting to develop simplified structural analysis [Pelliccio et al., 2017], as well as roughly estimate the energy needed to activate diverse mechanisms of collapse, e.g., partial or global façade overturning. Once a simple model is obtained (i.e., a general urban model), the single entities can be enhanced with further surveys and detailing. This might permit to perform a first-contact study of the elements that have been identified as the critical ones. Such elements might be further assessed through BIM modelling and even reaching the design of accurate finite element models. However, this enhanced model may still be linked to the original urban one. This capacity permits to have a phased project in which some parts are more detailed than others. Since HBIM models are a particular case of BIM, the generalities for their integration in a GIS environment would be applicable for both. However, it is important to keep in mind that HBIM models can represent physical decays that might be related to multiple environmental phenomena, such as water flows, wind, solar incidences or presence of vegetation, for example. Hence, the generation of proper links between environmental information in GIS databases and specific decays in HBIM models suggest a promising opportunity for the assessment and diagnosis of vulnerabilities, finding valuable correlations. Some researcher tried to deduct the decay process of a building by comparing an as-built and an idealised model. Nieto-Julián et al. [2020], for example, propose to build a purely theoretical model based on documents and historical research; and a second one that reflects the real current state of the building through geometric data obtained with laser-scan or photogrammetry techniques. This permitted to differentiate some aspects that may be part of an original conception or a process of decay. 2.5. VULNERABILITY ASSESSMENT METHODS. 33 Although the GIS-BIM-HBIM integration is relatively recent, some experiences reported in the literature reveal promising uses and potential opportunities in the field of management and protection of built heritage. Campanaro et al. [2016] discuss the use of a 3D-GIS model of a building in Pompeii to collect information about decay processes in the building’s façades – such as cracks or lack of verticality –, which is then used for mapping an overall risk indicator. Even though this experience did not include the integration with BIM environment, the processes generated by using the ArcMap and ArcScene modulus can be developed in BIM environments as well. De Ruvo [2019] also explores the suitability of the interoperability of HBIM-GIS frameworks in the cities of Quinson, Tolentino and Venice. It considers, among others, its usefulness in the representation of risks and hazards, considering the risk analysis in the context of the ResCult project. Another noteworthy example of the GIS-BIM-HBIM integration was developed by Pelliccio et al. [2017] for the Italian city of Sora. In this work, the authors use a comprehensive BIM 3D model for testing diverse environmental agents related to physical decays, including wind, solar incidence and water flow. In this specific case, the vulnerability analysis is concentrated in a limited part of the historical centre of Sora. However, the works mentioned above suggest that the use of GIS platforms for integrating multiple BIM models might also help to ensemble and frame a regional model. Another relevant aspect is related to the storage and management of the files that represent intermediate steps before the feeding of the GIS database. The GIS database would be able to be fed from different sources, such as on-site surveys and from data extracted directly from BIM-HBIM models by using a proper interface. Expectedly, the origin of BIM-HBIM models might be variable, since there is a vast number of actors interested in their creation and update, namely contractors, designers, government instances or academic institutions. Hence, it is possible to admit that the storage of BIM-HBIM models (as well as their background sources, such as point clouds) would be independent of the GIS database, only generating temporary links for feeding the database, which may be done by using cloud informatics services, for instance. Then, a preliminary design might consider an online cloud-based database with a distributed storage of the sources. 2.5 Vulnerability assessment methods. A vulnerability analysis of a historical centre is typically aimed at considering the effects of a determined event into this system. If the models of both historical centre and phenomena are representative, then it is possible to predict the interaction of the event and the system with a certain grade of reliability. The effects must be compared with a pre-established acceptability criterion in order to assess the need and 34 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). magnitude of mitigation. The disruption that the system may suffer as a consequence of the selected event can be studied through qualitative and quantitative analytical tools, for instance, Risk Matrix, Failure Mode and Effect Analysis (FMEA), Failure Tree Analysis (FTA) or Event Tree Analysis (ETA). 2.5.1 Risk matrix. A risk matrix is a semi-quantitative tool that allows relating the likelihood of a determined event against the expected effect that may have on the system, expressed in levels of severity. Different risk matrices can be constructed for a specific historical centre by confronting the relative frequency and intensity of past events that affected it, with their effects on the system. The acceptability of risk [Vrijling et al., 1998] is also addressed to determine the threshold from which a particular assumption of risk becomes reasonable. This is not only useful for assessing existing systems, but also for targeting performances and enhance specific aspects of the system. In terms of structures, for instance, this target safety [Holicky et al., 2015] is closely related to the existing codes and performance states. 2.5.2 Failure mode and effect analysis. In the context of risk assessment, one of the most relevant techniques for failure analysis is that of Failure Mode and Effect Analysis (FMEA). The formulation of FMEA intends to describe a system through its components and to separately analyse them in order to identify potential failure modes, their causes, and to assess the potential impacts of these failures in the entire system [Hollenback, 1977]. FMEA approach is especially useful to design a progressive and methodologic process to assess complex systems, in which a similar framework can be repeated through diverse subsystems or components. For FMEA analysis purposes, a failure mode is understood as a deviation from the required performance of a component, i.e., cessation of the ability to perform a critic function. This concept will have specific meanings depending on the semantic characterisations. A basic algorithm for FMEA application [Teng and Ho, 1996] may include: 1. Definition of the system’s scope. The division into components and the relationships (utilisations, functions, influences and interconnections) between these components are schematised. 2. Description of each component’s function, and their interactions with the others, resorting to block diagrams, for instance. 3. Selection of a primary component. 2.5. VULNERABILITY ASSESSMENT METHODS. 35 4. Selection of a function or a performance requirement to be tested. 5. Identification of all the potential failure modes associated with the selected component and function, determining under which operating conditions failure may occur. 6. Selection of a failure mode. Namely one considered as a significant risk. 7. Identification of the root causes of the selected failure mode. 8. Assessment of the direct, intermediate and end effects, tracking a chronological sequence. 9. Identification of the means or methods of detection, as well as existing control and mitigation measures. 10. Planning of supplementary measures and means that may improve or strengthen the element in the context of the selected failure. 11. Selection of a new failure mode. Repeat from point 7. 12. Once all failure modes were analysed, selection of a new function. Repeat from point 5. 13. Once all functions were considered, a new component will be selected. Repeat from point 4. 14. Sorting of potential failures. 15. Proposal of interventions. 16. Registering and documenting the entire FMEA process. 17. Monitoring of the system performance. This analysis may be especially suitable from the basis of the semantic codification of historical centres and their components. In fact, information such as the description of the component’s function, performance conditions and requirements, potential failures and the failure effects, is able to be coded together with the physical entity. In other words, the information needed for carrying the FMEA analysis can be integrated into the models and represented through them. For instance, buildings in which the electric or water supply is critical (e.g., a shelter), or buildings that may be compromised because of inaccessibility (e.g., a hospital). There are several literature proposals about the risk acceptability in the FMEA method context [Teng and Ho, 1996] , by using qualitative or quantitative indicators. One of the most important aspects to be taken into account is the existence of a societal risk that must be differentiated from individual risks. Societal risks are very sensitive to perception, even if several quantitative measures may be used for an objective approach [Vrijling, 1995]. Several expressions and representations (such as the FN diagram) are useful for contextualising a specific scenario [Stallen et al., 1996]. The results of the FMEA analysis may point out the most critical elements and their most likely failure mode. This critical failure can be further assessed resorting to Failure Tree Analysis and/or Event Tree Analysis. 36 CHAPTER 2. VULNERABILITY ASSESSMENT OF THE HISTORIC URBAN LANDSCAPE (HUL). 2.5.3 Fault tree analysis and event tree analysis. Fault Tree Analysis and Event Tree Analysis are logic trees (i.e., schemes) that represent the different cause-consequence relations existing between different events in the context of the main event. Fault Tree Analysis helps to calculate the probability of failure of a determined system from the analysis of the probabilities of determined disrupting events. On the other hand, the Event Tree Analysis helps to determine the probabilities of a system being able (or not) to cope with a determined disrupting event, taking as a base the chained effects that may be triggered from the main event. Then, both logic trees have in common a determined trigger event. Failure trees explain the events that may lead to the event itself, while event trees describe its consequences for the system. The combination of both threes may be found useful when analysing a very specific event (e.g., a particular earthquake) [Hollenback, 1977, Teng and Ho, 1996] When addressed to single buildings, the corresponding failure tree may regard the Ultimate Limit State of the structure, based on the different mechanisms of failure for masonry constructions. The event tree analysis may consider the potential effects that a collapse may have in the superior level, due to the presence of debris or loss of continuity in a certain road. Failure tree analysis helps to assess the events or chains of events that may imply a failure of the system because of the failure of their components. For instance, the city structural breakdown caused by the chained failure of technical networks, housing, business and/or public infrastructures that compromise the basic needs of the city. Time-dependant consequences of that failure can also be assessed resorting to an Event Tree Analysis. 2.6 Towards a comprehensive workflow for vulnerability assessment of HUL. By departing from the previously analysed information, it is possible to propose a comprehensive critical path divided into four groups of tasks: semantic descriptors selection, semantic description acquisition, modelling and assessment (Figure 2.5). The definition of the semantic descriptors involves the selection of the attributes that need to be coded in order to adequately describe the object in the context of a determined objective. It becomes relevant to select descriptors for the historical centre (streets networks or squares, for example) as well as its components (namely buildings). Since the goal is to carry a vulnerability assessment, it will be necessary to describe all elements that would be used in a Failure Mode and Effect 3.2. THE HISTORIC URBAN LANDSCAPE. 43 3.2 The Historic Urban Landscape. The transit on a predominantly globalising and urban world is an opportunity for questioning the role that urban heritage conservation has in the present-day and future urban dynamics. Cities are subjected to numeral pressure sources, such as touristic, demographic, economic, cultural, and infrastructure requirements [García-Hernández et al., 2017]. The growth of the cities is implicitly related to the cultural and economic variety. Most of the changes have been especially fast in the last century, creating a series of generalised and systematic problems that threaten the identity and culture of the communities. Even if globalisation imposes similarities in terms of financial dynamics, information technology, or urban development among the cities, their culture and heritage are still distinguishable and have unique resources in each one of them. These concerns lead to the design of an interdisciplinary and holistic approach: the Historic Urban Landscape (HUL). This approach pretends to guide the changes of the cities based on the recognition and identification of the natural and cultural, tangible and intangible, international and local values present in any city [UNESCO, 2016]. 3.2.1 Definition of the Historic Urban Landscape. The Historic Urban Landscape is both an approach and a definition: an approach for integrating the urban conservation in a generalised framework for sustainability; and a definition of the historic environment regarding its distinctive elements, identity, and richness [Van Oers, 2007]. The Recommendation on the Historic Urban Landscape [UNESCO, 2011] defines the HUL based on cultural and material traits. It recognises the city as the result of a historical layering of cultural and natural values and attributes, including its surrounding context. If analysed, the definition numbers a series of specific agents to be considered: «... 8. The historic urban landscape is the urban area understood as the result of a historic layering of cultural and natural values and attributes, extending beyond the notion of “historic centre” or “ensemble” to include the broader urban context and its geographic setting. 9. This wider context includes notably the site’s topography, geomorphology, hydrology and natural features, its built environment, both historic and contemporary, its infrastructures above and below ground, its open spaces and gardens, its land use patterns and spatial organisation, perceptions and visual relationships, as well as all other elements of the urban structure. It also includes social and cultural practices and values, economic processes and the intangible dimensions of heritage as related to diversity and identity.» 44 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. This definition is especially meaningful for contextualising the first of the so-called “Six Critical Steps”, which are considered as the minimum steps for implementing the HUL approach: “to undertake comprehensive surveys and mapping of the city’s natural, cultural and human resources” [UNESCO, 2016]. The examples of HUL implementation found in the literature demonstrate that these works of surveying and mapping are difficult to standardise due to the complexity of the HUL-associated phenomena [Rey-Pérez and Pereira Roders, 2020]. The challenge of having comprehensive surveys and maps implies dealing with multiple criteria for selecting significative attributes, having suitable structures and means for data coding and organisation, and designing adequate survey strategies. Besides, it is essential to assess the integration of the HUL in the context of transversal politics, such as the Sustainable Development Goals [Girard et al., 2015, Angrisano et al., 2016], national or local developing plans, etc. A comprehensive survey may include a wide variety of data, metrics, and indicators. For instance, the categorisation of the promoters and consumers. Landscape and cultural heritage indicators. Or even indicators associated with the performance of the different uses found in the HUL [Gravagnuolo and Girard, 2017]. This condition of the HUL implies the involvement of numerous stakeholders, experts, and professionals with specific interests and scopes towards the historic city. Many of these actors implicitly have specific approaches, namely associated with different ways for modelling and surveying the city. This divergence on the characterisations of the Historic Urban Landscape brings the pertinency of discussing the potential conciliation and integration of multiple interests for a comprehensive characterisation of the HUL. 3.2.2 Experiences facing the challenge of encoding the HUL. The singularities of the historic city make it challenging to propose universal and standardised approaches. Azpeitia Santander et al. [2018] explore how the development of indicators is still an open debate, referring to some valuable antecedents, such as the ones of the conventions of Vienna and St. Petersburg (Table3.1). Even if there is a categorisation of the city agents, each agent would demand a proper approach for being described. Thus, it becomes important to distinguish if a determined agent is reasonably characterised by its simple categorisation or if it is necessary to provide further descriptions. For instance, depending on the purpose of the description, if it is enough to declare the species of a determined tree for having a series of generic properties, or more detailed information is required to do so. Another interesting approach for categorising the historic city is provided by Guzman [2020]. This proposal summarises a series of indicators and their corresponding units of measure, based on a SWOT (Strengths, Weaknesses, Opportunities, and Threats) analysis designed in the context of the sustainable 3.2. THE HISTORIC URBAN LANDSCAPE. 45 Cultural–Tangible/Intangible Economic Tangible/Intangible Buildings, open spaces, green spaces, public space, designed ensembles, parks and gardens, composition, silhouette (horizontality with movements, views, events, and activities. Performance in terms of revenues (taxes, tourism, GDP), expenditures (on conservation), relation to the metropolitan/regional economy, marketing potential, and city image. Social Tangible/intangible Ecological Tangible/intangible Accessibility to the city (for living, working, leisure, services) for the population, quality of the housing units, range of housing categories (social housing, middle incomes, high incomes), civic pride. Biodiversity, water, air, water quality, air quality. Table 3.1: Key indicators discussed in the 2007 convention of St. Petersburg as referred by Azpeitia Santander et al. [2018]. development of the historic city (Table3.2). This approach intends to provide a generalised framework for organising a series of indicators that reasonably describe the city. However, since this system is based on the idea of urban performance, it may have some limitations for representing certain types of information, such as geographical parameters. Nevertheless, it represents an interesting approach for pursuing a holistic and comprehensive framework for characterising the city. Strengths Weaknesses Urban size; Protected areas; Green areas or recreational parks; Number of public libraries; Number of theatres and music halls; Festival and religious parties; Number of museums. Road network; Population density; Literacy rate; Air pollution; Accessibility; Housing; Deterioration phenomena; Marginalisation rate; Community involvement in decision-making processes; Population with access to healthcare. Opportunities Threats Research and development; Financial organisation; Number of police. Natural risks; Number of automobiles—road traffic; Crime level (robbery). Conflictive factors Conflictive factors (cont.) Number of schools; Number of markets; Productive sectors; Recreational or sport areas; Electricity and light infrastructure; Water supply; Telephone (access, visual disruption); Investment for intervention; Modes of transport; Access to a sewage system; Population with a university degree; Number of hotels. Table 3.2: SWOT analysis for the historical city of Queretaro (México). Adapted from Guzman [2020]. 46 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. The proposal of Kokla et al. [2019] pursues the semantic formalisation of historical centres by interpreting the definition of the HUL, conceiving that the HUL has a complex network of relations with tangible and intangible elements enclosed in an anthropic and natural environment (Figure3.1). Furthermore, the HUL description opens a series of categories that can have individual descriptions and hierarchies depending on the multiple meanings and uses that the historic city has (Figure3.2). Figure 3.1: Semantic formalisation of “historic towns and urban areas.” Adapted from Kokla et al. [2019]. Figure 3.2: Semantic formalisation of the HUL. Adapted from Kokla et al. [2019]. 3.3. TOWARDS A SUITABLE LOGIC FOR ENCODING THE HUL. 47 This approach is significant for illustrating a wide sample of fields of interest that converge in the HUL. Furthermore, it is important to recognise that the elements iden-tified in this approach can be further expanded in numerous fields and specialised areas. For instance, the unfolding of the many activities related to cultural heritage. Once again, this schematic representation of the HUL is an approach that suggests the complex challenge of offering a comprehensive description of the city. The works of Vehbi and Hoşkara [2009] suggest that the identification of representative indicators may result from enquiring a series of causal events and their measurable consequences. For instance, the assessment of the health of a community by measuring the percentage of mothers with adequate prenatal care or the total chronic disease-related deaths per 100,000 habitants. This approach permitted to match the outputs of numerous indicators and the corresponding causing factors for generating a set of sustainable urban revitalisation indicators, divided into economic, environmental, and social indicators. These sets (Table3.3) are the product of a workflow that includes the parameters’ validation. Many of the attributes considered by these authors share a certain attachment to the building-scale information. In fact, a significant number of urban-scale phenomena can be tracked from information gathered from a building and its permanent or temporary inhabitants. Hence, framing the data acquisition at the scale of a single building would permit the development of numerous surveys with complementary and variated purposes, including the quality of life associated with the cultural heritage [Battis-Schinker et al., 2021]. The review of these precedents reveals that the characterisation of the HUL is still an open and complex discussion that comprises a series of challenges related to the complexity of the city and the actors interested in its study. The design and implementation of generalised frameworks for the comprehensive characterisation of the city are still gaps, whose solution would be a key for the successful implementation of the HUL approach. 3.3 Towards a suitable logic for encoding the HUL. The characterisation of the current state of the historical city is a critical step towards implementing the HUL approach, which is complementary and interdependent to sustainable urban development. Such characterisation includes a vast number of indicators for modelling the dynamics of a city from the economic, social, cultural, and environmental points of view. Numerous indicators are strongly related to the dynamics between the people and the built environment, such as public services and house facilities. 48 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. Economic Environmental (Physical) Social Economy; Ratio of locally/nationally owned business in comparison with national/international business; Unemployment rate Employment diversity (Rate of privately owned business to public business); Local handcraft production rate Income level; Land and property prices (min./max. property prices); Property prices to income level; Rent prices to income level; Tourism; Ratio of tourism facilities in the area; Number and size of recreational; Cultural and spiritual sites; Development costs; Maintenance cost; Land value; Infrastructure; Environmental quality; Pollution levels: water, air, noise, visual pollution; Energy-transport-Energy-space heating; Transport; Accessibility level (block sizeinterconnected street, cul-de-sacs, sidewalks); Transportation (Variety of mode of transportation: bicycle, walking, vehicle); Proportion of car parking spaces to build up area; Pedestrian and bicycle-friendly streets; Public transportation; Land (obsolescence indicators— area/building condition); Structural condition of buildings; Proportion of various functions and green spaces; Ratio of built-up areas to open areas (Density of buildings); Ratio of buildings that worth to be preserved; Ratio of incompatible uses; Number/ratio of listed buildings; Rate of historical buildings with adaptive re-use; Proportion of preserved/restored buildings; Rate of the redevelopment of abandoned open space and old areas to the old pattern; Vacancy rate; Percentage of green space in built-up areas; Rate of multi-used buildings; Ratio of solid/voids; Type of public gathering spaces; Rate of use of local construction materials and techniques in new development; Rate and number of public open space with recent intervention. Income; Income level; Household income disparity Mixedincome; Poor households; Rent-to-income ratio; Crime; Level of safety; Crime ratio; Safety; Health; Sense of well-being; Proportion of population who find their living environment good; Ratio or recreational, leisure activities to the number of inhabitants; Variety of community facilities (educational, health facilities); Existing laws and regulations on urban revitalisation sufficient or insufficient; Accessibility of green space; Availability of local services; Accessibility of local services; Availability of local public open areas and services; Household connection to infrastructure; Number of people using recreation facilities (aware of social and cultural facilities); Household Homeownership rate; Age; Population Cultural/Racial diversity; Civil involvement; Duration in neighbourhood; Housing; Mixed choice of housing; Level of participation in the decisionmaking process; Floor area per person; Housing price to income ratio Housing tenure types; Housing affordability. Table 3.3: Sustainable urban revitalisation indicators. Adapted from Vehbi and Hoşkara [2009]. 3.3. TOWARDS A SUITABLE LOGIC FOR ENCODING THE HUL. 49 However, the task of reuniting apparently divergent attributes may be problematic when designing a unified scheme for encoding information. Hence, it is pertinent to analyse the step before the selection of the indicators and attributes to be surveyed and discuss how a particular set of characteristics of a physical entity can be organised and enriched regardless of the nature of data and the specific purpose of the model. In other words, it becomes pertinent to discuss purely ontological approaches for generally describe and analyse entities, departing from an unbiased interest. 3.3.1 The Aristotelian Causes as an approach for categorising the surveys. The accuracy of the description of an entity is necessarily dependent on the formulated questions. In this sense, the singularities of the cities (and their components) make it challenging to establish a standardised framework. Nevertheless, as discussed before, several authors agree on studying the impact that singular agents have in the historic city’s context from the economic, social, cultural, and environmental points of view. To systematise the data acquisition for understanding each one of these dimensions is not necessarily a single work but the sum of numerous sets of suitable and specific surveys. However, the discussion is still open, and there are several intents for proposing robust frameworks for formalising and organising the sets of characteristics that define the HUL and its elements. A plausible approach for categorising the layers of information to be surveyed from an entity can be understanding it through the Aristotelian principle of the four causes. This approach fundamentally accepts that there are four elemental answers to the question “why?” when analysing an entity: the material, formal, efficient, and final causes. This theory frames any entity in a hypothetical continuum of causes in which each one of the elements that compose a determined entity has its own sets of elemental causes [Gómez-lobo, 1996]. Generally speaking, the material cause regards the tangible matter that represents the object in the physical world. The formal cause is the arrangement, organisation shape, appearance, or spatial consequence of the entity. The efficient cause is the set of external agents whose intervention permits the existence and performance of the object [Heidegger, 1977]. The final cause is the purpose and/or function that an object has [Hocutt, 1974]. It is relatively easy to recognise these elemental components in simple and paradigmatic objects, but a more complex analysis may make it difficult to categorise the causes in a precise manner. A classic example consists of analysing a table. In this case, the material cause may be wood, metal, plastic, or any other material. The formal cause will be the geometrical description of the table. The efficient cause would be the carpenters that fabricated the table. A final cause for the table would be to eat on it, for 50 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. example. The assessment of a building is not that simplistic, of course. However, the categorisation of some sets of attributes would help organise the information. Furthermore, a nested organisation of the layers of information would permit the design of a unified framework. The interest of analysing entities by using this approach lies in the possibility of establishing a unified logic for successively exploring the parts of an entity until the level of information detail is considered satisfactory for a determined purpose. For example, we can declare that the material cause of a building is “masonry.” This answer may be satisfactory enough for a specific purpose, such as cataloguing, but insufficient for obtaining relevant information about the structure. Hence, it would become relevant to explore the causes of the “masonry”, i.e., the exact attributes that specifically belong to “that” masonry. This nested and enlargeable structure permits categorising the fundamental causes of an entity regarding the four different types of causes, being that a cause can be related to an unlimited number of instances able to be further analysed. It is possible to individually describe these instances or categories by categorising their own causes. The process of categorisation and description has a certain fractal development, in which it is possible to explore diverse elements of an entity with different levels of deepness without losing consistency. An example of this procedure is depicted in (Figure 3.3), where a plausible structure for finding a causal link between the HUL and the mortar of a determined building. Figure 3.3: Example of a nested analysis of causes. 3.3. TOWARDS A SUITABLE LOGIC FOR ENCODING THE HUL. 51 When analysed from its fundamental causes, the HUL may have a series of instances. For example, the material causes of the HUL (i.e., the tangible components of the HUL) may be itemised as a series of generic instances, such as green areas, water bodies, and buildings. Naturally, each instance can be further analysed and categorised. In this example, there is interest in unfolding the causes for the constructions, and consequently, a series of instances is proposed according to each fundamental cause of them. It is worth noting that this categorisation can be as exhaustive as necessary, framing more specific instances, such as “houses” instead of “buildings”. The analysis of the material causes of a building proposes a series of instances that determine the materiality of the construction. The process of determining these instances has been extensively explored in the past, namely in the context of the Building Information Modelling approaches. In fact, the standardisation of components and libraries (namely based on the IFC standards) offers a robust point of departure [European Commission, 2017]. Finally, when exploring the material causes of a wall, it is relevant to find a suitable position for establishing the instance of “mortar” and its further analysis. The causal analysis, however, can be supported by some generalising ideas. For a building, the material causes would be all the substances that give tangibility to the construction. The formal cause would be the spatial consequences of the building: its geometry, shapes, volumetrics. The efficient cause is the set of agents that possibilities the existence of the building. However, in a broad sense, it is not only who built it and how, but what does permit to keep the building’s ontological efficiency and/or performance. For instance, a house without a water supply is not properly working according to the “house concept”. It is important to recognise that the efficient cause is closely related to the expected performance of the entity. In other words, it is closely associated with the final cause, which represents the use of the building, its users, and its dynamics of use. To feasibly generate a flexible approach based on the Aristotelian principles, it is necessary to design and manage databases able to be enlarged as much as necessary, permitting to store and manage qualitative and quantitative data without losing the structured organisation. The control of the granulometry of the information for representing the built environment has already been explored for representation purposes in digital models. Approaches, such as the Level-Of-Detail for three-dimensional modelling, propose generalised frameworks for defining thresholds for the information contained in a model [Ramírez Eudave and Ferreira, 2021b]. However, this approach does not observe the flexibility for selectively having different levels of detail in multiple instances of the same model [Biljecki, 2017]. This customised compartmentalisation of the information is expected to satisfy the need of the various stakeholders of the HUL for an accumulative, multidisciplinary, and coherent framework. 52 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. While most of the examples found in literature build on a tree-structured organisation of the information, the Aristotelian approach is closer to a fractal development. Since the steps between instances are selfsimilar (i.e., based on the same set of four causes), the structure of successive steps (regarding the detail or granulometry of information) may be understood as a progressive unfolding of causes in different scales. A very representative schematization of this recursive process may be found in the so-called “Cantor dust”. This fractal progression departs from a square that is subsequently divided into four subsets. Suppose we admit that each step corresponds to the fragmentation of an instance in its four fundamental causes. In that case, it is possible to see that the structure represents the progressive nature of this analysis. Furthermore, from this geometrical schematization, one can observe that the process of fitting the descriptors needed by multiple stakeholders in this structure allows assembling a coherent and comprehensive description in upper levels. As shown in Figure 3.4, different stakeholders would need to analyse diverse aspects of the entity with varying levels of granulometry of data. For example, a determined stakeholder may be interested in knowing if a specific construction is made of masonry, whereas another stakeholder would need to specifically know which kind of stone is that masonry made of (Fig.3.5). Thus, if clearly structured and organised, the information generated by a stakeholder can be potentially valuable as a full integrative description. Figure 3.4: The “Cantor Dust” fractal as a schematic representation of the proposed approach. The level of specificity or granulometry of information is here schematized by departing from the more general level of information (0) and subsequent stages of more detailed analysis (1), (2) and (3). 3.4. WORKED EXAMPLE OF APPLICATION. 59 –Q18. Presence of voids [Y; N] –Q19. Transversal connection between layers [Y; N] –Q20. Presence of visible cracks [Y; N] –Q21. Depth of cracks: coating or structural [None; coating; structure] –Q22. Signs of repaired cracks [Y; N] –Q23. Cracks due to deformation and/or settlements [Y; N] –Q24. Bending and oblique cracks [Y; N] • FP8. Replacement of the floor system –Q25. Ratio of concrete floor systems/total of flooring systems (%) • FP9. Connection to orthogonal walls –Q26. Metallic strapping elements, tie rods [Y; N] –Q27. Good laying and locking of masonry [Y; N] –Q28. Deformations, disconnections, detachment, or embrittlement • FP10. Connection to orthogonal walls –Q29. Ratio of diaphragms efficiently connected to the façade (%) –Q30. Signs of deformation, rotting or retraction [Y; N] –Q31. Signs of fragility in the support zone [Y; N] –Q32. Signs of distortion [Y; N] –Q33. Lack of circulation safety [Y; N] • FP11. Impulsive nature of the roofing system –Q34. Span (m) –Q35. Height (m) –Q36. Symmetric or asymmetric [Symmetric; Asymmetric] –Q37. Thrust-cancellation elements [Y; N] –Q37. Strapping reinforce perimeter [Y; N] –Q38. Conservation status [bad; regular; good] –Q39.Strapping reinforced perimeter or tie rods [Y; N] • FP12. Elements connected to the façade –Q40. [none; light (light, signs); medium weight (equipment, AC); heavy (balconies, parapets)] • FP13. Improving elements 60 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. –Q41. Exterior stairs, arches, giants, etc. [Y; N] –Q42. Strengthening actions [Y; N] –Q43. Presence of reinforced plasters (e.g., with meshes) [Y; N] The grading for each parameter depends on a series of qualitative or qualitative attributes. Hence, a parameter is, in fact, the result of a series of basic conditions. Thus, it is necessary to establish a list of individual attributes that comprise all the information needed for categorising the parameters. The queries will be then set in the corresponding attributes table of the GIS database so that they can be individually obtained during a field campaign. As discussed in the previous section, each individual query can be framed in a specific section of the causal HUL schematization, which translates into the relationship between the attribute, indicator, or metric and the other elements of the building and the HUL. Then, it is necessary to consider the specificity of each query in the context of the deconstruction of the HUL in successive causal instances. The queries present a variety of types and potential answers. For example, there is a series of queries where the answer is meant to confirm or deny a specific aspect (e.g., if all the openings have a regular size). Some others need the introduction of quantitative information, such as the height and length of the façade, or imply the selection of a condition from a closed set of options. All these non-numerical queries can be determined in the QGIS software under the scheme of a map of values, which permits further data treatment. 3.5 Results. 3.5.1 Fitting of the instances in the causal structure of the HUL. It is worth recognising that the organisation of the parameters and queries of the Vulnerability Index Method are not comparable because the queries correspond to different instances and levels of granulometry. This independent development is expected to occur for almost any set of attributes and/or customised survey. On the other hand, the organisation based in the HUL causal structure is expected to permit the localisation of virtually any potential attribute, metric, or query used to describe the elements of the HUL. The organisation of the queries in the HUL scheme (Fig.3.9) unfolds from the HUL itself, at the top of the structure. It is possible to identify a series of intermediate steps that may have been considered between the HUL and the buildings. 3.5. RESULTS. 61 For this exercise, buildings are considered as one of the material causes of the HUL. It is possible to find that the material causes of the HUL are even more general, such as an initial categorisation between natural and artificial objects. Since the structure is expandable without losing the order and hierarchies of the granulometry, this is not considered an issue nor a limitation but a potential part of an enlarging process. Figure 3.9: Schematization of the HUL structure by using the Cantor Dust fractal representation, the correspondence with the queries, and the descriptive stages of granulometry of the information. The analysis of the building allows for a first approach to the most general queries, those related to the exterior geometry and spatial description of the building and its adjacencies. It is interesting to note that this organisation and the level of granulometry has a logic development that is similar to the Level-of-Detail approach, suggesting a potential convergence in future analysis. 62 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. As a material instance of the HUL, buildings are subsequently analysed in their four fundamental causes for finding the next level of information. Most of the queries for the parameters are related to the structure of the building. The material causes of the building were unfolded in three instances (that would be more for a different analysis): façade, flooring, and roofing system. These instances are primarily considered as material causes since they represent general parts of the tangible substance of the construction. The analysis of the first instance—the façade—, permits us to answer some more specific queries. A very illustrative query is Q40, which relates to the existence of elements connected to the façade. From the point of view of those elements, the façade is the support, and, in consequence, it is analysed from the final cause. The formal causes of the façade permit the extraction of metrics related to its dimensions, whereas the efficient causes permit to inquiry about the damages and pathologies that condition the structural performance of the façade. It is important to note that the analysis of the attributes of the masonry is conveniently divided into a subsequent level. The formal cause of the façade is unfolded in the first instance: the openings. Even if this unfolded instance is only analysed from its formal causes (namely the size, organisation, and alignment of the openings), it can be used in the future to include additional information with a different purpose. Just as for the formal cause of the façade, the material cause has been unfolded in a first instance, that of masonry. It is important to note that this approach is for masonry façades only, but the causal analysis does not exclude the possibility of unfolding several material-related instances. The causal analysis of the masonry permit to summarise its geometrical configuration and potential deformations (formal causes), the type of units used (material causes) and its construction system, quality, and status (efficient causes). The second instance of the building’s material cause—the flooring system—is analysed from its material, final cause, and efficient cause. The analysis of the efficient cause of the flooring illustrates how a series of conditions related to pathologies and decays of an element is an integral part of its efficiency. Finally, the third instance for the building’s material cause—the roofing system—is described in terms of geometry (formal causes) and construction elements that permit its performance as covering (efficient causes). When articulated together, the scheme and the queries illustrate a process of selection and organisation for attributes that can be replicated for different and variate purposes, though permitting the assembling of a comprehensive and complementary database. The instances and levels analysed for this example are only a first approach that would be enlarged as needed according to the requirements of other analysis and agents. 3.5. RESULTS. 63 3.5.2 GIS database setup. The church of São Miguel was selected and tagged based on an existing GIS database of the historical city of Guimarães. It was possible to draw a preliminary polygon resorting to the Open Street Maps public database as a primary source. The base map was created in QGIS (Version 3.12 Bucuresti). The representative polygons of the buildings were isolated in an independent layer, in which the queries were encoded in the corresponding attribute table (Fig.3.10). When programming the queries, it is essential to distinguish the type of variables to be encoded: those that correspond to real numbers (such as measurable properties), the logic true/false fields, and the multiple value ones. Figure 3.10: Screenshot of the attribute table and the empty fields to be obtained. Once the 43 queries were adequately encoded, the project was synchronised in the Mergin cloud storage service. It is important to remark that the present-day free hosting service is limited to 100 MB, which can limit the inclusion of attached files (namely photographs) as part of the survey. This issue, however, is not relevant in the context of this example. It is also worth noting that this workflow uses only one of the potential open-source-based software sets and does not exclude similar strategies with other tools. 64 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. 3.5.3 Data acquisition. When in the server, the project can be then downloaded and consulted locally using the app Input (version 0.9.2) on any mobile equipment running an Android or iOS system. This app allows to access the QGIS project and to capture the data gathered on-site by fulfilling the queries proposed for the attribute table. Since the app permits a real-time position in the georeferenced map, the constructions can be identified easily (Fig.3.11). The information gathered through the Input interface was synchronised with the cloud service, allowing its storage, consult, and managing within the QGIS interface (Fig.3.12). Besides, the QGIS software possibilities to export the data in various formats, facilitating further analysis and use for different stakeholders, namely for carrying external processes. The GIS database offers, in fact, the possibility of executing relatively simple operations fed by the data in the attribute table, but it may be limited for certain purposes. Figure 3.11: Screenshots of the app Input during the on-site survey work. 3.5. RESULTS. 65 Figure 3.12: Screenshot of the QGIS interface displaying the survey of the Church of São Miguel. Even if the data treatment for carrying the vulnerability assessment method exceeds the scope of this exercise, it is pertinent to note that the grading of the building can be carried out by feeding a simple conditional-based algorithm (even in a datasheet). This process can be repeated for covering sets of buildings, such as in an HUL. The information gathered during this process is shown in Table 3.4. Query Value Query Value Query Value Query Value Query Value Q1 8.15 Q10 None Q19 true Q28 true Q37 true/good Q2 7.65 Q11 None Q20 true Q29 100 Q38 false Q3 1.12 Q12 Stone Q21 N/A Q30 false Q39 false Q4 16 Q13 N/A Q22 true Q31 false Q40 None Q5 true Q14 Good Q23 true Q32 false Q41 false Q6 true Q15 False Q24 true Q33 false Q42 false Q7 true Q16 Well carved Q25 N/A Q34 6.15 Q43 false Q8 true Q17 true Q26 true Q35 1.38 - - Q9 true Q18 true Q27 true Q36 Symmetric - - Table 3.4: Data obtained during the survey. 66 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. 3.6 Discussion. The above-proposed exercise illustrates a sequential workflow for contextualising an approach for the Historic Urban Landscape in the Aristotelian theory of the causes by departing from a specific and realistic set of attributes to be surveyed for reaching a specific purpose. This process is expected to occur with different sets of attributes and multiple purposes, according to the variety of actors, stakeholders, and agents that converge in the Historic Urban Landscape. This exercise accepts that the agents interested in the HUL depart from their own surveys and models. However, this example illustrates how the enlargement and enrichment of the database would open a wider horizon for exploring attributes that are not usually considered. A worth noting issue identified during the process was the need to deconstruct some general attributes in smaller parts, namely those that consist of simple queries. This process is relevant in the context of data that different agents can further use. For example, we can accept that numerous agents would be interested in knowing the length of the façade while, in the example provided, this is a component used to evaluate some of the vulnerability assessment parameters, not a final product. A certain dose of subjectivity was found when classifying the queries in the HUL causal structure. This is one of the challenges for the first steps of defining a causal structure. However, it is important to note that the addition of more queries and the identification of their interdependence permits clarifying the logical development of the structure. Even if there would be some difficulties for a priori classification, the immediate upper and lower-level analysis may help identify the suitability of the classification. A more extensive set of indicators and instances would certainly help the design of a robust structure. Furthermore, already existing criteria and classifications, such as those of the Industrial Foundation Class (IFC) for BIM models and Level of Detail (LoD) for urban 3D models, would help provide standardised frameworks during the causal structure design. One of the implicit advantages of dividing the parameters into relatively simple queries is the compatibility of this approach with the proposed data acquisition procedure. Although, as demonstrated by the example provided, a relatively small survey may imply a wide set of queries, the proposed workflow permitted to easily acquire, store, and synchronise the information during an on-site campaign. The use of the default types of data that a QGIS file accepts for characterising entities in a layer was found enough for programming the queries. Furthermore, no incompatibilities were found while carrying the complete process of the database setup, the cloud storage, the on-site data acquisition in a mobile device, and the online synchronisation. This kind of surveys can be developed on existing GIS databases. 3.7. FINAL REMARKS. 67 Plus, it is not even necessary to have polygons or geometrical entities for acquiring information since any surveyor can store single points associated with a determined survey. This approach would be helpful in census and surveys non-related to buildings, such as urban furniture, vegetation, or events. Then, it is consistent with the flexibility and robustness needed for surveying the multiple layers of the HUL. Some limitations found during the exercise were related to the impossibility of editing the attribute table from the app Input. This may be an issue if there is a problem while programming the survey. Another potential issue is that the information stored by a determined surveyor would overwrite existing data, namely when the file was not synchronised and a determined entity appears as empty. In fact, the workflow depends on a manual frequent refreshing of the file for visualising the latest changes in the file. The complete experience would be enriched by the repetition of this process in the context of a very different survey related to the HUL for testing the enrichment and enlargement of the HUL causal structure when several and interdisciplinary approaches have occurred. 3.7 Final remarks. The interest in understanding and analysing the historical city as a complex phenomenon has supported the development of the Historic Urban Landscape approach. This methodology is intended to unify the multiple ways of understanding and living the city, including the point of view and needs of multiple stakeholders, helping decision-making. The HUL approach’s success primarily depends on the surveying and mapping of the natural, cultural, and human resources of the city. There is no unique approach for documenting the HUL since the stakeholders need specific sets of information for their processes. However, some generalising approaches have been proposed. A feasible approach for reading and describing the HUL involves recognising it as a series of entities linked by causal relations. The analysis of these relations through the scope of the Aristotelian theory of the causes permits the establishment of logical structures for exploring specific dimensions of the urban phenomenon in a generalised framework, respecting the relations between the components. This categorisation and description process is valuable for organising the information generated from multiple actors, articulating a comprehensive description of the urban entity. The Aristotelian causes approach, based on identifying the material, formal, efficient, and final causes of the constituent entities, permits to reach levels of detail of the descriptions as deep as necessary, nesting a series of surveys. The implementation of the Aristotelian causal structure for organising the information requires conceiving the HUL system as a series of nested instances. This exercise would be unusual in 68 CHAPTER 3. COMPREHENSIVE SEMANTIC DESCRIPTIONS FOR THE HUL. the context of specific and specialised approaches to the historical city but is especially meaningful while designing a common framework for understanding the HUL as a holistic, complex, and comprehensive set of phenomena. One of the interesting purposes of this approach is the potential conciliation and integration of already existing works and views on the historic city. Resorting to a standardised framework to survey the city also allows that the parameters (deconstructed in basic instances and causal-related queries) can be investigated utilizing relatively simple means, such as attribute tables within QGIS databases. Furthermore, the availability of open-source software and its integration with cloud-based storage permits sharing and managing GIS files from portable devices. This approach facilitates the on-site surveying of the city, which may support wide collaborative networks for documenting the HUL, reaching the goal of having a holistic, interdisciplinary, and complementary view of the historical city. 4.2. THE MEXICAN NATIONAL CATALOGUE FOR HISTORICAL MONUMENTS. 75 Figure 4.1: Screenshot of the Catalogue’s online public platform with the fields contained in the database: https://www.catalogonacionalmhi.inah.gob.mx/ • Type of datasheet. –Religious/funerary/other. • Localisation –State, Municipality, Region. –Street and number. –Type and name of settlement/village/neighbourhood. –Postal code. –Surrounding streets (including a schematic localisation plan). –National Monument Zone (if any). –Public Property Register number. –ITRF92 or WGS84 geographic coordinates, including height. • Identification –Classification as historical (built before 1900) or cultural monument (built after 1900). 76 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. –Original use. Category: domestic, commercial and services, public administration, justice facilities, financial and custom facilities, military building, culture and sports, schools, health facilities, religious, funerary, commemorative, civil infrastructure, hydraulic infrastructure, domestic and craft industry, agriculture and production, parks and gardens, public spaces; Genre (a subtype of the category, if any). –Architectonic typology (if any). –Name, when the building has a well-known denomination. –National Monument Zone (if any). –Present use. Category: domestic, commercial and services, public administration, justice facilities, financial and custom facilities, military building, culture and sports, schools, health facilities, religious, funerary, commemorative, civil infrastructure, hydraulic infrastructure, domestic and craft industry, agriculture and production, parks and gardens, public spaces; Genre (a subtype of the category, if any). –Architectonic typology (if any). –Urban use classification: housing, commerce, services, industry, infrastructure. –General and specific use, if any. –Lost asset. When there is an existing datasheet or register of a building that no longer exists when the survey is performed. • Legal aspects –Property: private or public. If public, user and responsible for the building. –Federal register and property register. –Federal (national) Declaration as a National Historic Monument. –Listed in a Historic Monuments Zone. –Located but not listed in a Historic Monuments Zone. –State declaration as monument / relevant building. –Municipal declaration as monument / relevant building. –Regulated by an urban plan. –Other registers existing in different databases or archives. • Religious and administrative aspects (if any). –Original patronage, devotion or advocacy. –Present patronage, devotion or advocacy. –Foundation. 4.2. THE MEXICAN NATIONAL CATALOGUE FOR HISTORICAL MONUMENTS. 77 –Administrative adscription. –Category: basilica, cathedral, parish, vicary, chapel, sanctuary, convent, hermitage, mission, bishopric, inquisition, mosque, synagogue, temple, etc. –Religious order, if any. • Historic information –Construction date (century). –Date of interventions (century). –Previous unrelated constructions. –Type of structure (pre-Hispanic, historical, other). –Location of the previous unrelated construction, if any. –Historical information (brief). • Sources –Testimonies, archives, documents, inscriptions, etc. • Environment –General description. • General terrain attributes –Number of storeys. –Total, built and free surface (in square meters). • Architectonic description. –Text field for stylistic and semantic descriptions. –Formal and material properties. –Architectonic plan: basilica, central patio, sided patio, Greek cross, Latin cross, polygonal, radial, single corridor construction, single nave, other. –Main façade: *Material: Adobe, brick masonry, stone masonry, tiles, metallic sheet, stone, metals, other. *Description. *Conservation state (good, regular, bad). –Vertical structure: *Material: Adobe, brick masonry, stone masonry, tiles, metallic sheet, stone, metals, other. *Description. 78 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. *Conservation state (good, regular, bad). *Thickness (up to two measures, minimum and maximum, if applicable) . –Floor structure: *Material: wood structure, metallic structure, iron, bricks, reinforced concrete slab, stone, mixed systems, etc. *Geometry: vault (and its varieties), plain *Approximate height (up to two measures, minimum and maximum, if applicable). *Description. *Conservation state (good, regular, bad). –Ceiling: *Material: wood structure, metallic structure, iron, bricks, reinforced concrete slab, stone, mixed systems, etc. *Geometry: vault (and its varieties), plain, oblique, etc. *Approximate height (up to two measures, minimum and maximum, if applicable). *Description. *Conservation state (good, regular, bad). –Floor finishing: *Material: cement, tiles, stone, brick, ceramic, wood, etc. *Description. *Conservation state (good, regular, bad). –Stairs: *Material: cement, tiles, stone, brick, ceramic, wood, etc. *Geometry. *Description. *Conservation state (good, regular, bad). • General state of conservation • Mobile cultural assets –If any • Risk Analysis (exposure, subtype and emergency plan, if applicable) –Geologic (landslides, earthquakes, tsunamis, vulcanism) –Hydrometeorological (hail, frost, terrain sinking, hurricanes, river or rain flood, snow, drought, storms, extreme winds) 4.2. THE MEXICAN NATIONAL CATALOGUE FOR HISTORICAL MONUMENTS. 79 –Physical-chemical (explosions, toxic leaks, fires, radiation) –Sanitary (pollution, epidemics, plagues) –Societal-organisational (aerial accidents, maritime accidents, terrestrial accidents, massive people concentrations, electricity interruption, manifestations, terrorism, etc. –Electric facilities, conditions and observations. –Hydraulic and drainage facilities, conditions and observations. –Gas facilities, conditions and observations. –Object susceptible to failing or slip, level of danger associated –Emergency stairs, state and observations –Fire combat equipment, existence and observations –Safeguard of cultural assets, present state and observations –Signals, existence, conditions and observations –External surrounding hazard sources, if any • Priority of attention –Urgent, corrective, preventive –General observations • Attachments –Localisation plan –Architectonic schematic plan –Photography of the main façade –Photography of a secondary façade, historic photography or interior detail –Others (maps, videos, documents) • Surveyors –Name, adscription, supervision The information on relevant structural and seismic performance-related aspects is limited since, as mentioned, the catalogue has a different purpose. Nevertheless, the catalogue data are potentially compatible with the use of simplified structural and vulnerability assessments approaches, namely those based on index-based indicators. After high-scale seismic events, more attention has been paid to the need for assessing the seismic vulnerability of historical constructions. On the 7th of September, 2017, an Mw= 8.2seismic event hit the south of Mexico. Just a few days after, on the 19th of September, an Mw= 7.1earthquake hit the centre of the country, affecting Mexico City and the surrounding areas. In an ominous coincidence, this earthquake coincided with the 80 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. 32nd anniversary of the most destructive earthquake in the history of Mexico City, and it happened just a few hours after the commemorative event. Preliminary information declared around 150,000 affected buildings [Alberto et al., 2018]. In the sequence of this seismic crisis, the information contained in the National Catalogue of Historical Monuments was used to conduct a preliminary estimation of the damages on the affected historical assets. From a total of 2340 damaged buildings, 1450 were labelled with moderate and severe damages. 61.4% of the damaged buildings were located in the states of Oaxaca, Puebla and Morelos [San Juan et al., 2017]. These states figure respectively in the third, fifth and seventeenth place in the 2015 national marginalisation index [Consejo Nacional de Población, 2016],which is a relevant aspect to explain the profound impact of these events in social and economic terms, as well as the challenge for the resilience process after the events. The recurrence of seismic events in Mexican territory is a significant threat for a broad universe of buildings with unknown seismic safety, and the reason why the possibility of using the information contained in the catalogue to support a nationwide seismic vulnerability assessment campaign is very promising and deserves in-depth analysis. 4.3 Seismic vulnerability assessment. There is currently an increasing interest in sustainable urban development, which is frequently associated with the reuse of existing constructions. However, the actions on existing constructions must be preceded by knowledge about their present state and structural safety. This information is not only relevant for carrying actions on individual buildings, but also for studying phenomena that involves clusters of historical constructions, such as the case of historical centres. The information regarding entire groups of buildings’ behaviour is especially meaningful for understanding post-seismic scenarios since the consequences of damages on individual constructions will configure entire urban-scale assessments. The anticipation of urban-scale effects of earthquakes is relevant for developing strategies on the use and management of the historical assets as cultural and economic resources. Forecasting damages would help to minimise losses by identifying the assets at risk, defining intervention priorities, and adopting appropriate preventive measures and emergency plans. Despite the numerous developments in structural analysis of historical constructions, the obtention of urban-scale assessments is still challenging, namely due to the wide diversity of typologies, conservation states, and singularities inherent to historical constructions. 4.3. SEISMIC VULNERABILITY ASSESSMENT. 81 Most of the existing assessment methods can be generically framed into three main groups: empirical (based on records, observations and meaningful typological features), analytical (based on simplified mechanic analysis) and hybrid, including sources such as expert judgment [Calvi et al., 2006]. Some approaches are intended to establish correlations between seismic intensities and estimated damages according to a limited set of architectonic and structural parameters, which facilitate to carry urban-scale assessments. These parametric approaches represent a reasonable effort in the context of wide samples, mostly as a first-level approach [Ferreira et al., 2019]. The parametric approaches are intended to present ranges for estimated damages as a function of specific earthquake intensity and an individualised vulnerability index. 4.3.1 Index-based methodologies to assess masonry buildings. Seismic vulnerability is a component of the seismic risk, together with the hazard—which can be defined as the probability of occurrence of a determined intensity seismic event in a certain site and during a determined period [Carreño et al., 2007]—and the exposure— which is related to the set of material assets and human lives that would be potentially affected. A comprehensive characterisation of risk depends thus on an accurate assessment of the vulnerability, which represents the sensitivity (expressed as a grade of damage) that a building would have towards a specific seismic event. In addition, out of the three risk components, vulnerability is the only one possible to be reduced through engineering intervention. Therefore, reducing the exposed assets’ vulnerability is the most viable and efficient means of reducing the overall seismic risk. A relevant system for assessing the seismic vulnerability of constructions is the Vulnerability Index Method, originally designed and applied in Italy. Developed in 1994 by the National Group for the Defence from Earthquakes (Gruppo Nazionale per la Difesa dai Terremoti—GNDT), this approach aimed at facilitating the data acquisition of existing buildings by means of a comprehensive but straightforward survey datasheet [Benedetti and Petrini, 1984, Gruppo Nazionale per la Difesa dai Terremoti—GNDT, 1994] The predominance of masonry among historical buildings, a material whose characterisation and evaluation are still difficult, represents a significant challenge when assessing their seismic vulnerability. Nevertheless, some general assumptions may be the basis for establishing simplified behaviour assumptions, such as a brittle response in tension, a frictional response in shear and anisotropy [Roca et al., 2010]. Based on the analysis of a large volume of damage data, a set of characteristics has been highlighted as especially meaningful for influencing the seismic response of masonry buildings. This set of attributes is the base for a field survey in which every parameter is classified with four different possible grades that 82 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. will feed an arithmetic calculation for obtaining a global index [Calvi et al., 2006]. Grades express how a determined condition increases or decreases the vulnerability of the constructions. Initially composed of eleven evaluation parameters, the approach has been reworked, expanded and calibrated by multiple authors over the last years. More recently, Ferreira et al. [2017c] and Maio et al. [2020] have proposed an updated version of this approach with fourteen parameters, see Table 4.2. Similarly to the original version of the method, the overall vulnerability index is given by the sum of the products of each parameter vulnerability class multiplied by a given weight that reflects the relative importance of the parameter in the overall formulation, see Eq. (4.1). The independence between the class values and the weights permits adjusting and calibrating the methodology to local typologies’ specificities in which each parameter may have a more significant or a smaller impact on the buildings’ seismic behaviour. Parameters Class Cvi Weight pi Relative weight A B C D Group 1 Structural building system 50/100 BP1. Type of the resisting system 0 5 20 50 2.50 BP2. Quality of the resisting system 0 5 20 50 2.50 BP3. Conventional strength 0 5 20 50 1.00 BP4. Maximum distance between the walls 0 5 20 50 0.50 BP5. Number of floors 0 5 20 50 0.50 BP6. Location and soil condition 0 5 20 50 0.50 Group 2 Irregularities and interaction 20/100 BP7. Aggregate position and interaction 0 5 20 50 1.50 BP8. Plan configuration 0 5 20 50 0.50 BP9. Height regularity 0 5 20 50 0.50 BP10. Wall façade openings and alignment 0 5 20 50 0.50 Group 3 Floor slabs and roofs 18/100 BP11. Horizontal diaphragms 0 5 20 50 0.75 BP12. Roofing system 0 5 20 50 2.00 Group 4 Conservation status and other elements 12/100 BP13. Fragilities and conservation status 0 5 20 50 1.00 BP14. Non-structural elements 0 5 20 50 0.75 Table 4.2: Vulnerability Index parameters, according to the calibration of Ferreira et al. [2017c]. I∗ vf = 14 ∑ i=1 cvipi(4.1) 4.3. SEISMIC VULNERABILITY ASSESSMENT. 83 The parameters are divided into four main groups: generalities of the structural system, irregularities and interaction, flooring and roofing systems, and conservation status and non-structural elements. Each parameter is associated with a specific grade for classification purposes. In the context of Mexican constructions, a proposal of adaptation has been explored by Salazar and Ferreira [2020]. In this work, the authors have reviewed and adapted the original formulation of every vulnerability parameter in order to make them suitable to the typologies and the codes currently in force in Mexico, such as the Mexico City Building Code (Reglamento de Construcciones para el Distrito Federal–RCDF), the National Institute for Anthropology and History (INAH) and the National Centre for Prevention of Disasters (Centro Nacional para la Prevención de Desastres–CENAPRED) recommendations. Each one of the parameters contains a series of qualitative or quantitative conditions that guide the selection of a determined vulnerability class. However, certain conditions may be unclear even with the help of the guides. Details on the evaluation parameters that compose this adapted approach have been extensively described by Salazar [2018] and are included in Appendix A. 4.3.2 A vulnerability Index Method to assess the seismic vulnerability of façade walls. To fulfil the survey datasheet is always a very demanding and challenging task for cataloguers due to the above-mentioned time, human resources and accessibility limitations. In fact, a large set of situations may prevent the surveyor from covering all the fields that compose the sheets. A relatively common strategy for dealing with the inaccessibility to buildings is to only gather information from the part of the building that is fully visible from the outside—the façades. A relevant number of entries of the National Catalogue of Historical Monuments are indeed in this situation. Nevertheless, this information is still potentially valuable for vulnerability assessment purposes. The methodology developed and calibrated by Ferreira et al. [2017b], and later applied by Aguado et al. [2018], is a variation of the Vulnerability Index Method for an assessment based on the information exclusively from the façade walls. This adaptation is mostly addressed to reveal the relative likelihood of the activation of out-of-plane mechanisms of collapse, which would also be meaningful in post-disaster scenarios. For example, allowing to obtain valuable urban-scale information, such as debris accumulation on the streets. This adaptation was initially developed and validated by Ferreira et al. [2014], based on the analysis of post-seismic data for identifying construction parameters that influence the level of damage suffered by masonry façades. The original work considered ten parameters associated with four classes and cor- 84 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. responding weights, based on the GNDT II approach. The application of this approach in 672 buildings permitted more refined vulnerability functions and curves. The suitability of this method was validated through its comparison against simplified mechanical models. Further calibrations of this method, such as the ones of Ferreira [2015] and Ferreira et al. [2015], led to the current 13-parameter version of the method (Table 4.3). Like the buildings’ approach, the vulnerability index for façade walls is obtained through the weighted sum of the 13 parameters class value Cvi multiplied by their corresponding weight factor Pi, which describes their relative importance Eq. (4.2)—the description of the vulnerability classes of each of the thirteen parameters that compose the method is provided in Appendix A. The vulnerability index is the base for the further estimation of damage grades. Parameters Class Cvi Weight pi Relative weight A B C D Group 1 Façade geometry, openings and interaction 16.7/100 FP1. Geometry of façade 0 5 20 50 0.50 FP2. Maximum slenderness 0 5 20 50 0.50 FP3. Area of openings 0 5 20 50 0.50 FP4. Misalignment of openings 0 5 20 50 0.50 FP5. Interaction between continuous façades 0 5 20 50 0.25 Group 2 Masonry materials and conservation 31.5/100 FP6. Quality of materials 0 5 20 50 2.00 FP7. State of conservation 0 5 20 50 2.00 FP8. Replacement of original flooring system 0 5 20 50 0.25 Group 3 Connection efficiency to other structural elements 33.3/100 FP9. Connection to orthogonal walls 0 5 20 50 2.00 FP10. Connection to horizontal diaphragms 0 5 20 50 0.50 FP11. Impulsive nature of the roofing system 0 5 20 50 2.00 Group 4 Conservation status and other elements 18.5/100 FP12. Elements connected to the façade 0 5 20 50 0.50 FP13. Improving elements 0 5 20 50 -2.00 Table 4.3: Parameters and vulnerability classes, according to Aguado et al. [2018]. I∗ vf = 13 ∑ i=1 cvipi(4.2) 4.5. APPLICATION TO ATLIXCO, PUEBLA. 91 is based on the Complementary Technical Code for Seismic Design of the Building [Gobierno del Distrito Federal, 2017], which is widely used in Mexico as a national standard for constructions. According to this document, a ductility value Q= 1.0should be adopted for unconfined or unreinforced masonry building typologies in Mexico, regardless of whether the masonry is composed of brick or natural stone units. It is pertinent to note that the use of this value is indicative and thus open to discussion, namely in the context of experimental data. However, and given the lack of more specific criteria in the regulations, this ductility value was taken conservatively as 1.0. Moreover, although the 2017 earthquake has damaged several other building typologies, this analysis is targeted explicitly to unreinforced masonry buildings. This decision to address this (and only this) typology is backed by the need to limit the scale variation among the samples as a means for reducing the uncertainly and the range of applicability of the vulnerability index method itself. Monumental and religious buildings were excluded from the analysis. 4.5.2 The building dataset. This section includes a general presentation of the nine buildings—representative of the unreinforced masonry typology—that served as the object of study herein. As mentioned before, these buildings’ seismic vulnerability has been assessed using information that is publicly available on the National Catalogue for Historical Monuments. The analysis has also been complemented with data from Google Maps. The level of damage suffered by the buildings after the 2017 Atlixco earthquake (from now on referred to as observed damage, dk) was initially based on after-event imagery from Google Street View and pictures and videos recorded after the event. Table 4.5 includes a picture of each one of the analysed buildings, before and after the event, as well as its INAH catalogue number and level of observed damage, dk. The selection of the (nine) buildings obeyed two fundamental criteria: (i) the building is included in the National Catalogue for Historical Monuments; and (ii) it has been captured by Google Street View imagery, both before and after the 2017 earthquake. The public datasheets [Instituto Nacional de Antropología e Historia México. Coordinación Nacional de Monumentos Históricos, 2020] permit to have an overview of the buildings in terms of their chronology, distribution, original and present uses and even some historical details (see Table 4.6). Even if the catalogue datasheets are rarely specific about the year of construction, they suggest a century of construction based on documental information (e.g., historic cartography) and/or stylistic attributes. Eight out of the nine selected buildings are from the 18th and 19th centuries. An exception is the sample I-21-00024, which corresponds to the Claustro de la Merced (Cloister of La Merced). This con- 92 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. ID Damage observations Pre-event picture Post-event picture I-21 00087 D2. Opening of some diagonal cracks. Cracking of masonry and localised plaster detachment. I-21 00083 D3. Opening of diagonal cracks. Cracking of masonry and localised plaster detachment. Cracking in parapets. I-21 00086 D4. The building presented severe damages and was demolished soon after the earthquake. I-21 00066 D4. The building partially collapsed. I-21 00024 D5. The building collapsed during the earthquake. I-21 00107 D2. Opening of some diagonal cracks. Cracking of masonry and localised plaster detachment. I-21 00079 D4. The building presented severe damages, including a partial collapse of the façade. I-21 00078 D2. Cracking of masonry and localised plaster detachment. I-21 00121 D4. The building presented severe damages and was demolished soon after the earthquake. Table 4.5: Set of selected buildings, their corresponding Catalogue number and pre and post-event picture. 4.5. APPLICATION TO ATLIXCO, PUEBLA. 93 ID Housing Plan distribution Storeys Century I-21-00087 Housing Corridor around patio 1 19th I-21-00083 Commercial Polygon 1 18th I-21-00086 Housing Corridor around patio 1 18th I-21-00066 Housing Corridor around patio 2 19th I-21-00024 Commercial and housing Corridor around patio 2 17th I-21-00107 Housing Corridor with side patio 1 19th I-21-00079 Commercial and housing Corridor around patio 2 18th I-21-00078 Housing Corridor around patio 2 19th I-21-00121 Housing Corridor around patio 2 18th Table 4.6: Summary of attributes available in the public INAH datasheet. struction is originally from the 17th century but was subsequently abandoned and occupied as collective housing during the 19th century, in the context of the secularisation of the Mexican State. Then, the building had a series of heterogeneous interventions and types of use, including commerce and housing during the 20th and 21st centuries. This antecedent would be relevant for explaining an eventual eccentricity on the analysis for this construction. Except for this building, the rest of the constructions are considered by the INAH as conceived for housing, even if some of them had mixed commercial and housing uses. All the analysed constructions have a maximum of two storeys and are typologically similar in their plan distribution. In fact, seven out of the nine samples correspond to the traditional distribution of rooms aligned to corridors around a central patio. One more sample has a lateral patio, which is sometimes a consequence of a property’s subdivision. Finally, sample I-21-00083 is described in its datasheet as a «polygonal» plan, which may correspond to an irregular orthogonal distribution of rooms with no patio. 4.5.3 Analysis and discussion of the results. Before going into the comparison between the analytical and the observed damages, it is pertinent to provide an overall picture of each building’s vulnerability results. As graphically presented in Figure 4.6, the nine buildings analysed present a relatively heterogeneous distribution of their vulnerability classes; except parameters FP13 and FP12, whose classes are roughly the same for all buildings. For this case study, this is strongly related to typological constants as well as reglementary restrictions on the use of façades (e.g., the prohibition of the use of external equipment or attached elements). In global terms, the sample has a mean seismic vulnerability index value, Ivf ,mean, of 45.12. The minimum value of vulnerability index was obtained for building I-21-00078, 21.11, and the maximum, 89.63, was obtained for building I-21-00024. The associated standard distribution, σIvf , is 20.36. 94 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. Class distribution (%) 0 20 40 60 80 100 Vulnerability Parameter FP1 FP2 FP3 FP4 FP5 FP6 FP7 FP8 FP9 FP10 FP11 FP12 FP13 Figure 4.6: Vulnerability class distribution for the nine buildings evaluated. Moving on to the analysis of the damages, whose values are also given in Table 4.7, it is interesting to see that the analytical, Dk, and the observed damages, dk, exhibit a pretty good correspondence. As can be observed in Table 4.7, the predicted Dkwas consistent with the observed dkfor all the nine buildings assessed. This good fitting does not exclude the need of calibrating the approach. In fact, the use of this method by departing from past calibrations is not intended to demonstrate the equivalence of the Mexican masonry typologies against those explored in the past, namely in Europe. Nevertheless, since the predicted damages are, in general, consistent with the observations, it is plausible to assume that the methodology can be accurately used to assess the seismic vulnerability of the constructions existing in the present case study, even with no further calibration. A relevant aspect of these results is that all the range between D2and D5was covered, which permits us to compare the representativeness for diverse levels of damage. Since buildings with no visible damage were excluded from the sampling, only values equal to or higher than D2are expected. The limitations found while grading some parameters are not neglectable. To verify the robustness and suitability of the results partially based on assumptions, it becomes necessary to assess how sensitive the results are if the assumed values change. In other words, it is necessary to carry a sensitivity analysis to verify if there are significant changes in the vulnerability and damage results. 4.5.4 Sensitivity analysis based on intervals. The assumption of values for the analysis presented the previous section was based on typological similarities among the samples, accepting some informed judgements during the survey. Therefore, it is important to measure the grade of uncertainty that a given imprecise or inexact criterion may have in the overall scoring. Having this in mind, a proposal based on ranges has been carried, offering a set of 4.5. APPLICATION TO ATLIXCO, PUEBLA. 95 FP Parameter I-2100087 I-2100083 I-2100086 I-2100066 I-2100024 I-2100107 I-2100079 I-2100078 I-2100121 1 Geometry of façade. A(0) A(0) A(0) A(0) A(0) A(0) B(5) B(5) C(20) 2Maximum slenderness. A(0) A(0) A(0) B(5) B(5) A(0) B(5) B(5) B(5) 3 Area of openings. A(0) B(5) A(0) C(20) B(5) B(5) B(5) B(5) B(5) 4Misalignement of openings A(0) B(5) D(50) A(0) D(50) A(0) D(50) A(0) B(5) 5Interaction between cont. façades C(20) D(50) D(50) D(50) D(50) D(50) D(50) C(20) D(50) 6 Quality of materials B(5) C(20) D(50) D(50) D(50) B(5) D(50) B(5) D(50) 7 State of conservation B(5) B(5) C(20) C(20) D(50) A(0) B(5) A(0) D(50) 8 Flooring replacement B(5) A(0) A(0) D(50) D(50) A(0) A(0) A(0) D(50) 9Connection to orthogonal walls C(20) C(20) C(20) C(20) D(50) B(5) C(20) B(5) B(5) 10 Connection to horizontal diaphragms C(20) C(20) C(20) C(20) D(50) C(20) C(20) A(0) D(50) 11 Impulsive nature of roofing system B(5) B(5) C(20) C(20) D(50) B(5) C(20) B(5) D(50) 12 Elements connected to the façade A(0) A(0) A(0) A(0) D(50) A(0) A(0) A(0) C(20) 13 Improving elements A(0) A(0) A(0) A(0) A(0) A(0) A(0) A(0) C(20) Ivf Normalised Vulnerability Index 27.59 33.70 54.44 54.44 89.63 22.96 51.11 21.11 51.11 µDDamage grade 2.10 2.55 3.92 3.92 4.85 1.77 3.74 1.65 3.74 Dk Analytical discrete damage D2D3D4D4D5D2D4D2D4 dk Observed discrete damage d2d3d4d4d5d2d4d2d4 Table 4.7: Summary of the vulnerability results. results corresponding to two plausible extreme values for those attributes for which there is any degree of uncertainty. Instead of giving a closed value, this approach assumes the uncertainty associated with parameters FP6, FP8, FP9 and FP10, accepting that each attribute may be associated with an upper and a lower boundary that reasonably contains the actual value for each variable. For some samples and attributes, there are some grades with very low uncertainty (i.e., the lower and upper bounds remain the same). For others, that uncertainty may be more significant. For instance, sample I-21-00066 has a relatively low number of uncertainties compared to the sample I-21-00087 (Table 4.8), which may be reflected in a wider range for the vulnerability index values. For six samples, the upper and lower estimated mean damage grades have the same value and have no differences regarding the observed damage and the estimated damage based on assumptions—see 96 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. FP Parameter I-2100087 I-2100083 I-2100086 I-2100066 I-2100024 I-2100107 I-2100079 I-2100078 I-2100121 6 Assumpted value B(5) C(20) D(50) D(50) D(50) B(5) D(50) B(5) D(50) Lower bound B(5) D(50) D(50) D(50) D(50) C(20) D(50) C(20) D(50) Upper bound A(0) C(20) D(50) D(50) A(0) A(0) C(20) A(0) D(50) Interval (no. of grades) 221143231 8 Assumpted value B(5) A(0) A(0) D(50) D(50) A(0) A(0) A(0) D(50) Lower bound D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) Upper bound A(0) A(0) C(20) D(50) B(5) B(5) C(20) A(0) C(20) Interval (no. of grades) 442133242 9 Assumpted value C(20) C(20) C(20) C(20) D(50) B(5) C(20) B(5) C(20) Lower bound D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) Upper bound A(0) B(5) B(5) C(20) B(5) B(5) D(50) B(5) C(20) Interval (no. of grades) 433233132 10 Assumpted value C(20) C(20) C(20) C(20) D(50) C(20) C(20) A(0) C(20) Lower bound D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) D(50) Upper bound A(0) B(5) B(5) C(20) C(20) C(20) C(20) B(5) C(20) Interval (no. of grades) 433222232 Total number of grades considered in the ranges of uncertainty 14 12 9 6 12 11 7 13 7 Table 4.8: Summary of the vulnerability results. Table 4.9; Figure 4.7. Moreover, the lower and upper bounds obtained never exceed one grade, and the tendencies are still consistent with the observed damages, always containing the observed and estimated damage value. It is worth noting that, since the damage levels are actually a discrete representation for intervals (refer to Table 4.5), they may not appropriately adjust to the cases where the damage grades are close to the superior or inferior thresholds. Then, in addition to the discrete damage grades, it is also useful to compare the mean damage grade, µD. Such analysis permits to better appreciate the amplitude of the range offered in the results, as well. In Fig. 4.8, it is possible to verify that there is consistency between the amplitude of the uncertainty, presented in Table 4.8, and the distance between the upper and lower values for the damage grade. It is relevant to observe that the suitability of using the discrete damage grades would be limited when working with intervals for grading the attributes instead of closed unique values, namely for those values that are close to the thresholds. The present approach also makes it possible to derive vulnerability curves based on «regions» (Fig. 4.9). 4.5. APPLICATION TO ATLIXCO, PUEBLA. 97 Figure 4.7: Ranges of values for analytical damage Dkand its comparison with the observed damage dk. I-2100087 I-2100083 I-2100086 I-2100066 I-2100024 I-2100107 I-2100079 I-2100078 I-2100121 Ivf Based on assumptions 27.59 33.70 54.44 54.44 89.63 22.96 51.11 21.11 51.11 lower bound 31.48 46.67 58.52 56.67 89.63 31.48 55.19 31.11 53.33 upper bound 18.52 28.15 49.63 54.44 57.59 21.67 51.85 20.00 50.00 µD Based on assumptions 2.10 2.55 3.92 3.92 4.85 1.77 3.74 1.65 3.74 lower bound 2.39 3.47 4.11 4.03 4.85 2.39 3.95 2.36 3.86 upper bound 1.48 2.14 3.65 3.92 4.07 1.68 3.78 1.57 3.67 Dk Based on assumptions D2D3D4D4D5D2D4D2D4 lower bound D2D3D5D5D5D2D4D2D4 upper bound D2D2D4D4D5D2D4D2D4 dk Observed after 2017 Earthquakes d2d3d4d4d5d2d4d2d4 Table 4.9: Summary of the vulnerability results. This kind of representation permits us to easily contextualise the tendency and amplitude of the variability for specific seismic intensity values. The overlap of the damage grades and the regions between the vulnerability functions are another expression of the consistency of both indicators. Once again, the regions may be more meaningful for understanding a field of vulnerability that intrinsically accepts uncertainties associated with the adequate grading of some attributes. This approach may be found valuable not only when it is impossible to have a comprehensive survey but also when the constructions’ attributes 98 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. 2.10 2.55 3.92 3.92 4.85 1.77 3.74 1.65 3.74 1.48 2.14 3.65 3.92 4.07 1.68 3.78 1.57 3.67 2.39 3.47 4.11 4.03 4.85 2.39 3.95 2.36 3.86 Discrete damage value Lower and upper bound values Damage grade, µD 0.0 1.0 2.0 3.0 4.0 5.0 Catalogue Number I-21-00087 I-21-00083 I-21-00086 I-21-00066 I-21-00024 I-21-00107 I-21-00079 I-21-00078 I-21-00121 Figure 4.8: Plot of the µDvalues according to assumptions (discrete value) and the corresponding ranges. Figure 4.9: Vulnerability curves according to the boundaries of each one of the samples. are close to the threshold between two contiguous grades. The use of these composed curves would also permit the development of subsequent works, such as the analysis of the nominal life of historical constructions (based on seismic return periods) and to analyse the impact, in terms of vulnerability, of certain interventions that would affect the Vulnerability Index Method attributes and qualification. 4.6. FINAL REMARKS. 99 4.5.5 Limitations of the approach. Even if the results suggest the validity of using this methodology, a series of limitations should not be forsaken. Firstly, because of its limited dimension, the sample used herein does not include all the possible attributes and damage levels. For this reason, it is not reasonable to see these results as definitive proof of reliability and accuracy. Another limitation is related to the application of the methodology itself, which, because it was performed in an indirect manner, encompasses some grade of uncertainty resulting from the assumptions made. In order to measure the impact of these two limitations, a simple validation procedure has been put into practice through: (1) comparing the information obtained following the procedure detailed in the previous sections with further sources of information, namely with a series of building catalogues developed by Martínez Bret [2018], Ruíz Herrera [2018], Tototzintle Huitle [2018] and Flores Rodríguez [2018]. These documents include the survey of most of the buildings of Atlixco in an early stage after the earthquake; and (2) an on-site survey of the buildings. Such a procedure allowed to confirm the suitability of the methodology presented in this work, since no inconsistencies were found between the information obtained from the National Catalogue for Historical Monuments and that obtained with more detailed, but not always available, sources of information. Another relevant and expectable limitation of the procedure results from the fact that Atlixco’s buildings do not belong to the same typology of masonry construction to which the vulnerability index method was initially designed for. Hence, some attributes do not accurately represent similar typological aspects, such as the geometry of the ceilings, for example. According to the results, this issue may be a minor limitation, but it definitively encourages enhancing and adapting the method in further research. 4.6 Final remarks. The existing Mexican National Catalogue for Historical Monuments represents a valuable source of information regarding the historical buildings considered as monuments by definition of the law. The nature and type of some of the catalogue fields are compatible with some recent approaches for assessing the seismic vulnerability of constructions, which is relevant in the context of a country with frequent seismic activity. The Vulnerability Index Method allows assessing the seismic vulnerability of a building on the basis of a set of parameters, which are weighted according to their role for the global seismic performance of the buildings. Because the method is based on a relatively fast and simple set of observations, it seems pertinent to analyse the possibility of using the vast amount of data in the current version of the National Catalogue of Monuments to feed it. The present chapter is specifically dedicated to that analysis. 100 CHAPTER 4. EXISTING CATALOGUES FOR SEISMIC VULN. ASSESSMENT OF HIST. BUILDINGS. The comparison between the fields that compose the National Catalogue data datasheet and the ones of the Vulnerability Index Method proves the feasibility of integrating the vulnerability-oriented fields in the existing survey datasheet. Furthermore, previous works found in the literature demonstrate the feasibility of adapting the Vulnerability Index Method and its variation for façades to provide a coherent workflow that involves Mexican normative related to seismic characterisation, typological properties of materials and structures, etc. The existing adaptations found in the literature also suggest the robustness of the general theoretical base of the approach and allows for the identification of the parameters to be calibrated. The information related to the damages and losses during the 2017 earthquakes would be a valuable source of information for calibrating the vulnerability curves to some Mexican contexts and typologies. The analysis applied to Atlixco, Puebla, demonstrated an encouraging scenario of suitability. The fitting of the observed and estimated damages does not necessarily imply that the methodology is ready to be extensively used with no changes. However, the results suggest that the existing approach could be adapted with minor calibrations. A more comprehensive sample would permit to carry sensitivity analysis to identify which parameters are more critical for explaining the behaviour of a determined typology of constructions. Besides, the analysis in a broader sample would permit defining the first insights for adapting the method to the most frequent features of Mexican building typologies; an adaptation of the method would imply not only a numerical calibration but also the adjustment of the descriptors to match the construction traditions of the country better. The logic of grading and classifying the building attributes are compatible with the idea of carrying out a typological classification based on specific features. Furthermore, it becomes necessary to contextualise the mechanical properties of the materials and structures for obtaining more accurate assessments. An inherent condition for analysing vast samples is the capacity of performing largescale data acquisition and management campaigns. The present experience revealed that the complementary and interdisciplinary use of diverse sources –such as the information contained in the National Catalogue for Historical Monuments, different scholar documents and on-site surveying—permit to get comprehensive descriptions for seismic vulnerability analysis purposes. A suitable workflow of acquisition, storage and management would favour the implementation of this approach. The major difficulties and limitations underlying the characterisation of some attributes were overcome by the adoption of a range-based criterion, which permitted us to obtain reasonably conservative results. Even so, and despite the results presented herein are promising regarding the suitability of this methodology to support large-scale seismic vulnerability analysis, further investigation on the limitations and accuracy of the approach should be conducted resorting to larger datasets. 5.2. SEISMIC RISK ASSESSMENT. 107 Despite the popularity of the classical definition of risk as a measure of the probability and severity of adverse effects or the triad Risk = Threat ×Vulnerability ×Consequence, Haimes [2009] identify risk as a vector with the units of the consequences. It is given as a function of time, the probability of the threat (initiating event), the probability of the consequences (given the threat), the state of the system (e.g., its vulnerability) and the resulting consequences. For example, in the context of seismic risk, the risk would be able to be understood as a vector in the units of damage or losses. Risk definition always needs the identification of the vulnerability of the systems and elements exposed to a given threat. Since no successful risk evaluation, mitigation or control can be carried out without the analysis and characterisation of the vulnerability, suitable approaches for framing the specific vulnerability of elements towards the studied phenomena must be designed and put in place. Just as for risk, the concept of vulnerability needs to be clearly defined in order to contextualise it in a determined analysis. 5.2.2 Definition of vulnerability. The concept of vulnerability has a broad utilisation throughout multiple and variate disciplines, namely to refer to a certain capacity for coping with change. The conceptual analysis of Brooks [2003] express that, in general, vulnerability is measured through outcome indicators, i.e., expectable losses and/or damages. This analysis also gathers some definitions of vulnerability, emphasizing that of the International Panel on Climate Change: «The degree to which a system is susceptible to, or unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate variation to which a system is exposed, its sensitivity and adaptive capacity». In the same line, Luca and Verderame [2016] defines seismic vulnerability as «… the degree of loss to a given element at risk (e.g., buildings) resulting from the occurrence of an earthquake event». The definition offered by Gavarini [2001] is even more specific since it is framed on the structural performance of constructions in historical centres: «…the seismic vulnerability of a building is a quantity associated with its ‘weakness’ in front of earthquakes of a given intensity so that the value of this quantity and the knowledge of seismic hazard allow to evaluate the expected damages from future earthquakes». This causal relation is highlighted by Vicente et al. [2005]: «Seismic vulnerability is an intrinsic property of the building structure, a characteristic of its own behaviour to seismic action described by a cause-effect relationship, in which the earthquake is the cause, and the effect is the damage suffered». 108 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. 5.2.3 Vulnerability in the context of urban scale: challenges and previous experiences. The measurement of the ability of the structures to cope with the demands imposed by earthquakes has to overcome, however, numerous challenges, such as the existence of great volumes of constructions on historical centres, the variety and complexity of structures (namely remarkable dissimilarities in materials and geometry), and the relations that exist between the constructions. In the most favourable but rare cases, common characteristics in a buildings sample are limited to a certain set of materials and relatively homogeneous structural systems [Oliveira, 2003]. According to Cocco et al. [2019], a feasible approach is to analyse large stocks of buildings in order to include the structural parameters that characterise the response of the buildings and involve the parameters’ variability. This approach is limited by the effort required to comprehensively investigate and model a large number of samples, which encourages the use of simplified procedures. The universe of historical buildings is hardly generalisable by any common characteristic (excepting, maybe, the antiquity), but there is a series of major groups that permit to group relevant sets of constructions in determined contexts. When analysing sets of historical buildings in the context of urban centres, it is possible to find some typological similarities that frame most constructions with a certain grade of variability, such as repetitive patterns for structural systems, geometric constants, etc. It is expectable to find a certain typological consistency in historical centres since architecture is namely a consequence of the local availability of resources (workforce, materials, technology), environmental conditions and cultural determinants. For the above-exposed reasons, the typological approach is very helpful for taking advantage of the similarities among a certain set of constructions (e.g., by analysing similarities that also determine the structural response towards an earthquake) while the typological deviations of each sample permit to individualise and characterise it. In other words, typological approaches permit framing some determinants in the context of the seismic performance of constructions by maintaining the individual-scale analysis of the buildings. Even if the characterisation’s uncertainty is still a challenge when performing typological analysis, an adequate definition of the fields to be characterised permits controlling and contextualising the uncertainty. Some sources of typological variety are listed by Valluzzi [2007] and include: type of masonry (e.g., regular brick or rubble stone), characterization of the properties of the materials (units and mortar, masonry as composite), knowledge of their mutual aggregation (in both thickness and façade), and the efficiency of connections among structural components (walls, floors, roof). Most of these characteristics 5.3. SEISMIC VULNERABILITY ASSESSMENT BASED ON PARAMETERS FOR THE MEXICAN CONTEXT. 109 can be parametrised (i.e., to be described based on parameters such as intervals or thresholds) and classified. Furthermore, it is possible to identify sets of characteristics that are determinant for explaining or conditioning the performance or behaviour of a building in fields such as habitability, energetical efficiency or, for instance, seismic behaviour. Parameter-based seismic vulnerability methods have been widely used worldwide, particularly in countries and regions where seismic activity is more significant or recurring, such as Italy, Greece or Nepal. The Gorkha earthquake (2015), one of the most destructive earthquakes of the recent past, is an outstanding example of a seismic event that has been intensively studied to consolidate and create new knowledge on the seismic vulnerability of various buildings typologies; often resorting to parameter-based vulnerability assessment methods. For example, Gautam et al. [2020] analysed more than 3000 schools by defining a series of structural and material parameters that were subsequently scored to obtain vulnerability classes. This work shows the suitability of typological classifications for designing robust descriptions beyond some typical taxonomical descriptions purely based on materials. 5.3 Seismic vulnerability assessment based on parameters for the Mexican context. The identification and parametrisation of meaningful typological similarities among constructions are indeed based on many of the simplified approaches and methodologies that have been proposed over recent years to assess the seismic vulnerability of existing structures. Some simplified parameter-based methods have presented models for establishing vulnerability functions among specific typologies of constructions, such as the proposal of the National Centre for Prevention of Disasters (CENAPRED) and the so-called Vulnerability Index Method. 5.3.1 Qualitative assessment of the vulnerability of housing units towards seismic actions (CENAPRED). The Basic Guide for the Elaboration of Statal and Municipal Threats and Risk Atlas [CENAPRED, 2006] offers a strategy for assessing the seismic vulnerability of housing units. However, this method is based on five levels of seismic vulnerability only, which constitutes a clear limitation – making it hardly usable as a tool to identify and prioritise intervention needs, for example. In Chapter 1.4, «Simplified criteria for qualitatively assessing the vulnerability of housing units towards seismic actions», this document includes 110 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. a simplified methodology based on a set of concepts related to the materiality of the constructions, the grade of social vulnerability, and the macroseismic region in which a determined municipality is located Values. The criteria behind these concepts are summarised in Table 5.4. However, the housing characterisation does not distinguish any details related to the construction system and/or the properties that may condition the seismic performance of the building. The word «masonry» is used in a generic way for any kind of stones, ceramic or cement blocks; plus, there is a poor specification for roofing systems. This methodology discusses (but does not involve) the suitability of a more comprehensive typological classification of the constructions. Another relevant limitation can be found in the microseismical zones (Figure 5.1). Even if the spectral accelerations of different soil types are considered, this approach does not permit obtaining intensity-based functions for damage, such as damage curves. This limitation is meaningful for developing vulnerability assessments based on different periods of return, restricting some approaches, such as the analysis that considers the constructions’ nominal life. Concept Criteria and grading Vulnerability Index IRF =Ivf (0.8 + IM 25 ) *The gap between 0.5 and 0.6 seems to be a typographic error of the Guide. The result of IRF relates to five so-called “risk levels”: ≤0.0≤IRF <0.2– Very low risk ≤0.2≤IRF <0.4– Low risk ≤0.4≤IRF <0.5– Medium risk* ≤0.6≤IRF <0.8– High risk* ≤0.8≤IRF ≤1.0 IM, grade of social vulnerability towards disasters, from the municipal classification of the National Institute for Statistics, Geography and Informatics (INEGI). IM= 1 for very low IM= 2 for low IM= 3 for medium IM= 4 for high IM= 5 for very high Physical vulnerability Ivf =ViPi VpPm. Viis the grade of the housing unit according to their materials Vpis the housing with the worst performance based on the criteria of Vi Piis the level of seismic exposure in the region of study Pmis the maximum possible level of exposure Vi, According to the materials of the structure Vi= 1 for masonry walls with rigid roofing system Vi= 2 for masonry walls with flexible roofing system Vi= 3 for adobe walls with rigid roofing system Vi= 4 for adobe walls with flexible roofing system Vi= 5 for walls of weak materials with flexible roofing system VpVpis the maximum possible value for exposure. Therefore, Vp= 5 Table 5.4: Components for the qualitative assessment of the vulnerability of housing units towards seismic actions. 5.3. SEISMIC VULNERABILITY ASSESSMENT BASED ON PARAMETERS FOR THE MEXICAN CONTEXT. 111 Figure 5.1: Seismic zones according to the Centro Nacional de Prevención de Desastres [Centro Nacional de Prevención de Desastres, 2006]. Even though the enhancement of the CENAPRED approach is beyond the scope of the present work, it is pertinent to recognise some positive experiences for enriching local approaches by using complementary tools. For example, Azizi-Bondarabadi et al. [2016] depart from an Iranian local parameter-based vulnerability assessment framework, discussing its capabilities and concluding its admissibility. The authors propose the addition of a complementary approach, the GNDT II method, for overcoming some of the limitations found in the original approach, namely those related to the absence of vulnerability curves, then proposing new damage factors specifically calibrated for the study area (associating local damage/cost ratios) and finally coming out with a set of fragility curves to three structural typologies identified among the school buildings. This work demonstrates the feasibility of enriching the Iranian-method index by linking it to a more complex method, which may be adopted in a similar way in the Mexican context. 112 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. 5.3.2 The Vulnerability Index Method. The physical damage and loss after a seismic event constitute a very valuable source of information for understanding the behaviour of the structures. This has improved the knowledge on the traditional building technology, antiseismic empirical or semi-empirical design techniques, the success and failure of building interventions and retrofitting techniques and more recently as valuable data for scientific approaches. The analysis of wide sets of data has allowed correlating meaningful relations between the intensity of earthquakes, some specific characteristics of constructions and the damages they present after a seismic event. The nature of this regressive analysis has allowed identifying several geometrical and material properties that are critical of the seismic vulnerability of traditional masonry building stock, such as for the activation of out-of-plane mechanisms. The Italian National Group for the Defence from Earthquakes (Gruppo Nazionale per la Difesa dai Terremoti – GNDT) analysed large sets of data related to the characteristics of constructions that were damaged or lost as the consequence of earthquakes. These analyses allowed identifying essential features that were lately consolidated in a straightforward datasheet [Benedetti and Petrini, 1984, Gruppo Nazionale per la Difesa dai Terremoti—GNDT, 1994]. This characterisation is based on a series of typological ranges that are common to several cities in Italy but also relatable to other countries across Europe due to the similar traditional construction technology. Despite the inherent difficulties for a generalised mechanical characterisation of masonry, some general assumptions can be made, such as assuming brittle response in tension, frictional response in shear and anisotropy [Roca et al., 2010]. Hence, it becomes possible to generalise some mechanisms and configurations as the basis for parameter-driven assessments. The general approach of this method is to consider a set of geometrical and material parameters. Each parameter is then able to be associated with the most suitable of four possible classes (A, B, C and D) that encloses the variability of the parameter assessed. Each parameter is associated with a weight pi, which represents how crucial/important the characteristic is ruling the seismic performance of the building. For each parameter, the weight is multiplied by the value of the class cvi, and the weighted sum of all parameters lead to a global vulnerability index [Calvi et al., 2006], as shown in Equation (6.1). The original proposal considered eleven parameters, but successive adaptations increased that number up to fourteen parameters divided into four groups: structural building system, irregularities and interaction, flooring and roofing systems and conservation status and non-structural elements (see Table5.5). I∗ vf = 14 ∑ i=1 cvipi(5.1) 5.3. SEISMIC VULNERABILITY ASSESSMENT BASED ON PARAMETERS FOR THE MEXICAN CONTEXT. 113 Parameters Class Cvi Weight Relative A B C D piweight Group 1 Structural building system 50/100 BP1. Type of the resisting system 0 5 20 50 2.50 BP2. Quality of the resisting system 0 5 20 50 2.50 BP3. Conventional strength 0 5 20 50 1.00 BP4. Maximum distance between the walls 0 5 20 50 0.50 BP5. Number of floors 0 5 20 50 0.50 BP6. Location and soil condition 0 5 20 50 0.50 Group 2 Irregularities and interaction 20/100 BP7. Aggregate position and interaction 0 5 20 50 1.50 BP8. Plan configuration 0 5 20 50 0.50 BP9. Height regularity 0 5 20 50 0.50 BP10. Wall façade openings and alignment 0 5 20 50 0.50 Group 3 Floor slabs and roofs 18/100 BP11. Horizontal diaphragms 0 5 20 50 0.75 BP12. Roofing system 0 5 20 50 2.00 Group 4 Conservation status and other elements 12/100 BP13. Fragilities and conservation status 0 5 20 50 1.00 BP14. Non-structural elements 0 5 20 50 0.75 Table 5.5: Vulnerability Index parameters, according to the calibration of Ferreira et al. [2017c]. As schematised in Figure 5.2, in which the circumference has been proportionally divided according to the weight of each parameter, certain aspects of the construction are significantly more determinant than the rest, such as parameters BP1 and BP2. The concentric circles are also proportional to the numerical value according to the class. It is convenient to keep in mind that the progression between classes can be more significant between classes C and D. If the area of the entire circle is 650 units and the regions are shaded according to the grade of each parameter, then it graphically represents the true value of I∗ vf as a proportional shaded surface. This method has been applied to numerous case studies, such as the ones of Coimbra [Vicente, 2008] [Aguado et al., 2018], Seixal [Ferreira et al., 2013], Horta [Ferreira et al., 2017c] and Leiria [Blyth et al., 2020], Portugal; Annaba, Algeria [Athmani et al., 2015]; Timisoara, Romania [Mosoarca et al., 2019]; Osijeck, Croatia [Hadzima-Nyarko et al., 2016] etc., demonstrating the suitability of this approach to be adapted based on adjustments of the weights and damage curves according to the typological characteristics [Giuliani, 2020]. The concerns related to the constraints for accomplishing an extensive survey (e.g., not having complete access to the building site or drawings) encouraged the design of a complementary façade scale evaluation approach. This method is mostly focused on the potential assessment of out-ofplane mechanisms (by assessing building features that make them more prone to such) and was widely explored initially [Vicente, 2008], that proposed an initial set of 10 parameters that have been expanded 114 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. in the following years [Ferreira et al., 2014, 2017a, Aguado et al., 2018]. The logic and procedures behind this method are the same that those applied for the building scale method, considering a total of 13 parameters (see Table 5.6). Figure 5.2: Schematic representation of the parameters and their weight. Parameters Class Cvi Weight Relative A B C D piweight Group 1 Façade geometry, openings and interaction 16.7/100 FP1. Geometry of façade 0 5 20 50 0.50 FP2. Maximum slenderness 0 5 20 50 0.50 FP3. Area of openings 0 5 20 50 0.50 FP4. Misalignment of openings 0 5 20 50 0.50 FP5. Interaction between continuous façades 0 5 20 50 0.25 Group 2 Masonry materials and conservation 31.5/100 FP6. Quality of materials 0 5 20 50 2.00 FP7. State of conservation 0 5 20 50 2.00 FP8. Replacement of original flooring system 0 5 20 50 0.25 Group 3 Connection efficiency to other structural elements 33.3/100 FP9. Connection to orthogonal walls 0 5 20 50 2.00 FP10. Connection to horizontal diaphragms 0 5 20 50 0.50 FP11. Impulsive nature of the roofing system 0 5 20 50 2.00 Group 4 Conservation status and other elements 18.5/100 FP12. Elements connected to the façade 0 5 20 50 0.50 FP13. Improving elements 0 5 20 50 -2.00 Table 5.6: Parameters and vulnerability classes, according to Aguado et al. Aguado et al. [2018]. 5.3. SEISMIC VULNERABILITY ASSESSMENT BASED ON PARAMETERS FOR THE MEXICAN CONTEXT. 115 Although the identification of the characteristics that critically determine the seismic vulnerability is very dependent on typological similarities, it is possible to admit a certain level of similarity among historical constructions in determined contexts. For example, Mexican housing constructions in historical centres are typologically related to the use of local materials and traditional construction techniques and solutions that are also common in European countries. The Viceroyalty of New Spain (1521-1821) represents a period in which a permanent exchange of architectonic and urban tendencies moulded the physiognomy of thevhistorical cities in Mexico. These constructions have numerous similarities with their chronological counterparts in the Iberian Peninsula. This cultural bridge is a strong indication that this approach can be suitable to assess the historical constructions in the Mexican urban context. A limitation, however, is that the criteria and analysis for the parameters are strongly focused on typological constants of the European constructions. This situation does not hinder the application to the Mexican building stock. However, the description and grading of all the parameters are needed in order to facilitate an objective framework for its application. The works of Salazar and Ferreira [2020] explored the assessment of buildings in the historical neighbourhood of La Merced in Mexico City by using the Vulnerability Index Method approach. This experience established a set of typologies (four geometrical types and nine material types) for generalising a set of 166 constructions (however, several limitations are implicitly present when assuming generalisations throughout a sample). This typological approach was potentially limited, for example, when there are morphological modifications due to building interventions (interior space changes, refurbishments, etc.). In this context, the analysis of individual buildings instead of groups is still a gap in literature addressed to Mexican assets. It is worth highlighting that this approach has also been used as a conceptual basis for developing more specific methods, such as the Seismic Vulnerability Index for Vernacular Architecture (SVIVA), developed by Ortega et al. [2019]. The SVIVA formulation includes a definition of the parameter’s vulnerability classes according to numerical reference models. The relative weight of each parameter was obtained using multiple linear regression and expert opinion. Although this approach was initially targeted at Portuguese vernacular architecture, the authors admit its applicability to similar structures outside the Portuguese context. In fact, it is plausible to assume that this approach would also be suitable to analyse the present case of study. The main reason for adopting here the Vulnerability Index Method proposed by Ferreira et al. [2017c] lies in the fact that the vulnerability parameters are more explicitly unfolded in this method. Despite that, the data generated in this work can likely be used to generate vulnerability results with the SVIVA approach, providing, therefore, a good base for future works of validation or numerical calibration based on the effects of the 2017 earthquakes in the Mexican constructions. 116 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. Data availability for single buildings is a very relevant challenge for applying the Vulnerability Index Method to large sets of constructions. However, the Mexican context provides an advantage due to the existence of the National Catalogue of Historical Monuments. Developed by the National Institute for Anthropology and History (Instituto Nacional de Antropología e Historia, INAH), this catalogue is a valuable resource that groups more than 100,000 datasheets that individually describe historical monuments in Mexican territory. Since Mexican law determines that any construction built before the year 1900 must be considered as a historical monument [Cámara de Diputados, 2012], this catalogue is wide. The experiences of Ramírez Eudave and Ferreira [2021d] provided a first insight on the use of this catalogue for feeding the Vulnerability Index Approach, supporting its suitability by comparing the expected damages and the effects of the 19th of September, 2017 earthquakes in Atlixco, Puebla, that had a Modified Mercalli Intensity of 7.5 in this city [United States Geological Survey USGS, 2017]. This city is a representative case of a Mexican historical city centre, recognised as a Zone of Historical Monuments (Zona de Monumentos Históricos) by the INAH. This denomination is given due to a relevant concentration of historical monuments and provides a special framework for its safeguarding. Even if the proof-of-concept study was based on a limited set of nine constructions, it strongly suggests the suitability of the methodology for assessing Mexican constructions. Although the catalogue ideally covers almost all the parameters to be used by the Vulnerability Index Method, this experience also found a strong limitation since most of the datasheets is incomplete. A more extensive and comprehensive experience was possible in the scope of a series of surveys performed in the previous months before the 2017 earthquakes. 5.4 Materials and methods: the dataset. A comprehensive catalogue for the historical monuments of the city of Atlixco, Puebla, following the model of the INAH datasheet, was used by Martínez Bret [2018], Ruíz Herrera [2018], Tototzintle Huitle [2018] and Flores Rodríguez [2018] and within the scope of their work an improved datasheet was made (see Figure 5.3) that contains additional and valuable information fields, such as general measurements and façade drawings (see Table 5.7). These works were mostly developed in the months that preceded the earthquakes of September of 2017, giving a valuable testimony of the general conservation state of the constructions in a pre-damage stage. Furthermore, some of the datasheets also include descriptions of the damages suffered because of the earthquakes. The use of this information allowed to significatively enlarges the first insight on the use of the Mexican Catalogue by applying the Vulnerability Index Method approach to a total of 83 constructions. 5.5. RESULTS AND DISCUSSION. 123 Figure 5.5: Summary of results (1/5). The interval between µ− Dand µ+ Dis represented as the white box. Figure 5.6: Summary of results (2/5). The interval between µ− Dand µ+ Dis represented as the white box. 124 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. Figure 5.7: Summary of results (3/5). The interval between µ− Dand µ+ Dis represented as the white box. Figure 5.8: Summary of results (4/5). The interval between µ− Dand µ+ Dis represented as the white box. 5.5. RESULTS AND DISCUSSION. 125 Figure 5.9: Summary of results (5/5). The interval between µ− Dand µ+ Dis represented as the white box. For both assessments, samples SECI-24 and SECIII-16 presented the highest and lowest mean damage grade value, respectively. These figures also present the µ− Dand µ+ Dthat respectively correspond to the level of damage observed in the building, according to the criterium explained in Table 8. This representation illustrates the overlap between the intervals of the mean damage grade obtained from the vulnerability index of the building and the actual level of damage suffered by the building. When the ranges of the mean damage grades and the real discrete damages are plotted together, it is evident that a vast majority of values present an overlap. This is meaningful because it reveals a level of consistency towards the whole assessment. It is convenient to consider that the proportion of this overlap depends on numerous circumstances, such as the amplitude of the range for the discrete Dkvalues and the accumulation of uncertainties during the assessment process. Nevertheless, the tendencies exhibit a satisfactory pattern of proportionality and coherence. It is worth noting that there can be a relevant distortion when dealing with the boundaries of the discrete damage values, given that the same semantical description may cover a wide variety of damages and patterns. This has been recognised as a potential limitation for using and calibrating this method, opening the possibility of developing more objective strategies for assessing damages in post-seismic scenarios. 126 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. Furthermore, it is important to mention that these results depend on the definition of the ductility factor, Q1, assumed equal to 1. Although this is the value suggested by the Mexican code for unreinforced masonry, the adoption of a ductility factor value based on experimental evidence is a most desirable improvement for any application of the method to any case. As shown in Figure 5.10, there can be a relevant variation in the mean damage grades for some building cases, namely the ones with the lowest and highest mean damage grades. Adopting different values for the ductility factor, for example, a less conservative value, i.e., a value higher than the Q=1.0 considered here, will reduce the global dispersion of the mean damage grades of the set of buildings assessed. Figure 5.10: Variability of the Mean damage grades for the 83 constructions by using different values of ductility Q(The buildings were arranged from the lowest up to the highest mean damage grade value). Although the mean damage grade values are important references for single-building qualitative assessment, it is possible to offer a contextualisation for an entire group of buildings, typologically similar in the same affected site, by means of a cartesian uncertainty/mean damage grade map. This representation facilitates the correlation between the distribution of the uncertainty level and the level of damage throughout a series of constructions, easily identifying how reliable the assessment of the vulnerability class is. Since this map offers a contextualised comparative, it becomes possible to hierarchise further surveys (e.g., on those buildings that apparently present a high mean damage grade value combined with rela- 5.5. RESULTS AND DISCUSSION. 127 tively high uncertainty) and even immediate remedial actions (e.g., on those constructions that combine a relatively low uncertainty and a high mean damage grade. In order to do so, the uncertainty has been graded by imposing numerical values associated with the levels: QC0 = 0;QC1 = 0.33;QC2 = 0.67;QC3 = 1.00. The uncertainty index UI is given by the sum of the value of the QC corresponding to each parameter multiplied by the weight that the parameter has in the seismic vulnerability method. Hence, uncertainty is proportional to the impact of the parameter. Finally, the result is divided by 15 (the sum of all (pi) partial weights) to have a normalised value from 0 to 1, in which 1 represents an absolute uncertainty of all parameters, as per Eq. (5.7) UI = 14 ∑ i=1 QCi×pi 15 {0,1}(5.7) This phenomenon corresponds to a pattern in which the built environment is still rich in vernacular expressions and traditional construction techniques, especially in historical cities. It is relevant to point out that the investment for attending non-residential historical monuments (namely churches) would exceed 600 million USD [Lindero, 2017], concentrated in states with a high level of marginalisation. In fact, more than 50% of the historical buildings damaged by the 2017 Earthquakes are located in the states of Puebla and Oaxaca [García et al., 2018]. The upper and lower limit values for this index were a minimum UImin = 0.07 for sample SECIII16, a maximum UImax = 0.54 for sample SECII-29 respectively and a mean of UI = 0.37 with a standard deviation σ= 0.08. This approach would facilitate the identification of the buildings assessed that combine a low quality of information but seem to have an elevated vulnerability index, becoming cases for further detailed appraisal and assessments. On the other hand, a combination of a low uncertainty index and a low mean damage grade is desirable. The thresholds for this representation can be decided, for example, according to ranges of acceptability in terms of repair costs, establishing criteria for risk acceptability. Figure 5.11 presents an example based on the universe herein discussed. The samples close to the bottom-left corner combine relatively low mean damage grades and low uncertainty, which is the safest condition in the context of the urban assessment. In contrast, samples in the dominion of the top-left corner combine relatively high mean damage grades with a relatively low degree of uncertainty. This means that these buildings are more likely to present a poor seismic performance – due to their high seismic vulnerability – and, consequently, deserve the most immediate attention. Samples in the top-right dominion present a high mean damage grade and relatively high uncertainty. 128 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. Figure 5.11: Distribution of the sample according to the calculated mean damage grade against the uncertainty index. This may indicate that the damage grade is amplified by the lack of reliable information. These buildings deserve particular attention as well, namely by carrying out urgent surveying actions devoted to double-checking their vulnerability, potentially resorting to more detailed assessment methodologies. Finally, the points closer to the bottom right corner of the graph refer to low vulnerable buildings. Although relatively high levels of uncertainty are associated with these buildings, their reassessment is not a priority, given that they are unlikely to suffer significant levels of damage. As is easy to understand for this analysis, this type of representation helps prioritise interventions, allowing for more efficient and rational use of the available resources. Table 5.10 summarises the distribution of damage grades throughout the studied universe. The numerical values associated with the uncertainty are used for calculating mean values for the 83 constructions assessed. These mean values UIpare then multiplied by the relative weight (pi) of each parameter for assessing the relative impact that the uncertainty had on the vulnerability assessments carried out in this work. This exercise points out that the most significant distortions due to epistemic uncertainties are on parameters BP1, BP12 and BP2. A detailed survey campaign specifically devoted to minimising the uncertainty on these parameters would be reflected in a substantial reduction of the overall uncertainty throughout the universe of study. Both, the Uncertainty Index and the Conservative Mean Damage Grade, constitutes feasible strategy to deal with uncertainty when having a partial availability or lack of data. The example of the National 5.5. RESULTS AND DISCUSSION. 129 Parameters (BP) Class Cvi piUIpUI*pi A B C D Group 1. Structural building system 1. Type of resisting system 3.61% 65.06% 28.92% 2.41% 2.50 0.43 1.07 2. Quality of the resisting system 10.84% 6.02% 74.70% 8.43% 2.50 0.36 0.90 3. Conventional strength 3.61% 31.33% 61.45% 3.61% 1.00 0.32 0.32 4. Maximum distance between walls 28.92% 44.58% 21.69% 4.82% 0.50 0.30 0.15 5. Number of floors 40.96% 57.83% 0.00% 1.20% 0.50 0.30 0.15 6. Location and soil conditions 0.00% 100.00% 0.00% 0.00% 0.50 0.87 0.44 Group 2. Irregularities and interaction 7. Aggregate position and interaction 26.51% 32.53% 10.84% 30.12% 1.50 0.22 0.33 8. Plan configuration 9.64% 14.46% 25.30% 50.60% 0.50 0.32 0.16 9. Height regularity 62.65% 22.89% 13.25% 1.20% 0.50 0.38 0.19 10. Wall façade openings and alignment 61.45% 2.41% 20.48% 15.66% 0.50 0.12 0.06 Group 3. Floor slabs and roofs 11. Horizontal diaphragms 49.40% 25.30% 24.10% 1.20% 0.75 0.47 0.35 12. Roofing system 12.05% 53.01% 20.48% 14.46% 2.00 0.47 0.93 Group 4. Conservation status and other elements 13. Fragilities and conservation status 50.60% 32.53% 16.87% 0.00% 1.00 0.38 0.38 14. Non-structural elements 26.51% 69.88% 2.41% 1.20% 0.75 0.22 0.17 Table 5.10: Example of conservative grades obtained after combining different grades and QC levels. Catalogue for Historical Monuments is an example of this, given that there is a vast universe of buildings that have been documented with variable levels of accuracy. Even if, in many cases, the information is not considered enough for performing detailed seismic vulnerability assessment, the information they provide is relevant and represent a valuable opportunity to complement with aspects to be enhanced. Such vulnerability assessment involving the evaluation of a large number of parameters and enclosing information related to historical monuments in individual datasheets, clearly benefits from a graphical representation in a simplified and easy-to-use scheme that is helpful for easily communicating the vulnerability of a construction together with the level of uncertainty that its survey has at a determined moment in order to provide a more accurate description during field campaigns. Figure 5.2 shown in Section 3.2 is herein revisited to exemplify the assessment for the building SECII-27 (Figure 5.12). 130 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. In this visual example, parameters such as BP2 obtained a grade A. Thus, since grade A has a numerical value of zero, no section is shadowed in the corresponding sector. Parameter BP6 was evaluated initially with grade B, which is associated with a numerical value and consequently, the most inner circle sector or slice is colour filled (dark gray). Nevertheless, as referred previously, uncertainty regarding this grading is to be considered and the light-shadowed sectors or slices (in light gray) imply that the parameter can potentially be graded as C or D. Figure 5.12: Example of a graphical representation for the building SECII-27. This implies a Quality Check assessment QC = 2 or QC = 3. Furthermore, the presence of the light-shadowed sectors indicates the need for further surveying of this parameter, BP6. Given that the sizes of the sectors or slices are geometrically proportional to the grade values and relative weights, the observation of such a graphical output allows correlating that a more colour-filled figure implies a higher vulnerability index. Additionally, another useful interpretation of this graphical output is uncertainty on the assessment of some parameters, indicated by the presence of lighter coloured sectors. 5.6 Final remarks. Seismic risk represents a relevant agent for human settlements, given the grade of destructiveness that a single event can impose. Furthermore, experiences such as the 2017 Earthquakes in Mexico demonstrate the relevance of promoting awareness through simple to more complex approaches to assess the structural and seismic vulnerability of traditional constructions to cultural assets. It is acknowledged that 5.6. FINAL REMARKS. 131 previous works devoted to risk prevention and assessment in the Mexican context are noteworthy; however, the large-scale assessment of historical constructions is still a gap. Nevertheless, the existence of a National Catalogue of Historical Monuments presents an interesting opportunity for performing simplified approaches, resourcing to the Vulnerability Index Method, herein presented, adapted and exposed. The present document implemented the Vulnerability Index Method approach for assessing a total of 83 buildings/constructions in the historic city of Atlixco, Puebla. Adopting this method demanded performing a detailed analysis of all parameters for the building and façade approach to assess the necessary slight redefinition and adapting criteria. A nuclear part of this exercise has been the adjustment of the parameters in order to fit the criteria to the Mexican built environment. This process demanded more than a translation but decoding the rationale and the justification behind the criteria. Despite the great value of the available information, the sources of data presented several gaps during the data-acquisition process. Given the difficulties for addressing them by performing on-site visits, a framework for contextualising and documenting the uncertainty was developed. This strategy is aimed to be adopted as a response to the lack of information, a common issue when reviewing existing databases. The implementation of a strategy based on assuming the uncertainty for defining plausible ranges for the vulnerability index and consequently the mean damage grades, instead of closed values resulted coherent and suitable for testing the methodology against the observed damages after the 19th of September 2017 event. The results of the assessment exhibit a good correspondence between the forecasted mean damage grades with the real and observed damage in buildings. This qualitative correspondence suggests the suitability of using this approach for assessing masonry structures in Mexico, accepting the eventual need of adapting the criteria for vulnerability class grading (such as herein presented). The strategies herein presented for contextualising and documenting the uncertainty during data acquisition are more than an instrumental assumption. Those strategies are meant for using existing information as a starting point regardless of its accuracy or quality, not overlooking the issue and taking it into account for a more robust analysis. Even if the present state of information is considered insufficient for performing largescale seismic vulnerability assessments, the information contained can be complemented with additional information that could be seen as to be included into the current updates of the datasheets of the National Catalogue. If information is scarce and potentially leading to a range of values for the seismic vulnerability indices and mean damage grade values, the definition of a qualitative comparison and other graphical analysis would facilitate the use of that information for decision-making. The results and the ability to use this method for assessing the seismic vulnerability of masonry constructions in Mexico may be limited because of the representativeness of the buildings of Atlixco when 132 CHAPTER 5. VULN. ASSESSMENT OF MEXICAN HUL: SUITABILITY AND UNCERTAINTY ANALYSIS. compared to different built environments in the country. However, it is important to note that no relevant typological divergences from the typical constructions of the central region of the country were found. All the parameters of the method were suitable for assessing the constructions, supporting the adoption of this method for other environments in which the parameters (and their grading conditions) are reasonably applicable. The hypothesis of extrapolating the method to another construction environment that fits the conditions considered in Appendix B is then sustained. Nevertheless, it is convenient to consider the relevance of testing the approach in the context of experimental data for more accurate grading and selection of ductility values. Another opportunity for enhancement lies in defining a more-suitable scale for damages, consistent with the common failures and mechanisms found in the Mexican built environment, including the adoption of continuous grade scales instead of closed discrete grades for assessing some quantitative parameters. The ranges for real/observed damage and the forecasted/calculated mean damage grade are often overlapped, representing a good precedent for performing a series of potential future works that would improve the application of this approach. An experience with more accurate data (i.e., with a better resolution) would facilitate performing sensibility analysis for testing numerous possibilities under the basis of the results herein presented. For example, adjustments on the parameter’s weights, the impact of the suppression of some parameters that present no variability among the samples or even the inclusion of new parameters, using data from the Mexican National Atlas of Risk or social vulnerability indicators for involving more stakeholders interested in the impact of seismic risk in human settlements are foreseen.