Dans le cadre de l'analyse et de la gestion intégrée des risques sismiques, deux approches sont proposées. La première, une méthode pseudo - adaptative de réponse modale (PSA), qui estime la réponse sismique des bâtiments à portiques, avec une précision acceptable et un temps de calcul et d'analyse réduit. En effet, dans le cadre de l'analyse de pushover multimodale (MPA), la courbe de capacité se construit sur la base d'une approche énergétique et le changement des propriétés modales après plastification est évalué à partir des vecteurs de déplacement pendant l'analyse de pushover. L'estimation des réponses en termes de déplacements absolus et relatifs, forces de cisaillement et rotations est satisfaisante comparativement aux résultats d'analyses non linéaires complètes. La seconde approche porte sur l'évaluation post-sismique des dommages structuraux à partir de dommages locaux observés. Elle est fondée sur une relation postulée entre le dommage et la probabilité résiduelle de ruine, à deux niveaux : l'étage et le bâtiment complet. Quatre portiques sont analysés et les résultats sont comparés à une approche mécanique qui estime l'endommagement du système à partir de la perte de raideur de la courbe de capacité. Les résultats obtenus montrent de bonnes estimations du niveau de dommage global. Ainsi, cette approche pourrait bien faire partie d'un outil d'aide à la décision dans le cadre des programmes d'évaluation urbaine des dommages qui requièrent des estimations simultanément rapides et précises / Within the overall framework of seismic risk analysis and management two approaches are presented. First, the Pseudo-Adaptive Uncoupled Modal Response Analysis (PSA) aims to provide improved estimates of seismic response for framed buildings, with an acceptable accuracy and a reduced calculation time duration. It relies on an energy-based equivalent displacement to develop the capacity curve and a pseudo-adaptive feature that considers changes in modal shapes after yielding, within the framework of the widely used Modal Pushover Analysis. According to the results, PSA is able to provide good estimates of structural responses such as displacements, storey drifts, shear forces and rotations, in comparison to a complete Nonlinear Time History Analysis. Second, a strategy for post-seismic evaluation of structural global damage is proposed on the basis of observed local damages and the postulation of adequate relationships between damage and residual probability of failure at two levels: a storey level prior to a building level. Three factors are proposed to reflect the influence of components damage at each of those levels. The obtained results appear as good predictions of the global damage. Accordingly, this strategy has the potential for being a first step within the implementation framework of a decision-making tool for rapid and accurate estimates of structural damages

Performance-Based (PBE) and Consequence-Based (CBE) are new approaches to seismic design, evaluation and risk assessment, in which design criteria are devised to achieve stated performance objectives, and regional losses to civil infrastructure are mitigated through selective interventions for critical components of a civil infrastructure. These new approaches give engineers more flexibility in achieving performance goals but require substantial additional computational resources to fully achieve performance goals. As a step toward making such approaches feasible, this dissertation develops a number of computationally efficient methods for performing finite element-based structural system dynamic response analysis and reliability assessment. The Enhanced Uncoupled Modal Response History Analysis (EUMRHA) procedure developed herein is an efficient response analysis procedure to make the analysis of dynamic structural response to earthquakes in the nonlinear range less time-consuming. This technique is used to investigate the potential for aftershocks to cause additional damage to steel moment frame buildings, utilizing a technique designed to enhance the efficiency of Monte Carlo simulation in estimating low-probability events. Relatively simple probabilistic tools are proposed for purposes of rapid structural evaluation and condition assessment of damaged buildings. Finally, an analysis-based inspection scheme based on an associated probability model of connection damage is proposed for assessing the safety condition of existing buildings, and a procedure to assess the likely performance of an un-repaired building during a future earthquake is developed.

Two approaches were developed for detecting earthquake damaged buildings from post-event aerial photographs using shadow analysis and perceptual grouping. In the first approach, it is assumed that the vector boundaries of the buildings are not known a priori. Therefore, only the post-event aerial photographs were used to detect the collapsed buildings. The approach relies on an idea that if a building is fully damaged then, it will not generate a closed contour. First, a median filter is applied to remove the noise. Then, the edge pixels are detected through a Canny edge detector and the line segments are extracted from the output edge image using a raster-to-vector conversion process. After that, the line segments are grouped together using a three-level hierarchical perceptual grouping procedure to form a closed contour. The principles used in perceptual grouping include the proximity, the collinearity, the continuity and the perpendicularity. In the second approach, it is assumed that the vector boundaries of the buildings are known a priori. Therefore, this information is used as additional data source to detect the collapsed buildings. First, the edges are detected from the image through a Canny edge detector. Second, the line segments are extracted using a raster-to-vector conversion process. Then, a two-level hierarchical perceptual grouping procedure is used to group these line segments. The boundaries of the buildings are available and stored in a GIS as vector polygons. Therefore, after applying the perceptual grouping procedure, the damage conditions of the buildings are assessed on a building-by-building basis by measuring the agreement between the detected line segments and the vector boundaries.

Urbanisation will require significant expansion of underground infrastructure, which results in unavoidable ground displacements that affect the built environment. Predicting the interaction between a tunnel, the soil and existing structures remains an engineering challenge due to the highly non-linear behaviour of both the soil and the building. This thesis investigates the interaction between a surface structure and tunnelling-induced ground displacements. Specifically, novel three-dimensionally printed building models with brittle material behaviour, similar to masonry, were developed and tested in a geotechnical centrifuge. This enabled replication of building models with representative global stiffness values and realistic building features including strip footings, intermediate walls, a rough soil-structure interface, building layouts and façade openings. By varying building characteristics, the impact of structural features on both the soil and building response to tunnelling in dense sand was investigated. Results illustrate that the presence of surface structures considerably altered the tunnelling-induced soil response. The building-to-tunnel position notably influences the magnitude of soil displacements and causes localised phenomena such as embedment of building corners. An increase of the façade opening area and building length reduces the alteration of the theoretical greenfield settlements, in particular the trough width. Moreover, the impact of varying the building layout is discussed in detail. For several building-tunnel scenarios, building distortions are quantified and the crucial role of building features is demonstrated. Structures spanning the greenfield inflection point experienced more deformation than identical structures positioned in either sagging or hogging, and partitioning a structure either side of the greenfield inflection point is shown to lead to unconservative damage assessments. Results also quantify the significant extent to which structural distortions increase as façade openings and building length increases. Observed building damage and cracking patterns confirm the reported trends. The experimental results are used to evaluate the performance of available methods to assess the behaviour of buildings to tunnelling. Predictions ignoring soil-structure interaction are usually overly conservative, while approaches based on the relative stiffness of a structure and the soil result in inconsistent predictions, though some methods performed better than others. Practical improvements to consider structural details when assessing this tunnel-soil-structure system are finally proposed.

abstract: Recent extreme weather events such the 2020 Nashville, Tennessee tornado and Hurricane Maria highlight the devastating economic losses and loss of life associated with weather-related disasters. Understanding the impacts of extreme weather events is critical to mitigating disaster losses and increasing societal resilience to future events. Geographical approaches are best suited to examine social and ecological factors in extreme weather event impacts because they systematically examine the spatial interactions (e.g., flows, processes, impacts) of the earth’s system and human-environment relationships. The goal of this research is to demonstrate the utility of geographical approaches in assessing social and ecological factors in extreme weather event impacts. The first two papers analyze the social factors in the impact of Hurricane Sandy through the application of social geographical factors. The first paper examines how knowledge disconnect between experts (climatologists, urban planners, civil engineers) and policy-makers contributed to the damaging impacts of Hurricane Sandy. The second paper examines the role of land use suitability as suggested by Ian McHarg in 1969 and unsustainable planning in the impact of Hurricane Sandy. Overlay analyses of storm surge and damage buildings show damage losses would have been significantly reduced had development followed McHarg’s suggested land use suitability. The last two papers examine the utility of Unpiloted Aerial Systems (UASs) technologies and geospatial methods (ecological geographical approaches) in tornado damage surveys. The third paper discusses the benefits, limitations, and procedures of using UASs technologies in tornado damage surveys. The fourth paper examines topographical influences on tornadoes using UAS technologies and geospatial methods (ecological geographical approach). This paper highlights how topography can play a major role in tornado behavior (damage intensity and path deviation) and demonstrates how UASs technologies can be invaluable tools in damage assessments and improving the understanding of severe storm dynamics (e.g., tornadic wind interactions with topography). Overall, the significance of these four papers demonstrates the potential to improve societal resilience to future extreme weather events and mitigate future losses by better understanding the social and ecological components in extreme weather event impacts through geographical approaches. / Dissertation/Thesis / Doctoral Dissertation Geography 2020

Incremental damage assessment is a tool used to assess the justification for expensive modifications of inadequate dams. The input data to incremental damage assessment are the output from the breach analysis and flood routing. For this reason, flood routing should be conducted carefully. Distorted results from the flood routing technique or unstable modeling of the problem will distort the results of an incremental damage assessment, because an error in the estimated incremental stage will cause a certain error in the estimated incremental damages. The objectives of this study were (1) to perform a comprehensive survey of the available dam break flood-routing techniques, (2) to evaluate the performance of commonly used flood-routing techniques for predicting failure and no-failure stage, incremental stage, average velocities, and travel times, and (3) to develop a set of recommendations upon which future applications of dam break models can be based. Flood-routing techniques that are evaluated cover dynamic routing as contained in DAMBRK, and kinematic, Muskingum-Cunge, and normal depth storage routing as contained in the Hydrological Engineering Center (HEC 1). These techniques were evaluated against the more accurate two-dimensional flood-routing technique contained in the diffusion hydrodynamic model (DHM). Results and errors from different techniques for different downstream conditions were calculated and conclusions were drawn. The effect of the errors on the incremental stage and the errors in the incremental stage were estimated. Overall, the performance of one-dimensional techniques in predicting peak stages, and assessing a two-feet criterion showed that DAMBRK did best, and normal depth storage and outflow did worst. This overall ranking matches the degree of simplification in representing the true flood-routing situation. However, in some circumstances DAMBRK performed worst, and normal depth storage and outflow outperformed either the Muskingum-Cunge or kinematic techniques. Thus, it is important to understand the specific performance characteristics of all the methods when selecting one for a flood-routing application.

The discovery of carbon nanotubes in 1990s popularized a new area of research in materials science called Nanoscience. In the following decades, several carbon based nanoparticles were discovered or engineered and with the discovery of Graphene nanoplatelets (GNP) in 2010, carbon based nanoparticles were propelled as the most promising class of nanoparticles. High mechanical strength and stiffness, excellent electrical and thermal conductivity, and high strength to weight ratios are some of the unique abilities of CNTs and GNPs which allow their use in a wide array of applications from aerospace materials to electronic devices. In the current work presented herein, CNTs and GNPs are added to polymeric materials to create a nanocomposite material. The effects of this nanoparticle addition (a.k.a reinforcement) on the mechanical properties of the nanocomposite polymer materials are studied. Specifically, efforts are focused on studying fracture toughness, a material property that describes the material's ability to resist crack growth. Relative to the conventional metals used in structures, epoxy-based composites have poor fracture toughness. This has long been a weak link when using epoxy composites for structural applications and therefore several efforts are being made to improve their fracture toughness. In the first, second and third chapters, the enhancement of fracture toughness brought about by the addition of carbon nanotubes (CNTs) and graphene nanoplatelets (GNPs) was investigated. CNT-Epoxy and GNP-Epoxy Compact Tension (CT) samples were fabricated with 0.1% and 0.5% nanofiller weight concentrations. The potential synergistic effects of dual nanofiller reinforcements were also explored using CNT/GNP-Epoxy CT samples at a 1:3, 3:1 and 1:1 ratio of CNT:GNP. Displacement controlled CT tests were conducted according to ASTM D5045 test procedure and the critical stress intensity factor, $K_{IC}$, and the critical fracture energy, $G_{IC}$, were calculated for all the material systems. Significant enhancements relative to neat epoxy were observed in reinforced epoxies. Fracture surfaces were analyzed via scanning electron microscopy. Instances of CNT pullouts on the fracture surface were observed, indicating the occurrence of crack bridging. Furthermore, increased surface roughness, an indicator of crack deflection, was observed along with some crack bifurcations in the GNP-Epoxy samples. In the fourth chapter of Part I, the influence of pre-crack characteristics on the Mode-I fracture toughness of epoxy is investigated. Pre-crack characteristics such as pre-crack length, crack front shape, crack thickness and crack plane profile are evaluated and their influence on the peak load, fracture displacement, and the critical stress intensity factor, $K_{IC}$ is studied. A new method of razor blade tapping was used, which utilized a guillotine-style razor tapping device to initiate the pre-crack and through-thickness compression to arrest it. A new approach of quantitatively characterizing the crack front shape using a two-parameter function is introduced. Surface features present on the pre-crack surface are classified and their effects on the post crack initiation behavior of the sample are analyzed. This study aims to identify and increase the understanding of the various factors that cause variation in the fracture toughness data of polymeric materials, thus leading to more informed engineering design decisions and evaluations. Chapters six and seven of Part II investigate the SHM capabilities of dispersed MWCNTs in mock, inert, and active energetics. In these experimental investigations, the strain and damage sensing abilities of multi-walled carbon nanotube (MWCNT) networks embedded in the binder phase of polymer bonded energetics (PBEs) are evaluated. PBEs are a special class of particulate composite materials that consist of energetic crystals bound by a polymer matrix, wherein the polymer matrix serves to diminish the sensitivity of the energetic phase to accidental mechanical stimuli. The structural health monitoring (SHM) approach presented in this work exploits the piezoresistive properties of the distributed MWCNT networks. Major challenges faced during such implementation include the low binder concentrations of PBEs, presence of conductive/non-conductive particulate phases, high degree of heterogeneity in the PBE microstructure, and achieving the optimal MWCNT dispersion. In chapter seven, Ammonium Perchlorate (AP) crystals as the oxidizer, Aluminum grains as the metallic fuel, and Polydimethylsiloxane (PDMS) as the binder are used as the constituents for fabricating PBEs. To study the effect of each constituent on the MWCNT network's SHM abilities, various materials systems are comprehensively studied: MWCNT/PDMS (nBinder) materials are first evaluated to study the binder's electromechanical response, followed by AP/MWCNT/PDMS (inert nPBE) to assess the impact of AP addition, and finally, AP/AL/MWCNT/PDMS (active nPBE-AL) to evaluate the impact of adding conductive aluminum grains. Compression samples (ASTM D695) were fabricated and subjected to monotonic compression. Electrical resistance is recorded in conjunction with the mechanical test via an LCR meter. Gauge factors relating the change in normalized resistance to applied strain are calculated to quantify the electromechanical response. MWCNT dispersions, and mechanical failure modes are analyzed via scanning electron microscopy (SEM) imaging of the fracture surfaces. Correlations between the electrical behavior in response to the mechanical behavior are presented, and possible mechanisms that influence the electromechanical behavior are discussed. The results presented herein demonstrate the successful ability of MWCNT networks as structural health monitoring sensors capable of real-time strain and damage assessment of polymer bonded energetics. / Doctor of Philosophy / The discovery of carbon nanotubes in 1990s popularized a new area of research in materials science called Nanoscience. Carbon nanotubes (CNTs) are one of several forms of Carbon, meaning a differently structured carbon molecule in the same physical state similar to diamonds, graphite, and coal. In the following decades, several carbon based nanoparticles were discovered or engineered and with the discovery of Graphene (GNP) in 2010, carbon based nanoparticles were propelled as the most promising class of nanoparticles. High mechanical strength and stiffness, excellent electrical and thermal conductivity, and high strength to weight ratios are some of the unique abilities of CNTs and GNPs which allow their use in a wide array of applications from aerospace materials to electronic devices. In the current work presented herein, CNTs and GNPs are added to polymeric materials to create a nanocomposite material, where the term "composite" refers to a material prepared with two or more constituent materials. The effects of this nanoparticle addition (a.k.a reinforcement) on the mechanical properties of the nanocomposite polymer materials are studied. Specifically, efforts are focused on studying fracture toughness, a material property that describes the material's ability to resist crack growth. Fracture toughness is a critical material property often associated with material and structural failures, and as such it is very important for safe and reliable engineering design of structures, components, and materials. Moving from a single function (i.e. mechanical enhancement) to a more multi-functional role, taking advantage of the excellent electrical and mechanical abilities of CNTs, a structural health monitoring system is developed for use in polymer bonded energetics (eg. solid rocket propellants). When a material undergoes mechanical deformation or damage, the measured electrical properties of the material undergo some change as well. Using sensor networks built with multiple CNTs dispersed within a polymeric material, a whole structure can be made into an effective sensor where by simply monitoring the electrical properties, the extent of material deformation and damage can be known. Such a system is geared towards providing early warning of impending catastrophic material failures thus directly improving the safety during material handling and operations.

Pitting corrosion is a prevalent form of corrosive damage that can weaken, damage, and initiate failure in corrosion-resistant metallic materials. For instance, 304 stainless steel is commonly utilized in various structures (e.g., miter gates, heat exchangers, and storage tanks), but is prone to failure through pitting corrosion and stress corrosion cracking under mechanical loading, regardless of its high corrosion resistance. In this study, to better understand the pitting corrosion damage development, controlled corrosion experiments were conducted to generate pits on 304 stainless steel specimens with and without mechanical loading. The pit development over time was characterized using a high-resolution laser scanner. In addition, to achieve scalable and automatic assessment of pitting corrosion conditions, two convolutional neural network-based computer vision algorithms were adopted and implemented to evaluate the efficacy of networks to identify existence of pitting damage. One was a newly trained convolutional neural network (CNN) using MATLAB software, while the other one was a retrained version of GoogLeNet. Overall, the experimental results showed that time is the dependent variable in predicting pit depth. Meanwhile, loading conditions significantly influence pit morphology. Under compression loading, pits form with larger surface opening areas, while under tension loading, pits have smaller surface opening areas. Deep pits of smaller areas are dangerous for structural members, as they can lead to high stress concentrations and early stress corrosion cracking (SCC). Furthermore, while the training library was limited and consisted of low-resolution images, the retrained GoogLeNet CNN showed promising potential for identifying pitting corrosion based on the evaluation of its performance parameters, including the accuracy, loss, recall, precision, and F1-measure.

This research reviews and extends the impedance-based structural health monitoring technique in order to detect and identify structural damage on various complex structures. The basic principle behind this technique is to apply high frequency structural excitations (typically higher than 30 kHz) through the surface-bonded piezoelectric transducers, and measure the impedance of structures by monitoring the current and voltage applied to the transducers. Changes in impedance indicate changes in the structure, which in turn can indicate that damage has occurred. Several case studies, including a pipeline structure, a composite reinforced aluminum plate, a precision part (gear), a quarter-scale bridge section, and a steel pipe header, demonstrate how this technique can be used to detect damage in real-time. A method to process impedance measurements to prevent significant temperature and boundary condition changes registering as damage has been developed and implemented. Furthermore, the feasibility of using the technique for high temperature structures and for condition monitoring of critical facilities subjected to a severe natural disaster has been investigated. While the impedance-based structural health monitoring technique indicates qualitatively that damage has occurred, more information on the nature of damage is necessary for remote structures. In this research, two different damage identification schemes have been combined with the impedance method in order to quantitatively assess the state of structures. One is based on a wave propagation modeling, and the other is the use of artificial neural networks. A newly developed wave propagation model has been developed and combined with the impedance method in order to estimate the severity of damage. Numerical and experimental investigations on 1-dimensional structures were presented to illustrate the effectiveness of the combined approach. Furthermore, to avoid the complexity introduced by conventional computational methods in high frequency ranges, multiple sets of artificial neural networks were integrated with the impedance-based health monitoring technique. By incorporating neural network features, the technique is able to detect damage in its early stage and to determine the severity of damage without prior knowledge of the model of structures. The dissertation concludes with experimental examples, investigations on a quarter-scale steel bridge section and a space truss structure, in order to verify the performance of the proposed methodology. / Ph. D.

Natural disasters strike anywhere, disrupting local communication and transportation infrastructure, making the process of assessing specific local damage difficult, dangerous, and slow. The goal of Building Damage Assessment (BDA) is to quickly and accurately estimate the location, cause, and severity of the damage to maximize the efficiency of rescuers and saved lives. In current machine learning BDA solutions, attention operators are the most recent innovations adopted by researchers to increase generalizability and overall performances of Convolutional Neural Networks for the BDA task. However, the latter, nowadays exploit attention operators tailored to the specific task and specific neural network architecture, leading them to be hard to apply to other scenarios. In our research, we want to contribute to the BDA literature while also addressing this limitation. We propose Global Voxel Transformer Operators (GVTOs): flexible attention-operators originally proposed for Augmented Microscopy that can replace up-sampling, down-sampling, and size-preserving convolutions within either a U-Net or a general CNN architecture without any limitation. Dissimilar to local operators, like convolutions, GVTOs can aggregate global information and have input-specific weights during inference time, improving generalizability performance, as already proved by recent literature. We applied GVTOs on a state-of-the-art BDA model and named it GVT-BDNet. We trained and evaluated our proposal neural network on the xBD dataset; the largest and most complete dataset for BDA. We compared GVT-BDNet performance with the baseline architecture (BDNet) and observed that the former improves damaged buildings segmentation by a factor of 0.11. Moreover, GVT-BDNet achieves state-of-the-art performance on a 10% split of the xBD training dataset and on the xBD test dataset with an overall F1- score of 0.80 and 0.79, respectively. To evaluate the architecture consistency, we have also tested BDNet’s and GVT-BDNet’s generalizability performance on another segmentation task: Tree & Shadow segmentation. Results showed that both models achieved overall good performances, scoring an F1-score of 0.79 and 0.785, respectively. / Naturkatastrofer sker överallt, stör lokal kommunikations- och transportinfrastruktur, vilket gör bedömningsprocessen av specifika lokala skador svår, farlig och långsam. Målet med Building Damage Assessment (BDA) är att snabbt och precist uppskatta platsen, orsaken och allvarligheten av skadorna för att maximera effektiviteten av räddare och räddade liv. Nuvarande BDA-lösningar använder Convolutional Neural Network (CNN) och ad-hoc Attention Operators för att förbättra generaliseringsprestanda. Nyligen föreslagna attention operators är dock specifikt skräddarsydda för uppgiften och kan sakna flexibilitet för andra scenarier eller neural nätverksarkitektur. I vår forskning bidrar vi till BDA -litteraturen genom att föreslå Global Voxel Transformer Operators (GVTO): flexibla attention operators som kan appliceras på en CNN -arkitektur utan att vara bundna till en viss uppgift. Nyare litteratur visar dessutom att de kan öka utvinningen av global information och därmed generaliseringsprestanda. Vi tillämpade GVTO på en toppmodern CNN-modell för BDA. GVTO: er förbättrade skadessegmenteringsprestandan med en faktor av 0,11. Dessutom förbättrade de den senaste tekniken för xBD-testdatauppsättningen och nådde toppmodern prestanda på en 10% delning av xBD-träningsdatauppsättningen. Vi har också utvärderat generaliserbarheten av det föreslagna neurala nätverket på en annan segmenteringsuppgift (Tree Shadow segmentering), vilket uppnådde över lag bra prestationer.