Polymeric composites reinforced with bio-materials have advantages over composites with synthetic reinforcements. Bio-based composites use low-cost and renewable reinforcements, have nonabrasive properties for machining, have improved damping characteristics, and have potential for energy recycling. However, the limited use of bio-based composites is because their mechanical properties are typically much lower than those of synthetic composites. The objective of this study was to combine state-of-the-art imaging tools with emerging numerical modeling methods for an integrated, multi-level characterization of bio-based reinforcements and their composites. Digital photography (2D) will allow collection of full-field digital images of the surface of sample composites, which will be used for characterization of the morphological structure of fillers (copper wire or wood particle) and of model composites. Mechanical experiments (tension load) on isolated fillers and on model composites will allow imaging of the deformed material. By correlating relative positions of thousands of surface features between consecutive images, digital image correlation (DIC) algorithms can be used to map surface deformation fields and calculate surface strain fields. Digital imaging methods can only record deformations and strains. The interpretation of those strains in terms of material properties, such as position-dependent modulus of a heterogeneous composite material, requires simultaneous modeling. The modeling must use morphology-based methods that can handle anisotropy, heterogeneity, and the complex structure of bio-based composites such as wood plastic composites. This research used the material point method (MPM) as a modeling tool. MPM is a particle-based, meshless method for solving problems in computational mechanics. The crucial advantage of MPM over other methods is the relative ease of translating pixels from digital images into material points in the analysis. Thus digital images (2D) used in our experiments were used as direct input to the MPM software, so that the actual morphologies, rather than idealized geometries, were modeled. This procedure removes typical uncertainties connected with idealization of the internal features of modeled materials. It also removes variability of specimen to specimen due to morphology variations. Full-field imaging techniques and computer modeling methods for analysis of complex materials have developed independently. This research Coupled imaging and modeling and used inverse problem methodology for studying bio-particulate composites. The potential of coupling experiments with morphology-based modeling is a relatively new area. This work studied the morphology and mechanical properties of copper wire (for validation experiments) and wood particles used for reinforcement in polymer composites. The goal was to determine the in situ mechanical and interfacial properties of copper wire and then wood particles. By comparison of DIC results to MPM, the conclusion is MPM simulation works well by simulating 3D composite structure and using Matlab software to do qualitative and quantitative comparisons. Copper validation tests showed that copper wire is too stiff compared to polymer such that the inclusion modulus had low effect on the surface strains (DIC experimental results). Wood particle worked better because modulus of wood is much lower than copper. By qualitative comparison of the wood particle specimens, we could deduce that the in situ properties of wood particles are lower than bulk wood. Quantitative analysis concentrated on small area and got more exact results. In a 90 degree particle quantitative study, MPM simulations were shown to be capable of tracking the structure of wood particle plastic, which involved failure. The entire approach, however, is not very robust. We can get some results for mechanical properties, but it does not seem possible to extract all anisotropic properties from a few DIC tests, as some researcher have suggested. / Graduation date: 2012

The feasibility of using superelastic shape memory alloys in the retrofit of multiple frame bridges is investigated. First, three shape memory alloy constitutive models with various levels of complexity are compared in order to determine the significance of including subloops and cyclic loading effects on the structural response. The results show that the structural response is more sensitive to the shape memory alloys strength degradation and residual deformation than the sublooping behavior. Next, two parametric studies are conducted to explore the sensitivity of hinge opening to the mechanical behavior of the superelastic shape memory alloys. The first study is focused on the hysteretic properties of the alloy that could vary depending on the chemical composition or the manufacturing process of the alloy, while the second study targets the changes in the mechanical behavior of shape memory alloys resulting from the variability in the ambient temperature. The results show that the hysteretic behavior of shape memory alloys has only a slight effect on the bridge hinge opening as long as the recentering property is maintained. A detailed study on the effect of temperature shows that a reduction in the ambient temperature tends to negatively affect the hinge opening while an increase in temperature results in a slight improvement. Next, a parametric study is conducted to examine the effectiveness of shape memory alloy retrofit devices in limiting hinge openings in bridges with various properties. In addition, a comparison is made with other devices such as conventional steel restrainers, metallic dampers, and viscoelastic solid dampers. The results illustrate that superelastic shape memory alloys are superior in their effectiveness compared to other devices in the case of bridges with moderate period ratios and high level of ductility, especially when subjected to strong earthquakes.

Atomistic simulations are employed in this thesis to investigate defect nucleation and free volume of grain boundaries and nanocrystalline materials. Nanocrystalline materials are of particular interest due to their improved mechanical properties and alternative strain accommodation processes at the nanoscale. These processes, or deformation mechanisms, within nanocrystalline materials are strongly dictated by the larger volume fraction of grain boundaries and interfaces due to smaller average grain sizes. The behavior of grain boundaries within nanocrystalline materials is still largely unknown. One reason is that experimental investigation at this scale is often difficult, time consuming, expensive, or impossible with current resources. Atomistic simulations have shown the potential to probe fundamental behavior at these length scales and provide vital insight into material mechanisms. Therefore, work conducted in this thesis will utilize atomistic simulations to explore structure-property relationships of face-centered-cubic grain boundaries, and investigate the deformation of nanocrystalline copper as a function of average grain size. Volume-averaged kinematic metrics are formulated from continuum mechanics theory to estimate nonlocal deformation fields and probe the nanoscale features unique to strain accommodation mechanisms in nanocrystalline metals. The kinematic metrics are also leveraged to explore the tensile deformation of nanocrystalline copper at 10K. The distribution of different deformation mechanisms is calculated and we are able to partition the role of competing mechanisms in the overall strain of the nanocrystalline structure as a function of grain size. Grain boundaries are observed to be influential in smaller grained structures, while dislocation glide is more influential as grain size increases. Under compression, however, the resolved compressive normal stress on interfaces hinders grain boundary plasticity, leading to a tension-compression asymmetry in the strength of nanocrystalline copper. The mechanisms responsible for the asymmetry are probed with atomistic simulations and the volume-averaged metrics. Finally, the utility of the metrics in capturing nonlocal nanoscale deformation behavior and their potential to inform higher-scaled models is discussed.

CNPQ / O presente trabalho tem como objetivo o desenvolvimento de um biosensor, utilizando redes de Bragg em fibras óticas, capaz de mensurar deformações em tecido ósseo. Inicialmente foram realizados testes de esterilização e desinfecção com sensores, após o processo completo, os sensores foram testados quanto a sua sensibilidade. Com isso os métodos de esterilização e desinfecção autoclave, óxido etileno, hipoclorito de sódio, cloramina T e ácido peracético, foram eficazes para eliminação e ou inativação dos microrganismos, sem que haja alterações no comportamento das FBGs. Com isso definem-se métodos aplicáveis para possíveis estudos, onde fibras óticas foram implantadas em animais in vivo para estudos de comportamento metabólico com a presença de materiais inorgânicos. Na continuidade dos trabalhos foram realizados testes de compatibilidade de fibras óticas inseridas no tecido subcutâneo de ratos. Baseado na análise descritiva dos cortes histológicos pode-se concluir que as fibras ópticas sem revestimento e com revestimento foram biocompatíveis quando implantadas em subcutâneo de ratos. Sendo assim, foi possível contemplar o desenvolvimento de um sensor baseado em redes de Bragg em fibras óticas para implantação direta no tecido ósseo. Posteriormente foram realizados testes-piloto para o desenvolvimento da metodologia de construção e implante no desenvolvimento de biosensores. Com todo o trabalho feito até à data, o desenvolvimento e construção de um biosensor que foi implantado in vivo foi possível. O biossensor foi fixado na região mandibular de um bovino quatro meses de idade. Uma semana após a implantação, os animais receberam diferentes tipos de alimentos e foi monitorizada durante a ingestão de alimentos. Com as deformações medidas foi possível a análise dos diferentes padrões mastigatórios e diferentes magnitudes de força e frequência mastigatória que o animal desempenhada durante o processo de mastigação. / This study aims to develop a biosensor using Bragg gratings in optical fibers, capable of measuring deformation in bone tissue for this took a lot of work. Initially sterilization and disinfection with sensors tests were performed after the entire process, the sensors were tested for their sensitivity. In the same study methods of sterilization and disinfection have been tested to assess its efficiency. Thus methods of sterilization and disinfection autoclave, ethylene oxide, sodium hypochlorite, chloramine T and peracetic acid were effective for removing or inactivating microorganisms and, without changes in the behavior of the FBGs. With this set up methods applicable for possible studies, where optical fibers are implanted in in vivo animal studies metabolic behavior in the presence of inorganic materials. Continuing the work was carried out optical fiber compatibility tests inserted in the subcutaneous tissue of rats. Based on the descriptive analysis of histological sections can be concluded that the optical fiber uncoated and coated were biocompatible when implanted in rat subcutaneous tissue. Thus it is possible to contemplate the development of a sensor based on Bragg gratings in optical fibers for direct implantation into the bone tissue. Following on from the work were carried out pilot tests for development of construction methodology and implant in biosensor development. With all the work done to date, the development and construction of a biosensor that was implanted in vivo was possible. The biosensor was fixed in the mandibular region of a bull four months old. A week after implantation, the animal received different types of food and was monitored during the intake of foods. With the deformations measured parse the different masticatory patterns and different magnitudes of force and masticatory frequency that the animal developed during the chewing process completing the objectives of this study.

Neste trabalho é apresentado um método para a identificação de falha causada por barras quebradas em rotores de motores de indução trifásicos através da análise de sua deformação dinâmica, utilizando redes de Bragg em fibra óptica (FBG) como elemento sensor. A análise da deformação dinâmica permite a detecção e diagnóstico de uma série de falhas mecânicas como desalinhamento, desbalanceamento, falhas em rolamentos e folga mecânica. Além de falhas mecânicas é possível identificar falhas elétricas como quebras ou trincas em anéis ou barras do rotor. Para as medidas de deformação dinâmica foram utilizados sensores baseados em redes de Bragg em fibra óptica que possuem como características principais a grande capacidade de multiplexação, imunidade à radiação eletromagnética e capacidade de operação a longas distâncias. Ensaios foram feitos em um motor de indução de pequeno porte (3 CV) afim de validar o método e futuramente aplicar em máquinas de médio e grande porte. O motor foi alimentado por três diferentes fontes de alimentação: rede elétrica da concessionária, geradores síncronos e inversor de frequência, para condições de 75% e 100% de sua carga nominal. Nos ensaios foram utilizados um rotor sem defeito de barra e, posteriormente, um rotor com uma barra interrompida. Através dos ensaios foi possível identificar a falha para as duas condições de carga ensaiadas e com as três diferentes fontes de alimentação. A falha de barra quebrada no rotor é identificada em duas regiões do espectro de frequência obtidos com os três conjuntos de ensaios realizados. A primeira região em torno da frequência mecânica de rotação do rotor e a segunda em torno do dobro da frequência elétrica da fonte de alimentação. O sistema se mostrou bastante sensível, com boa relação sinal ruído e com vantagens em relação aos principais métodos e sensores comumente utilizados para a identificação da falha de barras quebradas em motores de indução. / This work presents the rotor broken bar failure identification through strain analysis using fiber Bragg gratings (FBG) as sensor. The dynamic strain analysis allow the diagnostic and detect several mechanical failures like misalignment, unbalances, bearing failures and mechanical clearance, also electrical failures like rotor broken rings and rotor broken bar. Fiber Bragg gratings sensors were used to measure the dynamic deformation, some advantages of FBG are high multiplexing capability, electromagnetic interference immunity and long distance operation capability. Tests were conducted in a small induction motor (3 HP) in order to validate the method and allow it to be applied inthe future in medium and big size machines. The motor was fed through three different sources (directly from the electrical grid, synchronous generators and by one frequency inverter) at 75% and 100% of rated load condition. Tests were conducted with one healthy rotor and then with a rotor with a broken bar. The tests allow identifying the rotor broken bar failure in both load conditions. Rotor broken bar failure could be identified in two spectrum regions for all conducted tests. The first one was the region around the motor mechanical rotational frequency and the second one was the region around twice times power supply frequency. The system have been presented a high sensibility, a god signal to noise relationship (SNR) and some advantages over the main common sensors and methods used to identify and analyze the dynamic eccentricity and rotor broken bar failures in induction motors.

CNPq / Materiais tixotrópicos possuem aplicações industriais economicamente importantes. Modelos constitutivos descrevendo seu comportamento, propostos recentemente, são formulados por meio de um sistema acoplado de duas equações: equação constitutiva (baseada em modelos viscoelásticos clássicos) e a equação de taxa (que descreve a evolução microestrutural do material). O módulo de cisalhamento e o(s) coeficiente(s) de viscosidade são considerados, nesta classe de modelos, funções do parâmetro estrutural. As expressões utilizadas para tais funções são definidas satisfazendo limites assintóticos, de tal forma que o modelo seja fisicamente consistente. Entretanto, não há consenso quanto a forma em que as expressões são formuladas. Objetivou-se determinar o formato da função associada ao módulo de cisalhamento, a partir de dados de testes reológicos, utilizando um modelo de Jeffreys modificado. A obtenção da expressão do módulo de cisalhamento foi definida como um problema inverso e, portanto, a teoria e algumas estratégias associadas foram discutidas. Utilizou-se uma estrutura multiobjetiva juntamente com o método de regularização de Tikhonov e o critério de escolha de parâmetro curva L, para a obtenção de soluções de problemas mal-postos associados. Os algoritmos formalizados no trabalho foram implementados por meio de um código desenvolvido no MATLAB. Como resultados, uma nova proposta para a função associada ao módulo de cisalhamento foi obtida e os parâmetros associados ao modelo constituído desta nova proposta foram ajustados a dados de testes reológicos. / Thixotropic materials have economically important industrial applications. Recently proposed constitutive models describing its behavior are formulated by means of a two coupled equations system: the constitutive equation (based on viscoelastic classic models) and the rate equation (in which the microstructural evolution is described). The shear modulus and the viscosity coefficient(s) are considered, in such a class of models, as functions of the structural parameter. The expressions used for such functions are defined by satisfying some asymptotic limits, in a way that the model is physically consistent. However, there is no agreement as to the form in which the expressions are formulated. It is aimed to determine the form of shear modulus function, from rheological tests data, using a modified Jeffreys model. The obtainment of an expression for the shear modulus function is defined as an inverse problem and therefore the theory and some strategies associated were discussed. It is used a multi-objective framework together with the Tikhonov regularization method and the L-curve parameter-choice criterion in order to get the solution for associated ill-posed problems. The algorithms formalized throughout the work were implemented through a MATLAB developed code. As results, a new proposal for the shear modulus function were obtained and the parameters associated with the model constituted of this new proposal are fitted to rheological tests data.

CNPq / Materiais tixotrópicos possuem aplicações industriais economicamente importantes. Modelos constitutivos descrevendo seu comportamento, propostos recentemente, são formulados por meio de um sistema acoplado de duas equações: equação constitutiva (baseada em modelos viscoelásticos clássicos) e a equação de taxa (que descreve a evolução microestrutural do material). O módulo de cisalhamento e o(s) coeficiente(s) de viscosidade são considerados, nesta classe de modelos, funções do parâmetro estrutural. As expressões utilizadas para tais funções são definidas satisfazendo limites assintóticos, de tal forma que o modelo seja fisicamente consistente. Entretanto, não há consenso quanto a forma em que as expressões são formuladas. Objetivou-se determinar o formato da função associada ao módulo de cisalhamento, a partir de dados de testes reológicos, utilizando um modelo de Jeffreys modificado. A obtenção da expressão do módulo de cisalhamento foi definida como um problema inverso e, portanto, a teoria e algumas estratégias associadas foram discutidas. Utilizou-se uma estrutura multiobjetiva juntamente com o método de regularização de Tikhonov e o critério de escolha de parâmetro curva L, para a obtenção de soluções de problemas mal-postos associados. Os algoritmos formalizados no trabalho foram implementados por meio de um código desenvolvido no MATLAB. Como resultados, uma nova proposta para a função associada ao módulo de cisalhamento foi obtida e os parâmetros associados ao modelo constituído desta nova proposta foram ajustados a dados de testes reológicos. / Thixotropic materials have economically important industrial applications. Recently proposed constitutive models describing its behavior are formulated by means of a two coupled equations system: the constitutive equation (based on viscoelastic classic models) and the rate equation (in which the microstructural evolution is described). The shear modulus and the viscosity coefficient(s) are considered, in such a class of models, as functions of the structural parameter. The expressions used for such functions are defined by satisfying some asymptotic limits, in a way that the model is physically consistent. However, there is no agreement as to the form in which the expressions are formulated. It is aimed to determine the form of shear modulus function, from rheological tests data, using a modified Jeffreys model. The obtainment of an expression for the shear modulus function is defined as an inverse problem and therefore the theory and some strategies associated were discussed. It is used a multi-objective framework together with the Tikhonov regularization method and the L-curve parameter-choice criterion in order to get the solution for associated ill-posed problems. The algorithms formalized throughout the work were implemented through a MATLAB developed code. As results, a new proposal for the shear modulus function were obtained and the parameters associated with the model constituted of this new proposal are fitted to rheological tests data.

The need to protect marine structures from the high-intensity impulsive loads created by underwater explosions has stimulated renewed interest in the mechanical response of sandwich structures. The objective of this combined numerical and experimental study is to analyze the dynamic response of composite sandwich structures and develop material-structure-property relations and design criteria for improving the blast-resistance of marine structures. Configurations analyzed include polymer foam core structures with planar geometries. A novel experimental facility to generate high-intensity underwater impulsive loads and carry out in-situ measurements of dynamic deformations in marine structures is developed. Experiments are supported by fully dynamic finite-element simulations which account for the effects of fluid-structure interaction, and the constitutive and damage response of E-glass/polyester composites and PVC foams. Results indicate that the core-density has a significant influence on dynamic deformations and failure modes. Polymeric foams experience considerable rate-effects and exhibit extensive shear cracking and collapse under high-magnitude multi-axial underwater impulsive loads. In structures with identical masses, low-density foam cores consistently outperform high-density foam cores, undergoing lesser deflections and transmitting smaller impulses. Calculations reveal a significant difference between the response of air-backed and water-backed structures. Water-backed structures undergo much greater damage and consequently need to absorb a much larger amount of energy than air-backed structures. The impulses transmitted through water-backed structures have significant implications for structural design. The thickness of the facesheets is varied under the conditions of constant material properties and core dimensions. The results reveal an optimal thickness of the facesheets which maximizes energy absorption in the core and minimizes the overall deflection of the structure.

CAPES / O estudo de um componente estrutural, é mais realístico quando se admite que o componente já possua trincas. A área que estuda este fenômeno é a mecânica da fratura. O componente que possui trinca e é submetido a esforços cíclicos, tende a falhar por fadiga. Este estudo apresenta cotas que “envelopam” a solução numérica aproximada da evolução da trinca. São estimados momentos estatísticos das cotas superior e inferior, afim de se obter resultados mais realísticos com relação a propagação da trinca, visto a existência de incerteza sobre os parâmetros dos modelos de evolução da trinca. As cotas são determinadas via metodologia Fast Crack Bounds, sendo está comparada com a solução numérica aproximada obtida pelo método de Runge-Kutta de quarta ordem. A randomização dos parâmetros do modelo, é executada através de Simulação de Monte Carlo. Para a quantificação da incerteza, da cota superior, inferior e da solução numérica, são considerados exemplos “clássicos” da mecânica da fratura, onde a função de correção do fator de intensidade de tensão é conhecida: placa com largura infinita, placa com largura finita e trinca central e placa com largura finita e trinca na aresta. O trabalho apresenta os desvios relativos do primeiro e segundo momento estatístico, bem como os ganhos computacionais na resolução do problema de valor inicial que descrevem a propagação da trinca. Em todos os casos analisados, a metodologia Fast Crack Bounds apresentou menor tempo computacional, quando comparada à solução numérica do problema, sendo no mínimo 411,23% mais eficaz para o parâmetro a0 , até 8.296,29% para o parâmetro KC . / The study of a structural component is more realistic when it is admitted that the component already has cracks. The area that studies this phenomenon is the fracture mechanics. The component which is cracked and subjected to cyclic stresses tends to fail due to fatigue. This study presents upper and lower bounds that "envelop" the approximate numerical solution of the evolution of the crack. The statistical moments of the upper and lower bounds are estimated, to obtain more realistic results in relation to the crack propagation, considering the existence of uncertainty about the parameters of the evolution models of the crack. Upper and lower bounds are determined using the Fast Crack Bounds methodology, being compared to the approximate numerical solution obtained by the fourth-order RungeKutta method. The randomization of the model parameters and execution through the Monte Carlo Simulation. For the quantification of the uncertainty, the upper and lower bounds and the numerical solution, "classic" examples of fracture mechanics are considered, where the correction function of the tensile strength factor is known: Infinite width plate, finite width plate a centered crack and finite width plate a bordercracked. The work presents the relative deviations of the first and second statistical moments, as well as the computational gains in solving the initial value problem that describe the propagation of the crack. In all cases analyzed, the Fast Crack Bounds methodology presented lower computational time when compared to the numerical solution of the problem, being at least 411.23% more effective for the parameter a0 , up to 8,296.29% for the parameter KC .

This thesis focuses on the development of a new method for the continuous<p>monitoring of civil engineering structures in order to locate small damages automatically. A<p>review of the very wide literature on Structural Health Monitoring (SHM) points first out that<p>the methods can be grouped in four categories based on their need or not of a numerical model,<p>as well as their need or not of information of the damaged structure to be applied. This state<p>of the art of the SHM methods highlights the requirement to reach each levels of SHM, which<p>is in particular for the localization of small damages in civil engineering structures the needs<p>for a non-model based output-only damage sensitive feature extraction technique. The origin of<p>the local sensitivity of strains to damages is also analyzed, which justifies their use for damage<p>localization.<p>A new method based on the modal filtering technique which consists in combining linearly<p>the sensor responses in a specific way to mimic a single degree of freedom system and which<p>was previously developed for damage detection is proposed. A very large network of dynamic<p>strain sensors is deployed on the structure and split into several independent local sensor networks.<p>Low computational cost and fast signal processing techniques are coupled to statistical<p>control charts for robust and fully automated damage localization.<p>The efficiency of the method is demonstrated using time-domain simulated data on a simply<p>supported beam and a three-dimensional bridge structure. The method is able to detect and<p>locate very small damages even in the presence of noise on the measurements and variability<p>of the baseline structure if strain sensors are used. The difficulty to locate damages from acceleration<p>sensors is also clearly illustrated. The most common classical methods for damage<p>localization are applied on the simply supported beam and the results show that the modal filtering<p>technique presents much better performances for an accurate localization of small damages<p>and is easier to automate.<p>An improvement of the modal filters method referred to as adaptive modal filters is next<p>proposed in order to enhance the ability to localize small damages, as well as to follow their<p>evolution through modal filters updating. Based on this study, a new damage sensitive feature<p>is proposed and is compared with other damage sensitive features to detect the damages with<p>modal filters to demonstrate its interest. These expectations are verified numerically with the<p>three-dimensional bridge structure, and the results show that the adaptation of the modal filters<p>increases the sensitivity of local filters to damages.<p>Experimental tests have been led first to check the feasibility of modal filters to detect damages<p>when they are used with accelerometers. Two case studies are considered. The first work<p>investigates the experimental damage detection of a small aircraft wing equipped with a network<p>of 15 accelerometers, one force transducer and excited with an electro-dynamic shaker. A<p>damage is introduced by replacing inspection panels with damaged panels. A modified version<p>of the modal filtering technique is applied and compared with the damage detection based principal<p>component analysis of FRFs as well as of transmissibilities. The three approaches succeed<p>in the damage detection but we illustrate the advantage of using the modal filtering algorithm as<p>well as of the new damage sensitive feature. The second experimental application aims at detecting<p>both linear and nonlinear damage scenarios using the responses of four accelerometers<p>installed on the three-storey frame structure previously developed and studied at Los Alamos<p>National Labs. In particular, modal filters are shown to be sensitive to both types of damages,<p>but cannot make the distinction between linear and nonlinear damages.<p>Finally, the new method is tested experimentally to locate damages by considering cheap<p>piezoelectric patches (PVDF) for dynamic strain measurements. Again, two case studies are investigated.<p>The first work investigates a small clamped-free steel plate equipped with 8 PVDFs sensors, and excited with a PZT patch. A small damage is introduced at different locations by<p>fixing a stiffener. The modal filters are applied on three local filters in order to locate damage.<p>Univariate control charts allow to locate automatically all the damage positions correctly.<p>The last experimental investigation is devoted to a 3.78m long I-steel beam equipped with 20<p>PVDFs sensors and excited with an electro-dynamic shaker. Again, a small stiffener is added to<p>mimic the effect of a small damage and five local filters are defined to locate the damage. The<p>damage is correctly located for several positions, and the interest of including measurements<p>under different environmental conditions for the baseline as well as overlapping the local filters<p>is illustrated.<p>The very nice results obtained with these first experimental applications of modal filters<p>based on strains show the real interest of this very low computational cost method for outputonly<p>non-model based automated damage localization of real structures. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished