Column-supported embankments (CSE) enable accelerated construction on soft soils, high performance, and protection of adjacent facilities. The foundation columns transfer embankment and service loading to a competent stratum at depth such that loading on the soft soil can be reduced. This has the beneficial effects of reducing settlement and lateral displacement, and improving stability. Selection of column type depends on the design load, cost, constructability, etc., although unreinforced concrete columns are commonly used. A load transfer platform (LTP) is often included at the embankment base. This is a layer of coarse-grained fill that may include one or more layers of geosynthetic reinforcement. The LTP improves vertical load transfer to columns by mobilizing the shear strength of the LTP fill and the membrane effect of the geosynthetic. The geosynthetic reinforcement also responds in tension to lateral spreading. Herein, lateral spreading is defined as the lateral displacements occurring in response to lateral earth pressures in the embankment and foundation. Excessive lateral spreading can lead to bending failure of the concrete columns, tensile failure of the geosynthetic reinforcement, and instability of the system. Design procedures recommend inclusion of geosynthetic reinforcement to mitigate lateral spreading, with assumptions for the lateral thrust distribution, failure mode, and calculation of geosynthetic tensile capacity. The necessity and sufficiency of these assumptions have not been fully validated. In addition, unreinforced concrete columns have low tensile strength and can fail in bending, but recommendations for calculating column bending moments are not available. This research examines the limitations in CSE lateral spreading design with the goal of advancing fundamental understanding of lateral spreading mechanics. The research was performed using three-dimensional finite difference analyses. Limiting conditions for lateral spreading analysis were identified using case history records, and an undrained-dissipated approach was validated for the numerical analysis of limiting conditions (i.e., undrained end-of-construction and long-term excess pore pressure dissipated). The numerical model was calibrated using a well-documented case history. Additional analyses of the case history were performed to examine the lateral earth pressures in the foundation, column bending moments, and geosynthetic contribution to resisting lateral spreading. A parametric study was conducted to examine the lateral thrust distribution in 128 CSE scenarios. A refined substructure model was adopted for analyzing peak geosynthetic tensions and strains. Lastly, failure analyses were performed to examine the effect of different CSE design parameters on embankment failure height, failure mode, and deformations. The research produced qualitative and quantitative information about the following: (1) the percent thrust resistance provided by the geosynthetic as a function of its stiffness; (2) the geosynthetic contribution to ultimate and serviceability limit states; (3) the change in lateral thrust distribution throughout the embankment system before and after dissipation of excess pore water pressures; (4) the column-soil interactions involved in embankment failure; and (5) identification of two failure modes in the undrained condition. Design guidance based on these findings is provided. / Doctor of Philosophy / Column-supported embankments (CSEs) have been designated by the Federal Highway Administration as a critical technology for new highway alignment projects and widening of existing highways. CSEs enable accelerated construction and high performance in weak soils, which are factors critical to project success. In a CSE, columns are installed in the weak soil, followed by rapid construction of the soil embankment that provides the necessary elevation and foundation for the roadway. The columns transfer most of the embankment and traffic loading to a competent soil stratum at depth. Concrete without steel reinforcement is commonly used to construct the columns, although material selection depends on cost, constructability, expected load, etc. Layers of geosynthetic reinforcement can also be included at the embankment base. The geosynthetics help to transfer loads to the columns and resist excessive movement that could lead to instability. The entire embankment system should be designed for safety and economy. This research was motivated by uncertainties in design to mitigate lateral spreading. Lateral spreading refers to lateral displacements occurring in response to lateral earth pressures in the embankment and foundation. Excessive lateral spreading can lead to failure of the columns, geosynthetic reinforcement, and the entire embankment system. This research aims to advance fundamental understanding of lateral spreading in CSEs and to re-evaluate current design assumptions. Corresponding design guidance is provided.

This dissertation presents a sampling-based method that can be used for uncertainty quantification and deterministic or probabilistic optimization. The objective is to simultaneously address several difficulties faced by classical techniques based on response values and their gradients. In particular, this research addresses issues with discontinuous and binary (pass or fail) responses, and multiple failure modes. All methods in this research are developed with the aim of addressing problems that have limited data due to high cost of computation or experiment, e.g. vehicle crashworthiness, fluid-structure interaction etc.The core idea of this research is to construct an explicit boundary separating allowable and unallowable behaviors, based on classification information of responses instead of their actual values. As a result, the proposed method is naturally suited to handle discontinuities and binary states. A machine learning technique referred to as support vector machines (SVMs) is used to construct the explicit boundaries. SVM boundaries can be highly nonlinear, which allows one to use a single SVM for representing multiple failure modes.One of the major concerns in the design and uncertainty quantification communities is to reduce computational costs. To address this issue, several adaptive sampling methods have been developed as part of this dissertation. Specific sampling methods have been developed for reliability assessment, deterministic optimization, and reliability-based design optimization. Adaptive sampling allows the construction of accurate SVMs with limited samples. However, like any approximation method, construction of SVM is subject to errors. A new method to quantify the prediction error of SVMs, based on probabilistic support vector machines (PSVMs) is also developed. It is used to provide a relatively conservative probability of failure to mitigate some of the adverse effects of an inaccurate SVM. In the context of reliability assessment, the proposed method is presented for uncertainties represented by random variables as well as spatially varying random fields.In order to validate the developed methods, analytical problems with known solutions are used. In addition, the approach is applied to some application problems, such as structural impact and tolerance optimization, to demonstrate its strengths in the context of discontinuous responses and multiple failure modes.

<p>Opioid addiction has become a global epidemic and a national health crisis in recent years. In 2016, approximately 64,000 Americans under 50 years old were killed because of opioid overdoses. The aim of developing an abuse deterrent opioid is to render any form of manipulation that encourages abuse challenging and therefore, non-profitable. With this goal, the Food and Drug Administration (FDA) is extensively supporting research into the development of abuse deterrent technologies and prioritizing their production as a public health necessity. Abuse deterrent approaches include but are not limited to the following: (1) using a physical barrier (e.g., Polyethylene oxide PEO) that basically limit the release of the drugs in the blood or the digestive tract and prevent mechanical alteration of the drugs by crushing, grating, grinding, chewing etc, (2) using chemical barriers that employ gelling agents that prevent the aqueous or organic extraction of the drugs, and (3) combining the drug with an antagonist that blocks the post-abuse euphoria.</p> <p>PEO is a popular polymer used as a matrix in these complex opioid products. The polymer is responsible for the abuse deterrent properties as well as extended release behavior of opioid drugs. PEO hinders the extraction of Opioid drugs from Abuse Deterrent Formulations (ADF), makes it challenging to be injected, and resists mechanical stress and pulverization when crushed. PEO can be subjected to thermal processing such as thermal curing, compression molding, melt extrusion, and injection molding owing to its thermoplasticity.</p> <p>Assessment of the impact of using various manufacturing processes to develop ADFs and the effect of using various grades of this polymer is essential to improve upon the next generation of ADFs. There are three main categories of premarket studies: Category 1 – laboratory based (in-vitro manipulations and drug extractions), category 2 – pharmacokinetic and category 3 –clinical. These studies are required by the FDA to demonstrate that a given formulation exhibit abuse deterrent properties before a drug product is released to the market. In vitro laboratory based manipulation and extraction studies which are used to assess AD properties of these products are challenging, but essential for product development and generic abuse deterrent product approvals. It is important to realize that there is a great correlation between the laboratory based in vitro manipulation and extraction studies and the expectations of potential abuse and misuse of opioid drugs. The ability of these studies to mimic the manipulation techniques applied by abusers to defeat the abuse deterrent properties of a given formulation optimizes predictions on post-market abuse and misuse potential of ADFs. These studies should also correlate well with <i>in-vivo</i> studies since there is a direct correlation with the concentration (mg/mL in water) and the “high” obtained by an abuser. </p> <p>This research aimed to conduct laboratory based in vitro manipulation and extraction studies to investigate failure modes of PEO-based prescription opioids and Promethazine Hydrochloride PMZ HCl tablets. It highlighted the formulation components and the manufacturing parameters that might affect the dose dumping of Active Pharmaceutical Ingredients (APIs). Furthermore, this research identified model compounds that can be used as surrogates for Oxycodone and the best experimental setup that can be used to conduct smoking simulation experiments. Moreover, it provided an overview of the societal impacts of the opioid crisis in the state of Indiana.</p> <p>Investigations of the failure modes of the PEO-based prescription opioids and PMZ HCL tablets showed that physical manipulation techniques via chopping or grinding are much more effective in the destruction of the PEO matrix than thermal manipulation via the application of heat thus promoting the fast release. The factor with the most significant effect on the failure modes of PMZ HCL tablets was the application of physical manipulation, while the one with the lowest impact was the polymer grade. Moreover, producing PEO-based matrix tablets via Direct Compression DC significantly affected dose dumping behavior of the API from the drug products. The production of the PEO-based matrix tablets via DC was found to be favored over the usage of the melt extrusion method and molding techniques. It was clear that DC kept the integrity of the polymer, allowed for slow and controlled release fashion of the API, and rendered the extraction process relatively hard compared to the Hot Melt Extrusion HME and Molding techniques.</p> <p>Furthermore, the release profile of the investigated PMZ HCL products consisted of various phases of polymer swelling and API release. Thermal manipulations via the application of heat were found to accelerate the dose dumping behavior (90% release) of the APIs from the compressed, extruded, and molded PEO-based matrix formulations similarly. On the other hand, heating was much more effective in the extraction of APIs than chopping or grinding thus promoting the ability to<b> </b>draw a solution containing the API into a syringe for injection relatively easy and facilitate higher % API recovery.</p> <p>Among the formulation components that might have an impact on the AD properties of the PEO-based drug products are; the choice of the antioxidant, the use of complexing agents, chelating agents, and plasticizers. On the other hand, manufacturing process variables that might have a critical impact on AD properties of the PEO-based drug products include but are not limited to; processing temperature compared to the melting point of the polymer and time of exposure</p> <p>PMZ HCl was used as a model drug for Oxycodone in dissolution and extractability studies, while Caffeine and L-Nicotine were used as model drugs in smoking simulation experiments. The combination of the propane torch and Kugelrohr apparatus mimic the real-world scenario for smoking Opioids; however, this experimental setup caused thermal degradation rather than vaporization of some model drugs.</p> <p>According to the National Center for Health Statistics; a statistically significant increase in drug overdose death rates was reported in 2016 in the state of Indiana among other states. The number of deaths related to opioid pain relievers increased by 3732 folds in 2017 compared to the number of deaths in 2014. Moreover, Males were more affected by the opioid crisis than females. On the other hand, the age group 25-44 years, and white people were the most affected by the opioid crisis in Indiana. </p>

O desenvolvimento de métodos que permitam reproduzir numericamente o comportamento el de sistemas estruturais reais tem desa ado cientistas e pesquisadores a abordarem questões que vão além das condições de integridade do sistema. Neste sentido, uma série de novos parâmetros devem ser considerados durante um processo de otimização estrutural, zelando pela con abilidade em níveis aceitáveis enquanto que os custos esperados de falhas sejam minimizados. Aparentemente, os objetivos segurança e economia competem entre si, nesse contexto, a otimização do risco estrutural surge como uma formulação que permite equacionar este problema através do chamado custo esperado total. Nesta Tese, apresenta-se o desenvolvimento de uma técnica precisa para satisfazer os termos que compõem a função custo esperado total, isto é, uma técnica capaz de estimar a con abilidade de sistemas estruturais redundantes pela identi cação dos múltiplos modos de falha de uma maneira mais precisa que os métodos convencionais aproximados e de simulação. Além disso, análises inelásticas de estruturas de aço incluindo as não linearidades físicas e geométricas são consideradas utilizando o MCDG, além de leis constitutivas para prever o comportamento inelástico. Em posse destas informações, o algoritmo calcula o risco como sendo a probabilidade de falha multiplicada pela consequência econômica resultante desta falha. Ao risco são acrescentados os demais custos associados ao sistema estrutural, os quais não dependem dos parâmetros aleatórios do sistema e por isso denominados custos xos. Como produto, tem-se o custo esperado total, o qual corresponde a função objetivo do problema de otimização estrutural. Aplicações numéricas demonstram a precisão e e ciência da metodologia na avaliação da probabilidade de falha de problemas envolvendo funções de estado limite altamente não lineares com múltiplas regiões de falhas, assim como os efeitos causados pelas não linearidades físicas e geométricas nas análises probabil ísticas e na otimização do risco das estruturas. Os resultados demonstraram que, em relação às incertezas e consequências monetárias da falha, a estrutura ótima pode ser encontrada apenas pela formulação da otimização do risco, onde a con guração da estrutura e os limites de segurança são otimizados simultaneamente. A otimização do risco resulta numa estrutura ótima em termos mecânicos, custo esperado total e segurança. / The development of methods that allow to numerically reproduce the actual behavior of real structural systems has challenged scientists and researchers and urged them to address issues that extend beyond the integrity of a system. Therefore, a series of new parameters must be taken into account during the structural optimization process, looking for acceptable reliability levels while minimizing the expected failure costs. Apparently, safety and economy compete with each other, and in this context, structural risk optimization arises as a formulation that allows equating this problem through the so-called total expected cost. In this Thesis, an accurate technique is developed to satisfy the minimization of the total expected cost function i:e:; a technique that allows estimating the reliability of redundant structural systems by the identi cation of multiple failure modes in a more precise way than conventional approximation and simulation methods. Besides, inelastic analysis of steel trusses including both geometric and physical nonlinearities are considered using the generalized displacement control method, in addition to constitutive laws to predict the inelastic behavior. Based on this information, the algorithm calculates risk as the failure probability multiplied by the economic losses resulting from such failure. The other costs associated to the structual system are added to the actual risk and since they do not depend on system's random parameters, they are referred as xed costs. As result, one obtains the total expected cost, which corresponds to the objective function of the structural optimization problem. Numerical applications shows the accuracy and e ciency of the methodology in the evaluation of the failure probability of problems represented by high nonlinear limit state functions with multiple failure regions. It is also included the e ects of geometric and physical nonlinearities originated in the probabilistic analysis and risk optimization of truss structures. Results show that, in consideration of uncertainty and the monetary consequences of failure, the optimum structure can only be found by a risk optimization formulation, where structural con guration and safety margins are optimized simultaneously. Risk optimization yields a structure which is optimum in terms of mechanics and in terms of the compromise between cost and safety.

O desenvolvimento de métodos que permitam reproduzir numericamente o comportamento el de sistemas estruturais reais tem desa ado cientistas e pesquisadores a abordarem questões que vão além das condições de integridade do sistema. Neste sentido, uma série de novos parâmetros devem ser considerados durante um processo de otimização estrutural, zelando pela con abilidade em níveis aceitáveis enquanto que os custos esperados de falhas sejam minimizados. Aparentemente, os objetivos segurança e economia competem entre si, nesse contexto, a otimização do risco estrutural surge como uma formulação que permite equacionar este problema através do chamado custo esperado total. Nesta Tese, apresenta-se o desenvolvimento de uma técnica precisa para satisfazer os termos que compõem a função custo esperado total, isto é, uma técnica capaz de estimar a con abilidade de sistemas estruturais redundantes pela identi cação dos múltiplos modos de falha de uma maneira mais precisa que os métodos convencionais aproximados e de simulação. Além disso, análises inelásticas de estruturas de aço incluindo as não linearidades físicas e geométricas são consideradas utilizando o MCDG, além de leis constitutivas para prever o comportamento inelástico. Em posse destas informações, o algoritmo calcula o risco como sendo a probabilidade de falha multiplicada pela consequência econômica resultante desta falha. Ao risco são acrescentados os demais custos associados ao sistema estrutural, os quais não dependem dos parâmetros aleatórios do sistema e por isso denominados custos xos. Como produto, tem-se o custo esperado total, o qual corresponde a função objetivo do problema de otimização estrutural. Aplicações numéricas demonstram a precisão e e ciência da metodologia na avaliação da probabilidade de falha de problemas envolvendo funções de estado limite altamente não lineares com múltiplas regiões de falhas, assim como os efeitos causados pelas não linearidades físicas e geométricas nas análises probabil ísticas e na otimização do risco das estruturas. Os resultados demonstraram que, em relação às incertezas e consequências monetárias da falha, a estrutura ótima pode ser encontrada apenas pela formulação da otimização do risco, onde a con guração da estrutura e os limites de segurança são otimizados simultaneamente. A otimização do risco resulta numa estrutura ótima em termos mecânicos, custo esperado total e segurança. / The development of methods that allow to numerically reproduce the actual behavior of real structural systems has challenged scientists and researchers and urged them to address issues that extend beyond the integrity of a system. Therefore, a series of new parameters must be taken into account during the structural optimization process, looking for acceptable reliability levels while minimizing the expected failure costs. Apparently, safety and economy compete with each other, and in this context, structural risk optimization arises as a formulation that allows equating this problem through the so-called total expected cost. In this Thesis, an accurate technique is developed to satisfy the minimization of the total expected cost function i:e:; a technique that allows estimating the reliability of redundant structural systems by the identi cation of multiple failure modes in a more precise way than conventional approximation and simulation methods. Besides, inelastic analysis of steel trusses including both geometric and physical nonlinearities are considered using the generalized displacement control method, in addition to constitutive laws to predict the inelastic behavior. Based on this information, the algorithm calculates risk as the failure probability multiplied by the economic losses resulting from such failure. The other costs associated to the structual system are added to the actual risk and since they do not depend on system's random parameters, they are referred as xed costs. As result, one obtains the total expected cost, which corresponds to the objective function of the structural optimization problem. Numerical applications shows the accuracy and e ciency of the methodology in the evaluation of the failure probability of problems represented by high nonlinear limit state functions with multiple failure regions. It is also included the e ects of geometric and physical nonlinearities originated in the probabilistic analysis and risk optimization of truss structures. Results show that, in consideration of uncertainty and the monetary consequences of failure, the optimum structure can only be found by a risk optimization formulation, where structural con guration and safety margins are optimized simultaneously. Risk optimization yields a structure which is optimum in terms of mechanics and in terms of the compromise between cost and safety.

L’industrie des transports aéronautique et terrestre voit une augmentation constante de l’électrification de ses fonctions. Les actionneurs mécaniques ou hydrauliques sont au fil des évolutions technologiques remplacés par des actionneurs électriques.Les composants qui dominent le marché actuellement ne semblent plus capables de suivre la tendance. En effet, les composants de puissance à base de silicium règnent toujours sur le marché actuel, grâce à leur faible coût. Ce matériau commence par contre à atteindre ses limites théoriques en termes de performances. Dans ce contexte, différentes structures en semi-conducteurs à large bande interdite sont en train d’émerger afin de succéder au silicium.Cette étude a pour objectif d’évaluer la fiabilité des transistors de puissance à base de Nitrure de Gallium. Ces composants semblent être très prometteurs pour des applications moyennes puissances. Cependant, les mécanismes de défaillance dont peuvent souffrir ces composants ne sont pas encore suffisamment étudiés. L’étude consiste en l’application de vieillissements alliant contraintes thermiques et électriques. Ces vieillissements sont effectués à différentes conditions de tension et de température. L’objectif de cette méthode est, dans un premier temps, d’isoler l’effet de chaque facteur de stress sur l’état des composants, et dans un second temps, d’identifier les mécanismes de défaillances activés en fonction des conditions de vieillissement.Ce travail a permis d’identifier l’existence de différents mécanismes de défaillance pouvant être activés selon les conditions de vieillissement. En effet, il est apparu que la gamme de température de vieillissement utilisée influe grandement sur la prédominance des mécanismes de défaillance activés. Les résultats obtenus remettent en question les normes de qualification actuellement appliquées aux composants en Nitrure de Gallium. Ces normes devraient revoir à la hausse les températures de vieillissement utilisées afin de couvrir des gammes plus proches des températures d’utilisation pour ce type de composants. / The aeronautical and terrestrial transport industries know a steady increase in the electrification of their functions. In fact, the mechanical or hydraulic actuators are gradually replaced by electric ones.The components dominating the market today seem unable to follow the trend anymore. In fact, silicon-based power components still prevail in the current market, thanks to their low cost. However, this material begins to reach its theoretical limits in terms of performance. In this context, different wide bandgap semiconductor structures are emerging to take on from silicon.The aim of this study is to assess the reliability of power transistors based on Gallium Nitride. These components are very promising for medium power applications. However, the failure mechanisms of these components are not yet sufficiently studied. The study consists in the application of aging tests combining thermal and electrical stresses. These agings are carried out under different conditions of tension and temperature. The objective of this method is, firstly, to isolate the effect of each stressor on the state of the components, and secondly, to identify the failure mechanisms activated according to the aging conditions.This work made it possible to identify the existence of different failure mechanisms that can be activated according to the aging conditions. Indeed, it has emerged that the aging temperature range used influences the predominance of activated failure mechanisms. The results challenge the adequacy of current qualification standards for Gallium Nitride components. These standards should revise upwards the aging temperatures used to cover ranges closer to the operating temperatures of this kind of components.

Made available in DSpace on 2016-12-08T17:19:38Z (GMT). No. of bitstreams: 1 CAPA - SUMARIO.pdf: 307847 bytes, checksum: 431911d5eb22b3080df0c9228cba482d (MD5) Previous issue date: 2011-07-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / With increasing competition and reliability in modern equipment, is of vital importance to studies on the structural fatigue life in the time domain and frequency domain, the advance speed and cost reduction in computing greatly simplifies the prediction fatigue life. But it is also necessary to know the modes of fatigue crack growth, and its mechanisms. After a review appropriate to implement a program to estimate the loss of torsional stiffness in a specimen with a crack in Mode II, was developed a device torsional fatigue tests in an attempt to identify the onset of macroscopic crack and study its growth. With learning the operation of the drive system, the system data acquisition device testing, and design and construction of an electronic circuit to shutdown device tests at the beginning of the crack in the specimen, an adequate control and calibration of these components was performed, generating confidence in that joint. This document is the experimentally determined curve of fatigue specimens of SAE 1020 Steel, Grey Cast Iron and Aluminum 6063. They also studied the failure modes of cracks from the beginning until the final rupture of the specimen to verify the failure modes, the crack initiation and propagation of macroscopic materials. It still made a numerical prediction by finite element method the size of the crack. / Com aumento da competitividade e da confiabilidade em equipamentos modernos, é de vital importância os estudos em relação à vida em fadiga estrutural no domínio do tempo e no domínio da freqüência, o avanço da velocidade e redução de custos na área computacional facilita bastante a previsão de vida em fadiga. Mas é necessário também conhecer os modos de crescimento de trinca por fadiga, e seus mecanismos. Após revisão bibliográfica adequada, com implementação de um programa para estimar a perda de rigidez torsional em um corpo de prova com trinca no Modo II, foi desenvolvido um dispositivo de testes de fadiga torsional na tentativa de identificar o início da trinca macroscópica e estudar o seu crescimento. Com o aprendizado do funcionamento do sistema de acionamento, do sistema de aquisição de dados do dispositivo de testes, e a elaboração e construção de um circuito eletrônico para desligamento do dispositivo de testes ao início da trinca no corpo de prova, um adequado controle e calibração destes componentes executou-se, gerando confiabilidade nestes conjuntos. Neste documento é determinada experimentalmente a curva de fadiga de corpos de prova de Aço SAE 1020, Ferro Fundido Cinzento e Alumínio 6063. Também são estudados os modos de falha desde o início da trinca até a ruptura final do corpo de prova para verificar os modos de falha, início e propagação da trinca macroscópica desses materiais. É feita ainda uma previsão numérica através do Método dos Elementos Finitos do tamanho da trinca.

O desenvolvimento de métodos que permitam reproduzir numericamente o comportamento el de sistemas estruturais reais tem desa ado cientistas e pesquisadores a abordarem questões que vão além das condições de integridade do sistema. Neste sentido, uma série de novos parâmetros devem ser considerados durante um processo de otimização estrutural, zelando pela con abilidade em níveis aceitáveis enquanto que os custos esperados de falhas sejam minimizados. Aparentemente, os objetivos segurança e economia competem entre si, nesse contexto, a otimização do risco estrutural surge como uma formulação que permite equacionar este problema através do chamado custo esperado total. Nesta Tese, apresenta-se o desenvolvimento de uma técnica precisa para satisfazer os termos que compõem a função custo esperado total, isto é, uma técnica capaz de estimar a con abilidade de sistemas estruturais redundantes pela identi cação dos múltiplos modos de falha de uma maneira mais precisa que os métodos convencionais aproximados e de simulação. Além disso, análises inelásticas de estruturas de aço incluindo as não linearidades físicas e geométricas são consideradas utilizando o MCDG, além de leis constitutivas para prever o comportamento inelástico. Em posse destas informações, o algoritmo calcula o risco como sendo a probabilidade de falha multiplicada pela consequência econômica resultante desta falha. Ao risco são acrescentados os demais custos associados ao sistema estrutural, os quais não dependem dos parâmetros aleatórios do sistema e por isso denominados custos xos. Como produto, tem-se o custo esperado total, o qual corresponde a função objetivo do problema de otimização estrutural. Aplicações numéricas demonstram a precisão e e ciência da metodologia na avaliação da probabilidade de falha de problemas envolvendo funções de estado limite altamente não lineares com múltiplas regiões de falhas, assim como os efeitos causados pelas não linearidades físicas e geométricas nas análises probabil ísticas e na otimização do risco das estruturas. Os resultados demonstraram que, em relação às incertezas e consequências monetárias da falha, a estrutura ótima pode ser encontrada apenas pela formulação da otimização do risco, onde a con guração da estrutura e os limites de segurança são otimizados simultaneamente. A otimização do risco resulta numa estrutura ótima em termos mecânicos, custo esperado total e segurança. / The development of methods that allow to numerically reproduce the actual behavior of real structural systems has challenged scientists and researchers and urged them to address issues that extend beyond the integrity of a system. Therefore, a series of new parameters must be taken into account during the structural optimization process, looking for acceptable reliability levels while minimizing the expected failure costs. Apparently, safety and economy compete with each other, and in this context, structural risk optimization arises as a formulation that allows equating this problem through the so-called total expected cost. In this Thesis, an accurate technique is developed to satisfy the minimization of the total expected cost function i:e:; a technique that allows estimating the reliability of redundant structural systems by the identi cation of multiple failure modes in a more precise way than conventional approximation and simulation methods. Besides, inelastic analysis of steel trusses including both geometric and physical nonlinearities are considered using the generalized displacement control method, in addition to constitutive laws to predict the inelastic behavior. Based on this information, the algorithm calculates risk as the failure probability multiplied by the economic losses resulting from such failure. The other costs associated to the structual system are added to the actual risk and since they do not depend on system's random parameters, they are referred as xed costs. As result, one obtains the total expected cost, which corresponds to the objective function of the structural optimization problem. Numerical applications shows the accuracy and e ciency of the methodology in the evaluation of the failure probability of problems represented by high nonlinear limit state functions with multiple failure regions. It is also included the e ects of geometric and physical nonlinearities originated in the probabilistic analysis and risk optimization of truss structures. Results show that, in consideration of uncertainty and the monetary consequences of failure, the optimum structure can only be found by a risk optimization formulation, where structural con guration and safety margins are optimized simultaneously. Risk optimization yields a structure which is optimum in terms of mechanics and in terms of the compromise between cost and safety.

Les transistors IGBT sont les composants semi-conducteurs de puissance les plus couramment utilisés dans les fonctions de l'électronique de puissance. Ils sont fréquemment assemblés dans des modules contenant plusieurs puces et réalisant l'interconnexion à la fois électrique et thermique avec l'environnement. Dans de nombreuses applications, ces modules sont soumis à un cyclage thermique actif généré par les variations du régime de fonctionnement du système. Ceci est à l'origine de différentes dégradations d'origine thermomécaniques pouvant mener à la défaillance. Les tests de vieillissement sont l'un des moyens permettant de mieux comprendre les mécanismes de dégradations en imposant des conditions de stress connues mais également de construire des modèles de fiabilité comportementaux utiles dans l'estimation de durée de vie des systèmes.Le présent travail décrit la mise en œuvre d'une méthode de test originale destinée à des modules IGBT de la gamme 600V-200A et basée sur l'utilisation d'onduleurs fonctionnant en modulation de largeur d'impulsion. Elle exploite la variation de température de jonction provoquée par cette modulation pour générer un cyclage thermique d'une fréquence très supérieure (2Hz à 10Hz) à celles habituellement utilisées pour ce type de test. Ceci permet de réduire considérablement la durée de ces tests et d'accéder à des gammes basses de l'amplitude des cycles thermiques.Les essais de vieillissement imposent de suivre l'évolution de paramètres indicateurs afin d'estimer régulièrement l'état des composants testés et d'appliquer des conditions d'arrêt. Le paramètre utilisé dans le présent travail est la tension VCE des IGBT, qui est un bon révélateur de l'état des fils de bondings, maillons faibles de ces modules. En complément du développement de la méthode de cyclage proprement-dite, un dispositif de suivi automatique du paramètre VCE a été développé afin de limiter la durée des phases de caractérisations correspondantes. Ce dispositif permet également de mesurer la température de jonction de façon indirecte et de reconstruire le profil de température dynamique pendant le cyclage.Cet ensemble a permis d'obtenir des résultats exploitables sur une trentaine d'échantillons avec des amplitudes de cycle comprises entre 30°C et 50°C. Ils mettent en évidence un seul type de dégradation, la fissuration des attaches entre les fils de bonding et la métallisation d'émetteur avec, dans certains cas, le décollement complet du fil (lift-off). Des essais à différentes fréquences de cyclage pour la même amplitude ont été réalisés sur un groupe de ces échantillons. Si le nombre d'échantillons consacrés à comparaison n'est pas encore réellement suffisant, les résultats obtenus sont similaires et semblent donc démontrer que la fréquence de test n'impacte pas le mode de vieillissement dans la gamme de température étudiée. Cette observation est une première validation de la pertinence de la méthode proposée qui permet de réduire d'un facteur cinq à dix les durées de test. / IGBT transistors are the most used power semiconductor devices in power electronics and are often integrated in power modules to constitute basic switching functions. In various applications, IGBT power modules suffer thermal cycling (or power cycling) due to variations of operating conditions. This power cycling induces thermo mechanical stress that can lead to damages and then, to failures. Ageing tests are a means to identify and analyze the degradation mechanisms due to power cycling by imposing calibrated test conditions. In addition, their results can be used to establish empiric lifetime models that are useful for power converter designers.The present work describes the implementation of an ageing test method dedicated to IGBT modules operating in the 600V-200A range. This method takes advantage of particular operating conditions generated by pulse width modulation inverters in which the IGBT modules to be tested are introduced. The modulation induces a variation of IGBT die temperature, i.e. a power cycling, of which the frequency is significantly higher (2Hz to 10Hz) than the operating frequencies of classical test systems. By using this technique, the test length is reduced while low values of thermal amplitude can be reached.Throughout the ageing tests, the monitoring of ageing indicators is required to evaluate the sample health and to stop the operation when predefined conditions are reached. In the present work, the ageing indicator is the on-state voltage VCE across the IGBT device that is relevant in regard with wire bond degradations. Therefore, as a complement of the fast test method, an automated VCE monitoring system has been developed in order to fully take benefit of the high test-speed. In addition, this system is able to measure the junction temperature and to provide the temperature profile during the power cycling.This test bench has made possible the ageing process of three dozen of samples by applying thermal swing amplitudes in the 30°C-50°C range, that is not reachable with classical test benches operating in low frequency because of the unacceptable test length. The results show that only one kind of damage is generated by the present test conditions, i.e. the degradation of attaches between the emitter metallization and the wire bonds. In many case, complete lift-off have been observed. Some samples have been used to evaluate the influence of thermal swing frequency on the results. The latter are unchanged when the frequency varies between 2Hz and 0.2Hz, therefore it is a first validation of the fast test relevance.

Les composants de puissance à base de GaN sont connus par l’instabilité de leurs caractéristiques électriques, en particulier la tension de seuil et la résistance à l'état passant. Cela est dû aux effets des mécanismes de piégeage/de-piégeage des charges dans la structure. Le travail présenté dans ce mémoire se compose de deux grandes parties. Dans un premier temps, nous mettons en évidence l’effet d’un certain nombre de paramètres de commutation sur l'évolution de la résistance dynamique avec des cycles de commutation successifs. Nous analysons, en particulier, l’effet de la tension au blocage, la fréquence de commutation et la température sur l'évolution de la résistance dynamique. Dans un deuxième temps, nous présentons les résultats des essais de cyclage de puissance effectués en utilisant 80K de variation de température de jonction sur des puces de puissance MOS-HEMT Al2O3/AlGaN/GaN Normally-On. Ensuite, nous réalisons des caractérisations de pièges, basées sur des analyses de mesures de courants transitoires pendant le processus de vieillissement. Les résultats montrent qu’une dégradation irréversible affecte la tension de seuil avec une dérive vers des valeurs négatives. Ces dérives sont principalement attribuables au piégeage cumulatif avec des cycles de puissance, probablement induits par des électrons chauds, d’une manière progressive et non récupérable. / Power components based on GaN are known by the instability of their electrical characteristics, in particular the threshold voltage and the on-state resistance. This is due to the effects of trapping/de-trapping mechanisms in the structure. The work presented in this memoir consists of two main parts. At the first step, we highlight the effect of a number of switching parameters on the evolution of the dynamic resistance with successive switching cycles. In particular, we analyze the effect of blocking voltage, switching frequency and temperature on the evolution of dynamic resistance. In a second step, we present the results of power cycling tests performed using 80K of junction temperature swing on Normally-ON Al2O3/AlGaN/GaN MOS-HEMTs. Then, we perform trap characterizations, based on the analyses of transient current measurements, during the aging process. The results show that irreversible degradation affects threshold voltage with drift to negative values. These drifts were mainly attributed to cumulative trapping with power cycles, probably induced by hot electrons, in a progressive and non-recoverable way.