A comprehensive stress model of the filament winding fabrication process, previously implemented in the finite element program, WACSAFE, was implemented using the ABAQUS finite element software package. This new implementation, referred to as the ABWACSAFE procedure, consists of the ABAQUS software and a pre/postprocessing routine that was developed to prepare necessary ABAQUS input files and process ABAQUS displacement results for stress and strain computation. The ABWACSAFE procedure has a structure and general flow pattern similar to the WACSAFE program. ABAQUS now performs the fundamental finite element procedures needed in the fabrication stress model. The ABWACSAFE pre/postprocessing routine utilizes many subroutines from the WACSAFE program. Some subroutines are used in their original form while many were significantly modified. New subroutines have been written as well. / Master of Science

This research explores the nearly untapped research area of the analysis of fracture mechanics in residual stress zones. This type of research has become more prevalent in the field in recent years due to the increase in prominence of residual stress producing processes. Such processes include additive manufacturing of metals and installation procedures that lead to loads outside the anticipated standard operating load envelope. Abaqus was used to generate models that iteratively advanced toward solving this problem using the compact tensile specimen geometry. The first model developed in this study is a two-dimensional fracture model which then led to the development of an improved three-dimensional fracture model. Both models used linear elastic fracture mechanics to determine the stress intensity factor (K) value. These two models were verified using closed-form equations from linear elastic fracture mechanics. The results of these two models validate the modeling techniques used for future model iterations. The final objective of this research is to develop an elastic-plastic fracture mechanics model. The first step in the development of an elastic-plastic fracture model is a three-dimensional quasi-static model that creates the global macroscale displacement field for the entire specimen geometry. The global model was then used to create a fracture submodel. The submodel utilized the displacement field to reduce the model volume, which allowed a higher mesh density to be applied to the part. The higher mesh density allowed more elements to be allocated to accurately represent the model behavior in the area local to the singularity. The techniques used to create this model were validated either by the linear elastic models or by supplementary dog bone prototype models. The prototype models were run to test model results, such as plastic stress-strain behavior, that were unable to be tested by just the linear elastic models. The elastic-plastic fracture mechanics global quasi-static model was verified using the plastic zone estimate and the fracture submodel resulted in a J-integral value. The two-dimensional linear elastic model was validated within 6% and the three-dimensional linear elastic model was validated within 0.57% of the closed-form solution for linear elastic fracture mechanics. These results validated the modeling techniques. The elastic-plastic fracture mechanics quasi-static global model formed a residual stress zone using a Load-Unload-Reload load sequence. The quasi-static global model had a plastic zone with only a 0.02-inch variation from the analytical estimate of the plastic zone diameter. The quasi-static global model was also verified to exceed the limits of linear elastic fracture mechanics due to the size of the plastic zone in relation to the size of the compact specimen geometry. The difference between the three-dimensional linear elastic fracture model J-integral and the elastic-plastic fracture submodel initial loading J-integral was 3.75%. The J-integral for the reload step was 18% larger than the J-integral for the initial loading step in the elastic-plastic fracture submodel. / Master of Science / Additive manufacturing, sometimes referred to as 3-D printing, has become an area of rapid innovation. Additive manufacturing methods have many benefits such as the ability to produce complex geometries with a single process and a reduction in the amount of waste material. However, a problem with these processes is that very few methods have been created to analyze the initial part stresses caused by the processes used to additive manufacture. Finite element methods are computer-based analyses that can determine the behavior of parts based off prescribed properties, shape, and loading conditions. This research utilizes a standard fracture determination shape to leverage finite element methods. The models determine when a crack will form in a part that has process stresses from additive manufacturing. The model for crack initiation was first developed in two dimensions, neglecting the thickness of the part, using a basic material property definition. The same basic material property definition was next used to develop a crack initiation model in three dimensions. Then a more advanced material property definition was used to capture the impact of additive manufacturing on material properties. This material property definition was first used to establish the part properties as it relates to part weakening due to additive manufacturing. A higher accuracy model of just the crack development area was produced to determine the crack initiation properties of the additive manufactured part. Methods previously confirmed by testing were used to validate the models produced in this research. The models demonstrated that under the same loading parts with initial processes stresses were closer to fracture than parts without initial stresses.

The ability of a Formula SAE sports car to negotiate a turn in a race is influenced by many parameters which include car's overall geometry, its shape, weight distribution, type of suspension used, spring and shock absorber characteristics that are used in the tire properties, static and dynamic loading. Steady-state cornering implies that the forces acting on the vehicle are unchanging for a given time. The suspension uprights form a connection between the wheel assembly and the suspension linkages. The criticality of the upright is that it is considered an un-suspended body, but in fact, it is subjected to very high stresses. The dynamic load imposed on the vehicle from various road conditions, cornering, braking and suspension assembly constraints generate stress on the upright body. The equations of motion generally govern vehicle dynamics. For a kinematic and rigid body dynamics analysis, a multibody dynamics (MBD) approach is popular. The results of the dynamic analysis yield internal loads which are used to analyze suspension components for structural stiffness and strength. Automotive companies with relatively lower structural loads have made the MBD approach popular because it is supposed to be computationally less expensive. Elastodynamics is an alternative approach to solving dynamics equations while considering the components to be elastic. This approach can capture the inertial and elastic responses of the components and the load path with varying positions of the components in a mechanism. In this research, a quarter-car suspension is modeled in a finite element code (Abaqus®), focusing on the vehicle upright but still modeling the connections and interactions of the quarter-car suspension system of a FSAE vehicle. The BEAM element modeling used for the suspension members captures the bending response. The overall model is created by making computationally conscious decisions, debugging and refining the interactions and connections to be representative. The modeling technique to create elastodynamic models is explored and established with a versatile set of suspension components and interactions providing a good experience with finite element modeling. The models are created with incremental steps and early steps are verified with hand calculations. A further vehicle verification and validation plan is the next immediate priority to gain confidence in the model for accurate simulations which can be used to predict accurate structural and dynamic results. With extending the model capabilities and computational capabilities, a quarter-car suspension model is powerful enough to run the entire track simulations for formula races and even durability load cases for commercial vehicles. Fatigue loading and abusive test cases would be the load cases to investigate possible failure modes. The quarter-car suspension model is a framework with different interactions, connections, components, boundary conditions and loads that are representative for different suspension configurations in different vehicles. The best practices of this modeling exercise are established and scalability to defeature or add details while preserving the connection behavior is achieved. / Master of Science

A dynamic simulation of the response of an inflatable dam subjected to a flood was carried out to determine the survivability envelope of the dam where it can operate without rupture, or overflow. A fully nonlinear free-surface flow was applied in two dimensions using a mixed Eulerian-Lagrangian formulation. An ABAQUS finite element model was used to determine the dynamic structural response of the dam. The problem was solved in the time domain which allows the prediction of a number of transient phenomena such as the generation of upstream advancing waves, and dynamic structural collapse. Stresses in the dam material were monitored to determine when rupture occurs. An iterative study was performed to find the service envelope of the dam in terms of the internal pressure and the flood Froude number for two flood depths. It was found that the driving parameter governing failure of the dam was the internal pressure. If this pressure is too low, the dam overflows; if this pressure is too high, the dam ruptures. The fully nonlinear free-surface flow over a semi-circular bottom obstruction was studied numerically in two dimensions using a similar solution formulation as that used in the previous study. A parametric study was performed for a range of values of the depth-based Froude number up to 2.5 and non-dimensional obstacle heights up to 0.9. When wave breaking does not occur, three distinct flow regimes were identified: subcritical, transcritical and supercritical. When breaking occurs it may be of any type: spilling, plunging or surging. In addition, for values of the Froude number close to 1, the upstream solitary waves break. A systematic study was undertaken, to define the boundaries of each type of breaking and non-breaking pattern, and to determine the drag and lift coefficients, free surface profile characteristics and transient behavior. / Master of Science

[pt] A estabilidade de poços através de zonas de sal é um aspecto relevante em ambientes de perfuração offshore no Brasil. O fluxo convencional no planejamento de um poço de petróleo não reconhece a natureza complexa do estado de tensões in-situ em torno destes corpos de sal. Portanto, é necessária uma avaliação fiável das tensões in-situ considerando tanto a escala de campo (global) quanto as principais estruturas presentes no overburden. Neste trabalho, a análise de estabilidade de poços é realizada em três etapas. Primeiro, é realizada uma análise numérica a escala global para avaliar as tensões in-situ considerando a geometria de um corpo de sal. A seguir, são introduzidas as tensões in-situ em um modelo local, chamado subestrutura, através de duas técnicas de transferência de tensões propostas, denominadas as técnicas do Inverso Ponderado da Distância (IPD) e do Gradiente de Tensões (GT). O termo subestrutura é definido como uma linha curva no espaço composta por um conjunto de pontos, se assemelhando a uma seção ou trajetória completa de um poço. Finalmente, a janela operacional do poço é calculada acoplando os resultados de tensões da modelagem numérica com equações elásticas. Neste trabalho as técnicas IPD e GT são também utilizadas para transferir tensões em submodelos localizados dentro de um modelo global, visando realizar futuros estudos de submodelagem de estabilidade de poços. O termo submodelo consiste em uma malha de elementos finitos de um tamanho menor e um refinamento maior em relação ao modelo global. / [en] Wellbore Stability drilling through salt zones is an important current endeavor in many areas offshore of Brazil. The conventional well design workflow does not recognize the complex nature of the stress field near these salt bodies. Therefore, a reliable assessment of the in-situ stresses must be carried out considering a field (global) scale of the problem and the presence of major structures in the overburden. The proposed stability analysis is carried out in three stages. Firstly, a global finite element analysis is employed to evaluate the in-situ stresses at a global scale considering the geometry of a salt body. Secondly, the global scale in-situ stresses are introduced in a local model, that we call substructure, by using two proposed stress transfer techniques called the Inverse Distance Weighted Technique (IDWT) and the Stress Gradient Technique (SGT). We define Substructure as a set of points forming a section or a complete trajectory of an oil well. Finally, optimal mud weights are calculated combining numerical stress results with analytical elastic equations. These two stress transfer techniques are also proposed to be used to transfer stresses to submodels inside a global model domain for submodeling wellbore stability purposes. The term submodel is defined as a finite element mesh with a smaller size relative to the size of the global model.

During the insert process when the windings of an electric motor are pushed into the stator slots, some detrimental phenomena can occur that affect the efficiency and life of the motor. To detect these phenomena and optimize the process, a simulation would be useful. An investigation of the possibility to perform a simulation, using an appropriate numerical method for the insert process of distributed windings in a permanent magnet synchronous motor, was performed. During the project, a literature study was carried out to investigate existing methods and key-parameters for the simulation of the process. Explicit finite element method has been shown to be a suitable numerical method for simulating another winding process. An explicit finite element analysis was performed with a simplified model of the stator, the wires, the transmission tool and the needles by using the software Abaqus/CAE. In order to reduce the computational time, beam elements were used to model the wires and the other parts as rigid bodies. The model accounted for example contact and provided numerical results. The result was a suitable model. However, it needs to be developed further.

O ensaio SPT (Standard Penetration Test), é o ensaio de campo geotécnico mais usado no Brasil e grande parte do mundo. A ampla utilização deste ensaio no âmbito da engenharia geotécnica deve-se à sua simplicidade, baixo custo, grande experiência prática e facilidade de aplicação dos seus resultados. Os principais objetivos do ensaio são a determinação do índice de resistência do solo (NSPT) e amostragem. Apesar de ser muito utilizado na estimativa da capacidade de carga de fundações, o ensaio SPT tem sido questionado pelo fato de que o índice NSPT é utilizado em correlações empíricas baseadas em observações práticas, sem nenhum fundamento científico. Além do mais, seus resultados apresentam significativas dispersões. Para analisar racionalmente os resultados do ensaio e de possibilitar a comparação dos diferentes resultados de distintas equipes, é necessário conhecer as quantidades de energia envolvidas na penetração do amostrador no solo. Tais análises requerem o conhecimento da eficiência do equipamento (η) e a força de reação dinâmica do solo à cravação do amostrador (RD). Neste cenário, o presente trabalho aborda a interpretação racional dos resultados de ensaios SPT a partir de simulações numéricas realizadas com o software Abaqus/Explicit®. Esse programa fornece os deslocamentos do amostrador (Δρ), força de reação dinâmica do solo (RD), e as forças de reação nas paredes laterais externa e interna do amostrador (R1 e R2). Baseando-se nos resultados dos modelos numéricos, foi possível calcular a eficiência do equipamento, a partir da força de reação dinâmica do solo, as resistências unitárias de atrito atuantes nas paredes e a resistência unitária na ponta do amostrador. Também foi possível determinar a relação entre as resistências unitárias de atrito desenvolvidas no interior e exterior do amostrador (fator de atrito a). Os resultados obtidos foram comparados com resultados de trabalhos experimentais e valores teóricos determinados com base no Princípio de Hamilton da conservação da energia. Também foi possível simular uma prova de carga dinâmica com energia crescente no amostrador, variando a altura de queda do martelo. Isso confirmou que a resistência mobilizada do solo para certa energia aplicada pode estar bem abaixo da ruptura e apenas representar um ponto na curva de resistência mobilizada versus deslocamento. / The Standard Penetration Test (SPT) is one of the most used geotechnical tests in the world. The wide use of this test in the context of geotechnical engineering is due to its simplicity, low cost, large practical experience and its ease of application of results. The main objectives of the test are the determination of soil resistance index (NSPT) and sampling. Despite being widely used in estimating the bearing capacity of foundations, the SPT test has been questioned by the fact that the NSPT index is used in empirical correlations based on practical observations, with no scientific basis. Furthermore, its results show significant dispersions. To analyze rationally the test results and to make possible to compare different results obtained from different equipment, it is necessary to know the amounts of energy existing during the penetration of the sampler into soil. Such analyses require information about the equipment efficiency (η) and the dynamic soil reaction force during the sampler penetration (RD). In this scenario, this work addresses the rational interpretation of SPT test results from numerical simulations performed with the Abaqus/Explicit software. This program provides the sampler displacements (Δρ), the dynamic soil reaction force (RD) and the external and internal reaction forces acting on the sampler walls (R1 e R2). Based on the results of the numerical models, it was possible to calculate the efficiency of the equipment, from the dynamic soil reaction force, the unit friction resistance acting on the sampler walls and the unit resistance at the sampler tip. In addition, it was possible to estimate the relationship between the unit friction resistance acting on the internal and external walls of the sampler (friction factor a). The numerical results were compared with experimental results and theoretical values, obtained using the Hamilton\'s principle of conservation of energy. Furthermore, it was possible to simulate a dynamic load test with increasing energy applied to the sampler, by varying the height of fall of the hammer. It was obtained the confirmation that resistance mobilized for a certain level of energy applied to the sampler may be below the failure load and represent only a point on the graph curve mobilized resistance versus displacement.

O estudo do comportamento de estruturas em situação de incêndio tem se tornado cada vez mais importante devido às graves consequências de incêndios que têm ocorrido no Brasil e em outros países. Por esse motivo, este trabalho tem como objetivo estudar o comportamento residual do concreto com pérolas de EPS sob altas temperaturas, pois se pretende utilizar esse concreto em painéis e em lajes pré-moldadas para a construção de edifícios de pequeno e de médio porte, por ele ser um concreto muito leve, com massa específica em torno de 1170 kg/m3, um pouco menos da metade da relativa ao concreto convencional. Para este estudo, foram utilizados corpos de prova cilíndricos e prismáticos, seguindo recomendações nacionais e internacionais. As análises foram realizadas para 200ºC, 400ºC e 600ºC, e foram avaliadas: massa específica, resistência à compressão, módulo de elasticidade estático, módulo de elasticidade dinâmico, resistência à tração por compressão diametral, resistência à tração na flexão e fator de tenacidade. Para esses dois últimos, foram utilizadas fibras de aço nas seguintes taxas: 0,3%, 0,6% e 0,9%. Os resultados foram comparados com os de concreto com EPS em temperatura ambiente e com os de concreto convencional, que já tem alguns resultados disponíveis na literatura técnica. Também foi realizada uma análise térmica numérica, utilizando o software ABAQUS 14, para calibrar duas propriedades: calor específico e condutividade térmica. Para isso, foram utilizados resultados obtidos na análise experimental. Os resultados numéricos e experimentais foram coerentes com os esperados. Os experimentais apresentaram redução de valor à medida que a temperatura aumentava, e comportamento pior em comparação ao concreto convencional. A adição de fibras aumentou a resistência à tração na flexão e também a tenacidade, tanto em temperatura ambiente quanto em temperaturas elevadas. / The study of the structures behavior in fire has become increasingly important due to the serious consequences of fires that have occurred in Brazil and in other countries. Therefore, this work aims to study the residual behavior of concrete with EPS beads under high temperatures, because it is intended to use it in panels and precast slabs for the construction of small and medium-sized buildings, for it is a very light concrete, with a density around 1170 kg/m3, slightly less than half that of conventional concrete. For this study, cylindrical and prismatic specimens were used, following national and international standards. The analyses were carried out to 200°C, 400°C and 600°C, and were evaluated: density, compression strength, static modulus of elasticity, dynamic modulus of elasticity, splitting tensile strength, flexural tensile strength, and toughness factor. For the latter two, steel fibers were used at the following rates: 0.3%, 0.6% and 0.9%. The results were compared with those of concrete with EPS at room temperature, and with conventional concrete which already has some results available in the technical literature. A numerical thermal analysis was also performed, using ABAQUS 14 software, to calibrate two properties: specific heat and thermal conductivity. For this, results obtained in the experimental analysis were used. The numerical and experimental results were consistent with those expected. The experimental results showed reduction of value as temperature increased, and worst behavior in comparison to ordinary concrete. The addition of fibers increased tensile strength in bending and also the toughness, both at room temperature and at elevated temperatures.

Advanced technologies for the automotive industry require improved designed precision in different areas. Research is needed in order to meet customer demand and satisfaction to increase durability, efficiency, and reliability. That is why continuous development in transmission system has been an exciting topic for many years. The gears in the transmission system demand high resistance against repeated loads acting on the teeth and the ability to engage without energy loss. The intention is to support that effort by investigating seven design cases of a parallel helical gear pair. This master thesis aims to study a gear design concept of adding a supporting ring to reduce the root bending- and contact stresses. To implement this study, seven different design cases were modelled to study the effect of changing the design. One or two support ring structures were added, or the thickness was increased of the gear considered exposed to high stresses. The purpose is to gain a comparative advantage in the automotive industry. M1 is a reference standard design, while the other models (M2-M5/P1-P3) are modified designs with additional rings or changed in thickness. Simulation is an effective and an useful tool to understand and visualize how the complex interaction of the transmission component appears to be. A finite element method (FEM) program was used to investigate these models. The gear pairs were imported into the pre-processor ANSA, the FEM program Abaqus 2017 was used as a solver, and the results were extracted from the post-processor META. To support the aim of this thesis, two of the seven FEM models were validated against a specialized gear calculated program, WindowsLDP, in order to determine the robustness of the simulation models. The transmission error (TE) measurements, root bending- and contact stresses were observed for the validation. Introducing the different models M3-M5 and P1-P3, the root and contact stresses were reduced by 1.2-4.4 and 0.07-4.3 %, respectively, compared to M1. The transmission error TE could differ as much as 85% in M2-M3 as compared to M1. Systematic errors were generated in model M2, therefore the low root stresses obtained in M2 should be considered carefully. Implementing the so-called misalignment measurement, tilting parameter, microgeometry, and profile modification related to crowning and tip relief is discussed and believed to reduce TE measurements, root- and contact stresses. These modifications have not been studied in this thesis. The LDP results showed a trend of higher values compared to the FEM results, which was suggested to be further investigated in the future. / Dagens avancerade teknik inom fordonsindustrin kräver förbättrad precision hos konstruktioner inom olika områden. Forskning behövs för att möta kundernas efterfrågan och för att öka hållbarhet, effektivitet och tillförlitlighet. Detta är varför kontinuerlig utvecklingen av växellådssystem har varit ett hett ämne i många år. Kugghjulstransmission skall ha högt motstånd mot upprepande belastning som förekommer på tänderna och skall även ha en minimal energiförlust. Detta examensarbete siktar mot att stödja den fortsatta utvecklingen inom området genom en fallstudie, mer specifikt för att studera sju olika kugghjulsmodeller som behandlar ett parallellt spiralformade kugghjul. Målet med detta examensarbete är att studera ett kugghjulskoncept där en ring appliceras för att reducera rot- och kontaktspänningar. Utförandet sker genom att studera sju olika modeller, för att veta hur stor påverkan designen utgör. En eller två stödringstrukturer appliceras, eller att öka kuggbredden på det kugghjul som anses vara det mest kritiska för höga spänningar. Denna studie görs för att kugghjulstillverkaren ska vara i utvecklingens framkant och kunna konkurrera inom fordonsindustrin. M1 ären referens och standard designmodell, medan de andra modellerna (M2-M5/ P1-P3) är modifierade design där med ytterligare ringar eller ändrad kuggbredd. Simulering är en effektiv metod för att förstå och visualisera komplexiteten av komponenter inom växellådan. Ett finita elementmetodens program användes för att undersöka dessa modeller, genom att importera geometrierna till pre-processorn ANSA, där Abaqus 2017 användes som en lösare, där sedan resultaten extraheras från post-processorn META. För att stödja denna studien användes två av de sju FEM-modellerna till att valideras mot ett annat specialiserat kugghjulsprogram inom kuggberäkning som heter WindowsLDP. Detta med avsikt att fastställa simuleringsmodellernas robusthet. Det så kallade överföringsfelet, rotböjnings- och kontaktspänningarna var ingående parametrar som behandlades under valideringen. Modellerna M3-M5/P1-P3 introducerades, där rotböjsspänningen och kontaktspänningen reducerades med 1.2-4.4 och 0.07-4.3% när de jämfördes med M1. Överföringsfelet (TE) kunde skilja upp mot 85% mellan M2-M3 jämfört med M1. Ett systematiskt fel uppstod i modell M2, modellens robusthet kunde därmed ej fastställas, då modellens resultat bör övervägas noggrant. Införande av så kallade växelförskjutning, lutning/vippning parametrar, mikrogeometrier och profilmodifieringen relaterat till kronning och tipavlastning, kommer att genera minskade TE-mätningar samt rot- och kontaktspänningar för de spiralformade kugghjulen. Dessa ämne har ej studerats under detta examensarbete. LDP-resultaten visade högre värden relativt jämfört med de FEM resultat, där en trend kunde observeras. Slutsatsen föreslog att detta bör undersökas ytterligare i framtiden.

Cette thèse s’intéresse à l’efficacité des renforcements par matériaux composites à base de fibres carbone (CFRP : Carbon Fiber Reinforced Polymer) par la technique NSM (Near Surface Mounted ; réalisation d’engravures sur la surface du béton) pour requalifier les structures corrodées. Elle est composée d’une partie expérimentale et d’une partie modélisation par la méthode des Eléments Finis (EF). La technique NSM consiste à réaliser des engravures sur la surface du béton où sont insérés les joncs de carbone rendus adhérents par remplissage d’une résine époxy. Les éléments étudiés dans cette thèse sont des poutres en béton armé naturellement corrodées pendant 28 années d’exposition à un environnement salin. Dans cette thèse, nous nous intéressons aux modes de rupture et à la capacité portante à la fois : en flexion et à l’effort tranchant. Le degré de corrosion a été mesuré par la méthode de perte de masses locales après la fin des essais mécaniques sur les barres longitudinales ainsi que sur les cadres d’effort tranchant. La première partie présente les résultats expérimentaux obtenus sur une poutre corrodée et une poutre témoin réparée ou renforcée en flexion par un jonc de carbone de 6 mm de diamètre inséré dans la surface tendue par la technique NSM. Les essais ont pour objectif d’étudier la capacité portante, la flèche à la ruine, la rigidité en flexion et le mode de ruine des deux poutres afin de vérifier l’efficacité du renforcement ou de la réparation... La seconde partie présente les résultats expérimentaux obtenus sur des poutres courtes permettant de mettre en avant la résistance vis à vis de l’effort tranchant. A partir des poutres longues testées dans la première partie, deux poutres courtes corrodées et deux poutres courtes témoins sont extraites. Une poutre courte corrodée et une poutre témoin sont réparées ou renforcées en flexion et une poutre corrodée et une poutre témoin sont en plus réparées ou renforcées vis-à-vis à l’effort tranchant par des joncs de carbone de 6 mm de diamètre par la technique NSM. L’éventuel glissement des armatures longitudinales sur les appuis a été mesuré durant les essais de flexion 3 points. Les résultats expérimentaux montrent que la corrosion des armatures longitudinales et la corrosion des cadres d’effort tranchant n’affectent pas le mode de ruine et modifient très peu les capacités portantes. La réparation vis-à-vis de l’effort tranchant avec la technique NSM change le mode de ruine de la formation d’une fissure diagonale due au glissement des barres d’acier tendues, à la formation d’une large fissure de flexion à mi-travée suivie de l’écrasement du béton comprimé... La troisième partie s’intéresse à la modélisation par la méthode de modélisation par EF du comportement global en flexion. Une première approche 2D est élaborée avec le code FEMIX développé à l’Université de Minho par le professeur Barros. Cinq poutres sont modélisées : 3 corrodées dont une réparée par la technique NSM et 2 témoins dont une renforcée par la technique NSM. La modélisation par EF donne de bons résultats sauf dans le cas de la poutre corrodée réparée dont le mode de ruine est non conventionnel (séparation du béton d’enrobage)... La quatrième et dernière partie est consacrée à l’étude numérique des poutres courtes. Quatre poutres sont modélisées : 2 corrodées réparées dont une réparée vis-à-vis de l’effort tranchant et 2 témoins renforcées dont une renforcée vis-à-vis de l’effort tranchant. La modélisation par EF est faite en 3D en utilisant le code FEMIX. Les résultats numériques reproduisent correctement les aspects importants du comportement force-flèche ainsi que les modes de fissuration à la ruine pour les poutres réparées vis-à-vis de l’effort tranchant ou non. / The thesis is composed of an experimental and Finite Element Modeling (FEM) study, investigating the effectiveness of the (Near Surface Mounted CFRP rods technique (NSM)) on the corroded RC beams. In the NSM technique, the CFRP rods are placed inside pre-cut grooves and are bonded to the concrete with epoxy adhesive. The thesis studies the failure modes and the mechanical performance of the corroded RC beams due to steel corrosion and finally, the prediction of the mechanical behaviour of the repaired corroded RC beams using nonlinear models based on FE numerical modelling. The corroded RC beams studied here in this study were exposed to natural corrosion for more than 28 years. The first part includes experimental results which were obtained on two beams: (one corroded and one control beams, both are 3 metres long) repaired or strengthened in bending with one 6-mm-diameter NSM CFRP rod. The beams were tested in a three-point bending test up to failure. Overall stiffness was studied. Ultimate capacity, ductility and failure modes were also reviewed. Finally some comparisons were made between repaired and non-repaired beams in order to assess the effectiveness of the NSM technique... The second part discusses the experimental results of two short corroded beams, which were tested under three-point bending until failure, along with two short control beams of the same characteristics (age, length and cross-section). One RC corroded deep beam was repaired in bending and the other one was repaired in both bending and shear with NSM CFRP rods. After the beams had been tested up to failure, the main steel bars and the stirrups were extracted from the beams and the loss of mass was measured and plotted for both the longitudinal and transverse reinforcement. The slip of tensile reinforcement at the end of the beams was also measured during the tests. The effect of corrosion and the effect of repairing with CFRP NSM rods in bending and shear on the behaviour of deep beams are discussed. Experimental results showed that both corroded and control deep beams repaired only in bending failed due to shear failure mode(diagonal tension failure), while corroded and control beams repaired in both bending and shear failed due to concrete crushing. The test results also showed that the corrosion of both longitudinal and transversal reinforcement hardly modified the mechanical response of deep beams... The third part shows experimental results and numerical modelling results of 2D finite element model using the FEMIX computer code were obtained on five, 3-metre-long beams: three corroded RC beams and two control beams. Two beams, one corroded and one control were repaired or strengthened in bending with NSM CFRP rod and were then tested in three-point bending up to failure. The FE numerical modelling results from FEMIX were compatible with the experimental ones except for the repaired corroded beam, for which a three-dimensional model using the commercial software ABAQUS was required... Finally, the last part presents 3D numerical modelling results in terms of load-deflection curves, and failure modes for 4 short corroded beams: two corroded beams and two control beams, half of the beams were let repaired or strengthened in bending only with NSM CFRP rods while the others were repaired or strengthened in both bending and shear with NSM technique. Results showed that the FE model was able to capture the main aspects of the experimental load-deflection curves of the RC beams, moreover it has presented the experimental failure modes and FE numerical modelling crack patterns and both gave similar results for both shear-repaired and non-shear repaired beams, three dimensional crack patterns were produced for shear-repaired beams in order to investigate the splitting cracks occurred at the middle of the beams and near the support.