Purpose: This paper will present insight to the methodology and results of the experimental characterization of Acrylonitrile Butadiene Styrene (ABS) using Fused Deposition Modeling (FDM). The work in this research explored the effects of print orientation, surface treatment, and ultraviolet (UV) light degradation with the utilization of Digital Image Correlation (DIC) on ABS tensile specimens. Design/methodology: ABS specimens were printed at three build orientations (flat (0 degrees), 45 degrees, and up-right (90 degrees)). Each of these specimens were treated with three different surface treatments including a control (acrylic paint, Cyanoacrylate, and Diglycidyl Bisphenol A) followed by exposure to UV light to the respective batches. This experiment design will provide tensile direction properties with the effect of thermoset coatings and UV degradation. Dogbone FDM specimens based on ASTM standard D638 type IV were printed on a Stratasys Dimension SST (Soluble Support Technology) 1200es 3D Printer and loaded into a MTS Landmark Servohydraulic Test Systems. Analysis was preformed on the fracture section of the tensile specimens utilized DIC and comparing Ultimate Tensile Strength (UTS) and Ultimate Fracture Strength (UFS). Findings: From the results UV light did not play a large factor in the strength of the specimens. The print orientation showed the largest anisotropic behavior where some specimens experienced as much as a 54% difference in ultimate tensile strength. Thermoset coated specimens experienced a maximum of 2% increase in strength for the Cyanoacrylate and Diglycidyl Bisphenol A specimens where the acrylic paint and natural did not. Several findings were of value when looking at the stress strain plots. Originality/value: This paper provides knowledge to the limited work on print build orientation, thermoset coatings and, UV light on ABS specimens. Very little to no work has been done on these three properties. This paper can serve as the foundation of future work on external applications on ABS plastics.

Three testing methods are proposed to characterize properties of fuel cell materials that affect the mechanical durability of proton exchange membranes (PEMs). The first two methods involved measuring the in-plane biaxial strength of PEMs and the biaxial hygrothermal stresses that occur in PEMs during hygrothermal cycles. The third method investigated the nonuniform thickness and compressibility of gas diffusion media which can lead to concentrated compressive stresses in the PEM in the through-plane direction. Fatigue and creep to leak tests using multi-cell pressure-loaded blister fixtures were conducted to obtain the lifetimes of PEMs before reaching a threshold value of gas leakage. These tests are believed to be more relevant than quasi-static uniaxial tensile to rupture tests because of the introduction of biaxial cyclic and sustained loading and the use of gas leakage as the failure criterion. They also have advantages over relative humidity cycling test because of the controllable mechanical loading. Nafion® NRE-211 membrane was tested at three different temperatures and the time-temperature superposition principle was used to construct a stress-lifetime master curve. Tested at 90°C, extruded Ion Power® N111-IP membrane was found to have longer lifetime than Gore™-Select® 57 and Nafion NRE-211 membranes under the same blister pressure profiles. Bimaterial specimens fabricated by bonding a piece of PEM to a substrate material were used to measure the hygral stresses, compressive and tensile, in the PEM during relative humidity cycles. The substrate material and its thickness were carefully chosen so that stresses in the PEM could be obtained directly from the curvature of the bimaterial specimen without knowing the constitutive properties of the PEM. Three commercial PEMs were tested at 80°C by cycling the relative humidity between 90% and 0% and by drying the membrane to 0%RH after submersion in liquid water. Stress histories for all three membranes show strong time-dependencies and Nafion® NRE-211 exhibited the largest tensile stress upon drying. Besides in-plane stresses, hard spots in gas diffusion media (GDM) can locally overcompress PEMs in the out-of-plane direction and cause electrical shorting. In this study, GDM samples sealed with an impermeable Kapton® film on the surface were compressed with uniform air pressure and the nonuniform displacement field was measured with a three-dimensional digital image correlation technique. Hard spots as a result of the nonuniform thickness and compressibility of the GDM were found and their severities as stress risers are evident. Locally, a nominal platen compression (similar to bipolar plate land compression) of 0.68 MPa can lead to compressive stress as large as 2.30 MPa in various hard spots that are in the order of 100s µm to 1 mm in size. / Ph. D.

The ability to quickly develop predictions of the time-to-failure under different loading levels allows designers to choose the best polymeric material for a specific application. Additionally, it helps material producers to design, manufacture, test, and modify a polymeric material more rapidly. In the case of polymeric pipes, previous studies have shown that there are two possible time-dependent failure mechanisms corresponding to ductile and brittle failure. The ductile mechanism is evident at shorter times-to-failure and results from the stretching of the amorphous region under loading and the subsequent plastic deformation. Empirical results show that many high-performance polyethylene (PE) materials do not exhibit the brittle failure mechanism. Hence, it is critical to understand the ductile mechanism and find an approach to predict the corresponding times-to-failure using accelerated means. The aim of this study is to develop an innovative rupture lifetime acceleration protocol for PE pipes which is sensitive to the structure, orientation, and morphology changes introduced by changing processing conditions. To accomplish this task, custom fixtures are developed to admit tensile and hoop burst tests on PE pipes. A pressure modified Eyring flow equation is used to predict the rupture lifetime of PE pipes using the measured mechanical properties under axial tensile and hydrostatic pressure loading in different temperatures and strain rates. In total, the experimental method takes approximately one week to be completed and allows the prediction of pipe lifetimes for service lifetime in excess of 50 years. / Master of Science

This thesis focuses on the effect of microstructural variation on the mechanical properties and deformation behavior of ductile iron. To research and determine these effects, two grades of ductile iron, (i) GJS-500-7 and (ii) high silicon GJS-500-14, were cast in a geometry containing several plates with different section thicknesses in order to produce microstructural variation. Microstructural investigations as well as tensile and hardness tests were performed on the casting plates. The results revealed higher ferrite fraction, graphite particle count, and yield strength in the high silicon GJS-500-14 grade compared to the GJS-500-7 grade. To study the relationship between the microstructural variation and tensile behavior on macroscale, tensile stress-strain response was characterized using the Ludwigson equation. The obtained tensile properties were modeled, based on the microstructural characteristics, using multiple linear regression and analysis of variance (ANOVA). The models showed that silicon content, graphite particle count, ferrite fraction, and fraction of porosity are the major contributing factors that influence tensile behavior. The models were entered into a casting process simulation software, and the simulated microstructure and tensile properties were validated using the experimental data. This enabled the opportunity to predict tensile properties of cast components with similar microstructural characteristics. To investigate deformation behavior on micro-scale, a method was developed to quantitatively measure strain in the microstructure, utilizing the digital image correlation (DIC) technique together with in-situ tensile testing. In this method, a pit-etching procedure was developed to generate a random speckle pattern, enabling DIC strain measurement to be conducted in the matrix and the area between the graphite particles. The method was validated by benchmarking the measured yield strength with the material’s standard yield strength. The microstructural deformation behavior under tensile loading was characterized. During elastic deformation, strain mapping revealed a heterogeneous strain distribution in the microstructure, as well as shear bands that formed between graphite particles. The crack was initiated at the stress ranges in which a kink occurred in the tensile curve, indicating the dissipation of energy during both plastic deformation and crack initiation. A large amount of strain localization was measured at the onset of the micro-cracks on the strain maps. The micro-cracks were initiated at local strain levels higher than 2%, suggesting a threshold level of strain required for micro-crack initiation. A continuum Finite Element (FE) model containing a physical length scale was developed to predict strain on the microstructure of ductile iron. The material parameters for this model were calculated by optimization, utilizing the Ramberg-Osgood equation. The predicted strain maps were compared to the strain maps measured by DIC, both qualitatively and quantitatively. To a large extent, the strain maps were in agreement, resulting in the validation of the model on micro-scale. In order to perform a micro-scale characterization of dynamic deformation behavior, local strain distribution on the microstructure was studied by performing in-situ cyclic tests using a scanning electron microscope (SEM). A novel method, based on the focused ion beam (FIB) milling, was developed to generate a speckle pattern on the microstructure of the ferritic ductile iron (GJS-500-14 grade) to enable quantitative DIC strain measurement to be performed. The results showed that the maximum strain concentration occurred in the vicinity of the micro-cracks, particularly ahead of the micro-crack tip. / Denna avhandling fokuserar på effekten av variationer i mikrostrukturen på mekaniska egenskaper och deformationsbeteende hos segjärn. För att undersöka dessa effekter, två olika sorter av segjärn, (i) GJS-500-7 och (ii) högkisellegerad GJS-500-14, gjutits till plattor av olika tjocklekar för att generera mikrostrukturvariationen. Mikrostrukturundersökning, samt drag- och hårdhetsprov gjordes på de gjutna plattorna. Resultaten visade att en högre ferritfraktion, grafitpartikelantal och sträckgräns i den högkisellegerade GJS-500-14-sorten jämfört med GJS-500-7. För att studera förhållandet mellan mikrostrukturell variation och spännings-töjningsbeteendet på makroskala, modellerades detta med hjälp av Ludwigson-ekvationen. De erhållna spännings-töjningsegenskaperna modellerades baserat på mikrostrukturell karaktäristika genom multipel linjärregression och variansanalys (ANOVA). Modellerna visade att kiselhalt, grafitpartikelantal, ferritfraktion och porfraktion var de viktigaste bidragande faktorerna. Modellerna implementerades i ett simuleringsprogram för gjutningsprocessen. Resultatet från simuleringen validerades med hjälp av experimentella data som inte ingick i underlaget för regressionsanalysen. Detta möjliggjorde att prediktera spännings-töjningsbeteendet och dess variation hos gjutna segjärns komponenter med liknande sammansättning och gjutna tjocklekar som användes i denna studie. För att kunna undersöka deformationsbeteendet på mikroskala utvecklades en metod för kvantitativ mätning av töjning i mikrostrukturen, genom DIC-tekniken (digital image correlation) tillsammans med in-situ dragprovning. I denna metod utvecklades en grop-etsningsprocess för att generera ett slumpvis prickmönster, vilket möjliggjorde DIC-töjningsmätning i matrisen och i området mellan grafitpartiklarna med tillräcklig upplösning. Metoden validerades genom benchmarking av den uppmätta sträckgränsen mot materialets makroskopiska sträckgräns mätt med konventionell dragprovning. Det mikrostrukturella deformationsbeteendet under dragbelastning karakteriserades. Under elastisk deformation avslöjade töjningsmönstret en heterogen töjningsfördelning i mikrostrukturen, och bildandet av skjuvband mellan grafitpartiklar. Sprickbildning initierades vid låg spänning och redan vid de spänningsnivåer som ligger vis ”knät” på dragprovningskurvan, vilket indikerar energidissipering genom både begynnande plastisk deformation och sprickbildning. Den lokala töjningen vis sprickinitiering skedde då den lokala töjningen översteg 2%, vilket indikerar att detta skulle kunna vara en tröskelnivå för den töjning som erfordras för initiering av mikro-sprickor. En kontinuum Finita Element (FE) modell utvecklades för att prediktera töjningen hos ett segjärn och dess fördelning i segjärns mikrostruktur. Materialparametrarna för denna modell optimerades genom att anpassa parametrarna i Ramberg-Osgood ekvationen. De predikterade töjningsfördelningarna jämfördes med de experimentell uppmätta töjningsmönstren uppmätta med DIC, både kvalitativt och kvantitativt. Töjningsmönstren överensstämde i stor utsträckning, vilket resulterade i att modellerna kunde anses vara validerade på mikronivå. För att kunna mäta töjningsmönster under dynamiska förlopp på mikronivå utvecklades en metod för att skapa prickmönster och att utföra in-situ CT provning i ett svepeletronmikroskop (SEM). Prickmönstret skapades genom avverkning med en fokuserad jonstråle (FIB), och provades på det ferritiska segjärnet (GJS-500-14 grad). Resultaten visade att maximal töjningskoncentration fanns i närheten av mikrosprickorna, framförallt framför sprickspetsen.

Les outillages de mise en forme à chaud sont soumis à des sollicitations thermomécaniques transitoires. Ils sont donc confrontés à la fissuration par fatigue. Le but de ce travail est d'étudier le comportement et la durée de vie en fatigue des aciers d'outillage, notamment leurs résistances à la propagation de fissure. Dans un premier temps, les chargements thermiques imposés aux outillages de fonderie sous pression de magnésium et d'emboutissage à chaud ont été évalués. Ensuite, le comportement et la durée de vie de l'acier à haute conductivité thermique HDC1 ont été étudiés et comparés à l'acier AISI H11 (acier de référence) en fonction de la température. L'acier HDC1 présente un adoucissement cyclique stable à 20°C et 300°C. Par contre, l'intensité d'adoucissement est plus importante à hautes températures. La durée de vie a été étudiée en utilisant les lois de Manson-Coffin et de Basquin. A hautes températures, l'oxydation devient un mécanisme d'endommagement primordial pour l'acier HDC1 et provoque des durées de vie plus courtes que celles observées sur l'acier AISI H11. La résistance à la propagation de fissure de fatigue a été déterminée dans des aciers à la température ambiante par le biais de critères : l'amplitude de facteur d'intensité de contrainte (∆K) et l'amplitude d'ouverture en pointe de fissure (∆CTOD). La méthode de corrélation d'images a permis de mesurer (∆COD) et d'évaluer (∆CTOD). L'ensemble de ces résultats a permis de mettre en évidence l'effet de fermeture de fissure et le comportement plastique en pointe de fissure. ∆CTOD présente un bon critère pour rationaliser la propagation de fissure dans les aciers étudiés.Enfin, la simulation numérique de la propagation de fissure en fatigue a été menée dans l'acier AISI H11 à 600°C par la méthode de relâchement des nœuds en éléments finis. L'effet des modèles de comportement monotone (élastoplastique) et cyclique (élastoviscoplastique) a été étudié sur le calcul de l'ouverture de fissure et la plasticité en pointe de fissure. L'influence du modèle de comportement est faible sur le calcul de l'amplitude d'ouverture de fissure ∆COD, du fait d'une plasticité confinée en pointe de fissure. Alors que, le modèle de comportement cyclique est mieux adapté pour décrire la plasticité en pointe de fissure. / Hot metal forming tools are subjected to cyclic thermomechanical loading and damage by complex fatigue/wear/oxidation interactions. Thermal solicitations were measured on high pressure die casting and hot stamping processes. Based on thermal measurements, the isothermal fatigue behaviour and lifetime of a new high conductivity steel HDC1 were investigated at different temperatures and strain amplitude then compared to AISI H11 steel. As AISI H11, continue cyclic softening was observed in HDC1 at all temperatures. The Manson-Coffin and Basquin laws were used for life prediction models under different temperatures. It was observed that the fatigue/oxidation interaction was a principal damage mechanism of the HDC1 steel at high temperature. Fatigue crack propagation in steels was investigated at room temperature in SENT specimens. A digital image correlation technique was used to evaluate crack opening (∆COD) and crack tip opening displacement (∆CTOD) ranges. Crack growth rate were investigated using ∆K (Paris law) and ∆CTOD criteria. It was observed that the cyclic crack tip plasticity control the crack propagation resistance. Crack closure could be evaluated by ∆CTOD.Finite element method by debond technique was used to model the crack propagation of AISI H11 at 600°C using both monotonic elasto-plastic (EP) and cyclic elasto-viscoplastic (EVP) constitutive laws materials. The comparison of ∆COD calculated and measured had shown that monotonic EP and cyclic EVP had no significant effect on the ∆COD, mainly due to the small-scale yielding conditions. It is however observed that the cyclic constitutive law was the best suitable model for the crack tip plasticity effect.

Dans cette étude, les propriétés mécanique et la microstructure du polyéthylène à haute densité (PEHD)/ fibres de bois (Pinmaritime) ont été caractérisés. Le comportement en traction et en flexion 4 points de composites fibres de bois/polyéthylène(WPC) avec et sans additif a été étudié en utilisant des mesures de champs par stéréo-corrélation d’image numériques. Nousavons tout d’abord comparé les mesures de déformation longitudinale par stéréo-corrélation d’image à celles mesurées parextensométrie mécanique au cours d’un essai de traction simple uni-axial. Les valeurs de déformation macroscopiquemesurées sont comparables à celles obtenues par extensométrie mécanique classique. Quatre formulations ont été testées(avec ou sans l'additif et avec 10% ou 30% de bois) au travers des essais de traction et flexion 4 points en utilisant laméthode par stéréo-corrélation d’image. Nous avons pu ainsi corréler la visualisation combinée des images enregistrées parles caméras au comportement mécanique ainsi qu’aux modifications des surfaces des composites. Les résultats ont montréque les singularités au niveau de la surface des échantillons ainsi que la présence des fibres de bois (dans les proportionsutilisées durant nos essais) n’ont pas d’effets sur la qualité du composite. La stéréo corrélation d’images nous permet d’avoirune information précise sur le cisaillement tant d’un point de vue quantitatif que qualitatif. L’essai cyclique a été utilisé pouranalyser l’endommagement du composite. Les courbes en traction et flexion ont montré un comportement global non linéaire. Le module d’élasticité (MOE) est évalué par la tangente à l’origine d’un modèle de Maxwell-Binhgam ajusté paroptimisation sur les courbes expérimentales. Les cartographies fausses couleurs du champ de déformation du WPC ont étéprésentées. D’autre part, on a analysé le comportement endommageable et le degré d’hétérogénéité pour différentspourcentages de bois à l’aide des valeurs de l’écart type spatial du champ de déformation longitudinale. Par cette technique,nous avons également pu étudier les propriétés mécaniques en traction et en compression à partir d’un essai de flexion 4points. Le second objectif était d’améliorer la performance des liaisons entre les fibres de bois hydrophiles polaires et lamatrice plastique de PEHD hydrophobe non polaire. Pour cette raison, la fibre de bois a été traitée par anhydride maléique depolyéthylène (MAPE) (greffage par copolymérisation) et acétylée (greffage de chaine de carbone). Dans cette partie, nousavons étudié l’effet de MAPE ainsi que la longueur de la chaine de carbone greffée sur les propriétés mécaniques du WPC.Par ailleurs, nous avons pu observer et caractériser la morphologie à l’interface fibre de bois/HDPE du WPC greffé enutilisant un microscope électronique à balayage (MEB). / Firstly we have compared longitudinal strenght obtained by stereo correlation with mechanical extensometer forthe same tests. It is shown that macroscopic values of the longitudinal strain are closed to those measured bymechanical extensometers using standard mechanical tests. Four injected WPC formulations (with or withoutadditive and with 10% or 30% of wood) are consider. WPC surface particularities and wood fibers distributionon samples have no effect on WPC mechanical performances. Also 3D digital image correlation givesinformations on Tensile performances (from quality and quantity point of view). Cyclic tensile tests have beenperformed in order to analyse the damage of material. A non-linear behaviour is shown. The Modulus ofElasticity (MOE) is provided by a Maxwell-Bingham model fitted to the experimental tensile curves. Color mapsof the spatial strain distribution are commented. Moreover, the damage behaviour and the degree ofheterogeneity for several percentages of fibre wood-based WPC are analysed thanks to the spatial standarddeviation of the longitudinal strain field. With this technology it was possible to study of the mechanicalproperties in tensile and compression during four bending test. The second objective was to improve the linkbetween PEHD matrix and wood fibres because of the incompatibility between the polar hydrophilic wood fibresand the non-polar hydrophobic polyethylene. For this reason, the wood fibre was treated by maleic polyethyleneanhydride (MAPE) (grafting by copolymerization) and was acetylated (grafting of carbon chain). In this part, theeffect on the mechanical properties, of maleic anhydride modified polyethylene (MAPE) and of the length of thecarbon chain graft (CCG) between a (HDPE) matrix and wood fiber is studied. Furthermore, Scanning ElectronMicroscope (SEM) is used to characterize the morphology of the wood fibre / HDPE matrix interface forspecimens with carbon chain grafted.

[pt] A avaliação da propagação de trincas de fadiga inclui a identificação da direção da trinca, o conhecimento do Fator de Intensidade de Tensões (SIF) equivalente, a determinação de uma taxa de crescimento de comprimento de trinca por número de ciclos da/dN e o estabelecimento de uma regra de propagação de trinca conectando SIF e da/dN, como uma regra de tipo Paris. Quando ocorrem cargas mistas e não proporcionais, esses parâmetros ainda não são totalmente compreendidos. Esta tese trata de algumas das variáveis que influenciam a propagação de trincas sob carregamento no modo misto não proporcional. A técnica de Correlação de Imagens Digitais (DIC) foi utilizada para a aquisição de imagens de corpos de prova submetidos a carregamento proporcional e não proporcional cíclico. Dois tipos de corpos de prova foram utilizados. Primeiramente, dois corpos de prova planos foram testados; um disk compact tension (DCT, em inglês) e um compact tension modificado (C (T) em inglês). Eles foram submetidos a carregamento cíclico induzindo o modo I de abertura de trinca ou modos I e II de abertura de trinca proporcionais. Em segundo lugar, os dados DIC adquiridos anteriormente, e em outro lugar, para cinco tubos finos sujeitos a carregamento cíclico foram analisados. Os tubos finos tiveram entalhes usinados a partir dos quais as trincas por fadiga iniciaram e se propagaram. Esses cinco tubos finos foram submetidos a diferentes casos de carga proporcional e não proporcional. Um corpo de prova tipo tubo fino foi exposto a carga axial e apresentou modo de abertura de trinca tipo I. Os outros quatro foram submetidos a carregamento de torção ou carga axial-torcional mista e exibiram todos os três modos de abertura de trinca I, II e III. Os campos de deslocamento adquiridos experimentalmente com a técnica DIC foram processados para calcular independentemente o SIF para cada modo de abertura existente usando formulações de mecânica de fratura elástica linear (MFLE). Uma formulação delas utilizou dados de deslocamento de campo completos adquiridos em pequenas áreas que circundavam a ponta da trinca. Outra formulação usou dados adquiridos de um par de pontos localizados ao longo dos flancos opostos das faces da trinca. Os SIFs determinados foram usados para encontrar os SIFs equivalentes e faixas de SIF equivalentes usando o critério da tensão máxima de tração (para ambas as versões 2D e 3D de combinações dos modos I-II e modos I-II-III respectivamente) que implicitamente incluíram o ângulo de propagação de trinca. Verificou-se que a inclusão do SIF no modo III experimentalmente determinado efetivamente faz diferença nas faixas do SIF e dos SIF equivalentes estimados. A curva da/dN versus faixa do SIF equivalente foi elaborado com as taxas de crescimento de trinca medidas experimentalmente e as faixas de SIF que foram encontradas usando a suposição amplamente aceita de que as trincas cresceram na direção que maximiza a tensão de tração. Para isso, extensões do modelo de Schollmann et. al. e bem como o modelo de Erdogan-Sih, que são geralmente aplicados ao carregamento proporcional, foram usados para determinar os SIFs equivalentes e faixas de SIF equivalentes para os casos de carregamento proporcional e não proporcional. Finalmente, a segunda zona da regra de Paris (da/dN versus faixa do SIF equivalente) foi plotada para os cinco casos de carregamento nos tubos finos mostrando que eles caíram dentro de uma faixa razoavelmente fina e dispersa. / [en] Fatigue crack propagation assessment includes identifying the crack direction, knowing the equivalent Stress Intensity Factor (SIF) range, determining a crack length growth rate per number of cycles (da/dN), and establishing a crack propagation rule connecting the equivalent SIF and da/dN rate, such as a Paris type of rule. When mixed and non-proportional loading occur, those parameters are not fully understood yet. This thesis deals with some of the variables that influence crack propagation under non-proportional mixed mode loading. The Digital Image Correlation (DIC) technique was used to acquire images of test specimens subjected to cyclic proportional and non-proportional loading. Two types of specimen samples were used. Firstly, two different plate test specimens were tested; a disk compact tension (DCT), and a modified compact tension, C(T). They were subjected cyclic loading inducing crack opening mode I or proportional crack opening modes I and II. Secondly, the previously and elsewhere acquired DIC data for five thin tubes subject to cyclic loading were analyzed. The thin tubes had pre-fabricated slit-notches from which fatigue cracks initiated and propagated. Those five thin tubes were subjected to different cases of proportional and non- proportional loading. One tube specimen was exposed to axial loading and presented mode I crack opening. The other four were subjected to torsion loading or mixed axial-torsional loading and exhibited all three I, II and III crack-opening modes. The experimentally acquired DIC displacement fields were processed to independently calculate SIF for each existing opening mode using linear elastic fracture mechanics (LEFM) formulations. One formulation used full field displacement data acquired in small areas that surrounded the crack tip. Another formulation used data acquired from a pair of points located along the opposite crack flanks. The determined SIFs were used to find equivalent SIFs and equivalent SIF ranges using the maximum tensile stress criterion (for both 2D and 3D versions of combinations of modes I-II and modes I-II-III respectively) which implicitly included the crack propagation angle. It was found that the inclusion of the experimentally determined mode III SIF indeed makes a difference in the determined equivalent SIF and equivalent SIF ranges. A da/dN versus equivalent SIF ranges plot was drafted with the experimentally measured crack growth rates and the SIF ranges that were found by using the widely accepted assumption that the cracks grew in the direction that maximizes the tensile stress. For this, extensions of the Schollmann et. al. model as well as of the Erdogan-Sih model, which are generally applied to proportional loading, were used to determine equivalent SIFs and equivalent SIF ranges for the cases of proportional and non-proportional loading. Finally, the second stage of the Paris rule (da/dN versus SIF range) was plotted for the five thin tubes loading cases showing that they fell inside a reasonably thin scattered band.

L'automobile nécessite l'utilisation d'échangeurs thermiques permettant d'assurer au moteur des conditions de fonctionnement en température acceptables (autour de 90°C). La fiabilité de ces échangeurs ne peut être négligée car ils peuvent être à l'origine de complications mécaniques importantes en cas de dysfonctionnement. La maîtrise des divers modes d'endommagement des échangeurs thermiques liés aux conditions d’utilisation devient dès lors un enjeu incontestable pour les industriels en charge de leur production, tant du point de vue matériaux, que du comportement général de la structure en service (influence du procédé d'assemblage, design,...). Les échangeurs thermiques présentent aujourd’hui une sensibilité accrue aux sollicitations thermomécaniques cycliques induites en service, du fait, essentiellement, d’une constante diminution des épaisseurs des composants. Celle-ci est responsable d’une augmentation significative des contraintes internes pour des conditions en service identiques, pouvant avoir pour conséquence directe et irréversible la rupture d’un tube, témoin d’une incompatibilité matière/design/process. Deux configurations matières industrielles ont ici été étudiées. Il s’agit de structures tri-couches colaminées de type tube, constituées respectivement de deux et trois alliages d’aluminium distincts (4xxx/3xxx/4xxx ou 4xxx/3xxx/7xxx), pour une épaisseur totale de 270µm. Ces dernières ont été développées pour permettre l’utilisation du procédé de brasage comme procédé d’assemblage (alliage 4xxx). Toutefois, une telle architecture, combinée à un procédé thermique d’assemblage sévère (600°C), est responsable d’une modification profonde de la microstructure avec l’apparition de structures de solidification, responsables entre autres de nombreuses irrégularités de surface (appelées Gouttes de Placage Résiduelles - GPR) ainsi que d’importants gradients de propriétés mécaniques dans l’épaisseur. Ces travaux de recherche s’appuient sur une approche expérimentale et numérique développée pour étudier les mécanismes d'endommagement en fatigue relatifs à de telles structures fines hétérogènes. Associant diverses techniques expérimentales telles que la corrélation d’images numériques (2D-3D) ou la tomographie à rayons X (de laboratoire ou à l’ESRF), elle permet une analyse précise des mécanismes d’amorçage et de propagation des fissures de fatigue (sur éprouvette de fatigue classique ou de type échangeur thermique). Le rôle des différents placages dans chacune des phases de l’endommagement a ainsi été mis en évidence (4xxx : amorçage, 7xxx : propagation des fissures). Des simulations par la méthode des éléments finis nous ont permis de compléter ces observations en proposant une quantification précise de l’influence de l’état de surface (GPR) sur la tenue en fatigue des éprouvettes testées. Enfin, des essais de fatigue réalisés directement sur échangeurs thermiques ont permis de corroborer les résultats obtenus sur éprouvettes modèles. / The automotive industry, like many other industrial fields, requires the use of heat thermal exchangers to allow optimal thermal service conditions of the engine (around 90°C for a car engine). The exchangers’ reliability has to be guaranteed to avoid a decrease of the engine efficiency or detrimental mechanical damage resulting from too high service temperatures. It is therefore necessary to control the different damage modes of such thermal heat exchangers according to the conditions of use. Thanks to their good thermal, corrosion and mechanical properties, aluminium alloys have steadily replaced copper alloys and brass for manufacturing heat exchangers in cars or trucks. Such components have been constantly optimized in terms of exchange surface area and, nowadays, this has led to Al components in heat exchangers with a typical thickness of the order of 0.2 to 1.5 mm. With such small thicknesses, the load levels experienced by heat exchangers components has drastically increased leading to an important research effort in order to improve the resistance to damage development during service life. Two industrial materials made of 3 co-rolled aluminium alloys (total thickness 0.27 mm) have been studied. In spite of their small thickness, the materials exhibit a composite structure comprising a core material (3xxx alloy) and 2 clads (4xxx and/or 7xxx alloys according to material configuration: 4xxx/3xxx/4xxx or 4xxx/3xxx/7xxx). The lower melting point 4xxx alloy is used for producing the heat exchanger assembly during a brazing process while the 7xxx alloy improves internal corrosion resistance. Such complex architecture, combined to the severe brazing thermal treatment, leads to important microstructural modifications, mainly characterized by the formation of brazing joints or Clad Solidification Drops (CSD) on the surface. Both of them are responsible for significant gradients of the mechanical properties on the thickness. The present study is based on an original experimental and numerical approach developed to characterise the different fatigue damage mechanisms operating in such thin heterogeneous structures. Digital image correlation (2D-3D) and X-rays tomography (at different resolutions) have been used to analyze the crack initiation and propagation mechanisms, highlighting the impact of each clad on each damage step. While the 4xxx clad corresponds to preferential crack initiation zones, the 7xxx clad seems to affect significantly the crack propagation phase. Finite Elements simulations have been carried out to complete these experimental observations, putting forward an accurate quantification of the surface state influence (through the CSD). All the different results and observations made on fatigue samples with a simplified geometry have been finally confirmed by fatigue tests on thermal exchanger configurations.

Ce mémoire présente une méthode mise au point pour caractériser et analyser des fissures de fatigue présentant un fort caractère tridimensionnel dans des matériaux métalliques opaques. L'analyse consiste à déterminer avec précision la position du front de la fissure étudiée et à mesurer des valeurs de facteurs d'intensité des contraintes le long du front par projection sur les séries de Williams du champ de déplacement issu de la corrélation numérique d'images 3D obtenues par tomographie aux rayons X. La corrélation d'images 3D numériques est utilisée afin de mesurer le champ de déplacement en volume lors de la mise sous chargement d'une éprouvette fissurée fatiguée. La corrélation d'images nécessitant un mouchetis, le matériau retenu pour les expériences est la fonte à graphite sphéroïdal car il présente un mouchetis 3D naturel (les nodules de graphites) parfaitement imagé par tomographie aux rayons X. Le cyclage est appliqué à l'aide d'une machine de fatigue in situ permettant d'alterner des phases de propagation de la fissure avec des acquisitions tomographiques sous différentes charges. L'introduction d'un défaut artificiel (une entaille obtenue par usinage laser) permet de maîtriser l'amorçage et la propagation de la fissure in situ. La méthode de corrélation d'images 3D numériques employée dans ces travaux étant basée sur des éléments finis, nous avons cherché à tirer profit de différents outils développés dans le cadre de cette méthode. Les surfaces libres sont spécifiées afin de bien conditionner le maillage et un enrichissement dans l'esprit des X-FEM permet de renseigner la fissure dont la position est repérée grâce à la trace laissée dans le résidu de corrélation entre l'image avant cyclage et la dernière image acquise. Une régularisation mécanique est également introduite dans le calcul sous forme d'un filtre de longueur d'onde choisie. Le champ de déplacement mesuré avec précision est ensuite projeté sur les séries de Williams augmentées des termes correctifs de Leblond et Torlai qui prennent en compte la courbure du front de la fissure. L'annulation du terme super-singulier d'ordre -1 des séries de Williams est utilisée pour détecter la position du front de la fissure. Une procédure itérative a été mise en place afin de concilier l'enrichissement et la courbure du front avec la projection sur les séries de Williams. Une fois la position du front 3D de la fissure déterminée et les valeurs des facteurs d'intensité des contraintes associées calculées, les résultats obtenus sont confrontés à la littérature. / This manuscript describes a methodology used to compute Stress Intensity Factor values along the curved front of a fatigue crack inside a nodular cast iron. An artificial defect is introduced at the surface of a small sample. The initiation and growth of a fatigue crack from this defect during constant amplitude cycling is monitored in situ by laboratory x-ray tomography. The method for processing the 3D images in order to compute SIF values is described in detail. The results obtained show variations of the stress intensity factor values along the crack front.

Artificial Neural Networks (ANNs) can be viewed as a mathematical model to simulate natural and biological systems on the basis of mimicking the information processing methods in the human brain. The capability of current ANNs only focuses on approximating arbitrary deterministic input-output mappings. However, these ANNs do not adequately represent the variability which is observed in the systems' natural settings as well as capture the complexity of the whole system behaviour. This thesis addresses the development of a new class of neural networks called Stochastic Neural Networks (SNNs) in order to simulate internal stochastic properties of systems. Developing a suitable mathematical model for SNNs is based on canonical representation of stochastic processes or systems by means of Karhunen-Loéve Theorem. Some successful real examples, such as analysis of full displacement field of wood in compression, confirm the validity of the proposed neural networks. Furthermore, analysis of internal workings of SNNs provides an in-depth view on the operation of SNNs that help to gain a better understanding of the simulation of stochastic processes by SNNs.