Une augmentation de la température à jeune âge du béton peut être causée par la chaleur d’hydratation dans le cas d’éléments massifs ou par l’application d’un traitement thermique dans l’industrie de la préfabrication. Lorsque le couple critique température maximale/durée de maintien est dépassé, il y a formation différée de l’ettringite (DEF) expansive pouvant fissurer les pièces. Les travaux présentés dans ce mémoire ont pour but principal d’établir une limite sécuritaire pour des mélanges V-S pour béton de masse et V-P pour béton préfabriqué régulièrement utilisés par le Ministère des Transports du Québec. Ces bétons utilisent des liants binaires composés de fumée de silice de type GUb-SF et des liants ternaires composés de fumée de silice et de cendres volantes ou de laitier de haut-fourneau. La norme canadienne actuelle (CSA A23.1-14 et CSA A23.4-16) impose une limite fixe de température qui ne dépend ni de la durée de traitement ni du type de liant utilisé. Le lien, souvent énoncé dans la littérature, entre la réponse d’un liant à un traitement thermique et sa composition chimique et minéralogique, déterminé respectivement par fluorescence X et diffraction des rayons X couplée par la méthode de calcul Rietveld, a pu être confirmé. Le rapport SO3 solubles/Al2O3 est notamment ce qui permet la meilleure estimation de la sensibilité d’un liant. Le processus de formation différée de l’ettringite est lent et peut prendre plusieurs années. Donc, pour étudier le phénomène, des méthodes d’essais accélérés ont été testées. Des méthodes ont ainsi pu être proposées pour de futurs tests. Une méthode d’essai rapide d’expansion de barres de mortier permet d’identifier les liants les plus sensibles à la DEF. La formation de la DEF dans les liants les moins sensibles peut être appréciée par le gain de masse de prismes de béton et leur observation microstructurale au microscope électronique à balayage (MEB). Une méthode sur bétons de masse et bétons préfabriqués permet de classer le degré de sensibilité d’un liant face à la DEF selon le gain de masse des prismes de béton. Les liants ternaires montrent une bonne résistance face à la DEF. La sensibilité des liants binaires dépend du rapport SO3 solubles/Al2O3 du liant. / Abstract : A heat cure for concrete can be caused by hydration in massive structures or by thermal treatment in prefabrication industry. When the limit couple maximum temperature/duration of treatment is exceeded, the delayed formation of ettringite occurs and leads to cracking in concrete. The main goal of the work presented in this essay is to establish a safe batch limit for some mixtures, VS (mass concrete) and V-P (precast concrete) which are often used by MTQ. These concretes use some cements blended with silica fume and ternary blended cements with silica fume and fly ash or blast furnace slag. Indeed, the current Canadian standard (CSA A23.1) set a limit which doesn’t depend on duration of the treatment nor the type of bender. However, the link between composition of the used bender, with can be determined by X-ray fluorescence and X-ray diffraction, and the response to a thermal treatment has been confirmed. The soluble SO3/Al2O3 ratio allows to estimate if a bender is sensitive to heat cure. The delayed formation of ettringite is a long process which can last a decade. In order to study this phenomenon, several accelerated test methods have been used. Some of them are available for some future tests. An accelerated expansion method using mortar bar allows to spot the DEF-sensitive benders. Delayed ettringite formation in the less sensitive bender can be estimated by weight gain in concrete prisms and by observation with scanning electron microscope (SEM). The sensitivity level of a bender can by estimated using a method based on weight gain. Ternary blended cements show a greater ability to withstand DEF. The sensibility of binary blended cements depends on soluble SO3/Al2O3 ratio.

La temperatura de curado, utilizada por la industria de fibrocemento como catalizador en la hidratación del cemento para lograr alta resistencia inicial en sus productos, genera la hipótesis de expansión causada por formación tardía de etringita. Para el planteamiento de dicha hipótesis, el siguiente estudio contempló dos grupos de muestras con el objetivo de estudiar aisladamente su comportamiento químico y físico: uno de los grupos conformado por pasta cementicia con 35% de sustitución de calcario, y el otro conformado por fibrocemento con incorporación de PVA y celulosa. Por calorimetría isotérmica se observó la cinética de reacción de dos pastas, una de ellas curada a 23 ºC y la otra a 85 ºC donde esta última experimentó un efecto catalizador en sus reacciones. Ensayos de DRX y DTG/TG mostraron la presencia de etringita a 23 °C durante todas las edades de hidratación exhibiendo una tendencia de aumento en la intensidad de los picos y en la cantidad, respectivamente, en función del tiempo; sin embargo, a 85 ºC la presencia de esta fase fue inconstante durante las primeras edades de hidratación, a partir de mes de hidratación mostró aumento en cada una de las siguientes medidas realizadas mensualmente; a 85 °C hubo una menor formación de portlandita en comparación con los resultados obtenidos a 23 °C. Utilizando el método Rietveld y análisis químico fueron determinadas las cantidades y las composiciones químicas, respectivamente, de las fases del cemento anhidro y del calcario para alimentar el programa GEMS y simular la hidratación de la pasta estudiada en laboratorio para un tiempo infinito bajo la influencia de la temperatura. Simulaciones de la hidratación de la pasta se aproximaron a los resultados reportados por la revisión bibliográfica permitiendo predecir la presencia de fases en función de la estabilidad termodinámica. Resultados de porosimetría por inyección de mercurio mostraron una mayor concentración de poros, asociados a defectos, en muestras curadas a 85°C con respecto a las de 23 °C. Finalmente, medidas de variación longitudinal para muestras de fibrocemento elaboradas en laboratorio y en fábrica presentaron dispersión en los resultados de expansión para cada uno de los escenarios sin poder correlacionar todos los datos con el aumento de masa exhibido por cada una de las muestras; no obstante, se resalta el riesgo de deterioro del fibrocemento por el aumento de defectos causado por la temperatura así como la creación de todas las condiciones necesarias que favorecen la formación tardía de etringita a partir del proceso de fabricación empleado para este producto. / The curing temperature using by the fibercement industry as a catalyst in the hydration of cement, hypothesizes expansion caused by delayed ettringite formation. To approach this hypothesis, this study looked at two groups of samples in order to study its chemical and physical behavior individually. One group was composed of cement paste with 35% substitution of limestone filler, and the other by fibercement with addition of PVA and cellulose. In the isothermal calorimetry test was observed the kinetics reaction of both pastes, one cured at 23 °C and the other at 85 °C, the latter showed a catalytic effect in his reactions. XRD and DTG/TG showed the presence of ettringite at 23 °C for all hydration ages exhibiting a trend of increase in the peak intensity and amount, respectively in function of time, however, at 85 °C the presence of this phase was unstable during early ages of hydration until completing one month hydration where showed an increase in each of the following measurements monthly performed. At 85 °C there was a lower formation of portlandite when it was compared with results obtained at 23 °C. Using the Rietveld quantitative method analysis and chemical analysis were determined amounts and chemical compositions, respectively, of the phases of cement anhydrous and limestone filler to feed the GEMS software and simulate the hydration of paste studied in the laboratory for an infinite time under the influence of temperature. Simulations of paste hydration approached the reported data in literature allowing predicting the presence of phases according to thermodynamic stability. Additionally, results of mercury intrusion porosimetry showed a higher concentration of pores associated with defects, in samples cured at 85 ° C compared to 23 °. Finally, were taken measures of length offset of fibercement samples produced in laboratory and factory which had incorporation of PVA fibers and cellulose; they presented dispersion in the results of expansion for each of the scenarios unable to correlate with the mass gain exhibited by each sample. However, the risk of deterioration of the fibercement is highlighted by the increase of defects caused by temperature and the creation of all necessary conditions to favor the late formation of ettringite from the manufacturing process used for this product.

La réaction sulfatique interne (RSI) est une pathologie conduisant au gonflement et à la dégradation du béton de certains ouvrages. En tant que gestionnaire d'ouvrages, VINCI doit assurer la sécurité des usagers et des biens. Pour cela, une évaluation de l'aptitude au service des ouvrages concernés est nécessaire. Ce travail a pour but d'améliorer la modélisation des phénomènes mécaniques induits par la formation d'ettringite différée, notamment sous contrainte. Une campagne expérimentale a été réalisée sur éprouvettes prismatiques en béton réactif libres de contraintes, armées uniaxialement, armées triaxialement et soumises à un chargement uniaxial de compression de 14,5 MPa. Les phénomènes expansifs sont relativement isotropes en l'absence de contraintes et atteignent 0,6 % en fin de réaction. Le bridage induit par la présence des armatures conduit à une diminution importante des déformations dans les directions concernées, sans impact majeur sur le gonflement dans les directions libres : les déformations du béton initiées par la RSI sont de fait anisotropes sous chargement anisotrope. Une fissuration préférentielle est observée parallèlement à la direction gênée. L'application d'un chargement uniaxial de compression à hauteur de 14,5 MPa entraine un raccourcissement du béton dans la direction chargée, les déformations de fluage masquant les éventuelles expansions. La décharge de ces corps d'épreuve après environ 420 jours d'immersion a provoqué une reprise des gonflements de l'ordre de 0,10 % en 3 jours dans la direction chargée, sans effet visible dans les autres directions. Au cours des 200 jours suivants, des gonflements longitudinaux et transversaux similaires en termes de cinétique et d'amplitude ont été observés. Ce phénomène a été attribué au renouvèlement de l'eau de conservation ayant eu lieu au moment de la décharge, en particulier à une réaccélération du lessivage des alcalins. L'endommagement induit par la RSI sur le béton a été caractérisé par le biais d'essais de compression, traction par fendage et mesures de module d'Young avant et après gonflement. [...] / Delayed ettringite formation is a pathology leading to swelling and degradation of the concrete of certain structures. As structure manager, VINCI must ensure the safety of users and goods. To do so, an assessment of the serviceability of affected structures is necessary. This work aims to improve the modeling of mechanical phenomena induced by delayed ettringite formation, especially under stress. An experimental campaign was carried out on prismatic reactive concrete specimens. Some were plain, uniaxially reinforced, triaxially reinforced and subjected to an uniaxial compressive loading of 14.5 MPa. Expansions appeared fairly isotropic in stress-free conditions and reached 0,6 %. Restraint due to reinforcements led to decreased strains in the restrained directions. Expansions were only slightly impacted in transversal free directions. Therefore, DEF expansion under uniaxial stress is anisotropic. Cracks were observed parallel to the restrained direction. For prestressed concrete, creep strains hid possible strains induced by DEF. These specimens were unload after 420 days of immersion in water. It led to a strain increase of about 0,10 % in 3 days in the loaded direction, without any effect on transversal directions. Similar kinetic and range of expansion was measured in both longitudinal and transversal throughout the 200 following days. This phenomenon was understood as the consequence of an alkali leaching acceleration after storage water renewal that occurred during specimens unloading. DEF induced damage on concrete was characterize through compressive tests, brazilian tensile tests and Young modulus measurements before and after expansions. DEF effect on steel-concrete bond behavior was also measured through pull-out tests. Maximal steel-concrete bond stress appears slightly impacted by longitudinal expansions of about 0.25 %, but the bond shear modulus decreases significantly after DEF. All collected experimental data were used to fit a poromechanical model of DEF taking into account all the phenomena involved in the of reinforced concrete behavior (plasticity, damage, creep, shrinkage, distributed reinforcements...). [...]

Properly proportioned, placed and cured concrete can be durable under most conditions. However, deterioration of concrete does occur under certain environments. One of the problems that affect the durability of hardened concrete is delayed ettringite formation (DEF) which is an important problem encountered in precast concrete industry where high temperature curing is applied. Although there had been many researches on DEF, there are still many uncertainties about its chemistry and mechanism. In this study, the effects of partial cement replacement by different mineral admixtures (trass, blast furnace slag and fly ash), SO3/Al2O3 molar ratio and specific surface area of cement on DEF were investigated. For this purpose, 9 groups of control cements were prepared with 3 different specific surface areas and 3 different SO3/Al2O3 molar ratios. Different amounts of mineral admixtures were blended with the control cements. High temperature curing was applied to the cement pastes and the expansions of these pastes were measured periodically for 240 days. v The experimental results obtained were interpreted for a comparative analysis of the effects of the afore-mentioned parameters.

Concrete bridges in Texas have developed large cracks in bent caps and pretensioned trapezoidal bridge girders. The bridges show premature concrete deterioration due to alkali-silica reaction (ASR) and delayed ettringite formation (DEF). There is concern that deterioration due to ASR/DEF may lead to a loss of structural capacity. However, there are no quantitative guidelines to relate the level of concrete deterioration due to ASR/DEF to structural performance. Using such guidelines, the need for rehabilitation of beams with ASR/DEF cracking can be assessed. The goal of this research was to determine the shear capacity of pretensioned trapezoidal box girder specimens exhibiting varying degrees of ASR and/or DEF cracking and to use the shear testing results to evaluate the severity of the problem that may exist in Texas bridge structures. To achieve this goal, beams that were severely deteriorated due to ASR/DEF over a period of more than ten years were transported to the University of Texas for testing to failure. Both severely deteriorated and uncracked beams were tested in shear. The test results were used to evaluate the shear performance of trapezoidal box beams affected by ASR/DEF. In addition, three different types of forensic analyses were conducted on the beams to understand the nature of the ASR/DEF cracks and severity of the deterioration. After testing, it is found that the shear capacity of the test specimens was not significantly reduced even with heavy ASR/DEF cracking. Assessment using current US design provisions for bridges or buildings (ACI 318-08 and AASHTO LRFD 2008) and the proposed provision from an earlier project (TxDOT Project 5253) yielded conservative estimates of strength. Results from forensic analyses provided a qualitative indication of ASR/DEF damage but did not correlate with the observed levels of ASR/DEF deterioration. / text

Across the State of Texas and many other areas of the world, relatively young concrete structures have developed signs of premature concrete deterioration. Large cracks form on the surface of the concrete due to expansive forces from alkali-silica reaction (ASR) and delayed-ettringite formation (DEF). The goal of this project is to assess the effect of ASR/DEF on the trapezoidal box beam bridges in the US 59 corridor and Katy Central Business District (CBD) HOV lanes in Houston, TX. Five dapped-end beams were rejected during the casting process and have been in storage at a local precast yard for nearly fifteen years. These beams have been subject to accelerated deterioration and represent the potential severity of the ongoing ASR/DEF distress within the dapped end regions of the in-service trapezoidal box beams. The results from five load tests, corresponding strut-and-tie models, and forensic investigation are used to provide insights into the relationship between the severity of the deterioration and the capacity margin. / text

Submitted by Erika Demachki (erikademachki@gmail.com) on 2014-10-23T17:14:40Z No. of bitstreams: 2 Dissertação - Mário Henrique Gomes Brito - 2013.pdf: 6257761 bytes, checksum: 3ff568c727b44a8a7e645c8475216a03 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Jaqueline Silva (jtas29@gmail.com) on 2014-10-23T18:50:27Z (GMT) No. of bitstreams: 2 Dissertação - Mário Henrique Gomes Brito - 2013.pdf: 6257761 bytes, checksum: 3ff568c727b44a8a7e645c8475216a03 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2014-10-23T18:50:27Z (GMT). No. of bitstreams: 2 Dissertação - Mário Henrique Gomes Brito - 2013.pdf: 6257761 bytes, checksum: 3ff568c727b44a8a7e645c8475216a03 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2013-04-30 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / The overall objective of this research, which was essentially experimental, is to study the influence of the isothermal period and the maximum temperature of the thermal cure cycle steam under atmospheric pressure in the development of compressive strength of concrete blocks over time. The influences of two other types of curing are also being investigated; curing by enveloping with plastic sheet and open air curing, considered the last case scenario for reference. This study specifically aimed to further our understanding on how to; a) evaluate and quantify the importance of adopting procedures for curing the average resistance (Fbm) and characteristic (fbk) strength to compression of concrete blocks, comparing the processes of thermal curing and curing by enveloping plastic sheet to open air curing; b) evaluate statistically the aging effect of blocks over its compressive strength; c) evaluate the interaction effects of the variables "type of cure" and "age analysis" of the results of compressive strength of concrete blocks; and d) identify the main changes in the microstructure of the blocks subjected to a thermal curing process, especially observing eventual delayed ettringite formation in the internal structure of the material. To examine the influence of maximum curing temperature, two levels were adopted; 65°C and 85°C. As for analyzing the influence of isothermal period, three levels were adopted; 3 hours, 4 hours and 5 hours. The age analysis was evaluated in five levels; 1, 3, 7, 28 and 91 days. The results showed that the worst condition for curing, or curing type, was the open air curing (curing time approximately equal to 24 hours), which led to a lower overall average result of compressive strength of concrete blocks, proving the importance of adopting procedures of curing to optimize the mechanical performance of concrete blocks. Furthermore, it was also determined that the best curing conditions were the curing with enveloping plastic sheet (curing time approximately equal to 24 hours) and the thermal curing with steam under atmospheric pressure isotherm of 65°C and isothermal period of 3 hours (curing time approximately equals 8.5 hours), which were considered statistically equal. It was also observed that there was no significant influence of the curing in maximum temperature in relation to the compressive strength of the blocks. In the other hand, the isothermal period was very significant; the best condition was 3 hours, while 4 hours and 5 hours were considered statistically equal. Regarding the delayed ettringite formation, it was only verified for thermal curing conditions of isothermal of 85°C and isothermal period of 4 hours and 5 hours. / O objetivo geral desta pesquisa, de caráter essencialmente experimental, é estudar a influência do período isotérmico e da temperatura máxima do ciclo de cura térmica a vapor sob pressão atmosférica no desenvolvimento da resistência à compressão de blocos de concreto ao longo do tempo. Paralelamente, foram investigadas ainda as influências de dois outros tipos de cura, a saber: cura por envelopamento com lona plástica e cura ao ar livre, considerada esta última a situação de referência. De modo específico, o estudo visou ainda: a) avaliar e quantificar a importância da adoção de procedimentos de cura nas resistências média (fbm) e característica (fbk) à compressão de blocos de concreto, comparando os processos de cura térmica e cura por envelopamento com lona plástica com a cura ao ar livre; b) avaliar estatisticamente o efeito da idade dos blocos sobre a sua resistência à compressão; c) avaliar a interação dos efeitos das variáveis “tipo de cura” e “idade de análise” sobre os resultados de resistência à compressão dos blocos de concreto; e d) identificar as principais transformações ocorridas na microestrutura dos blocos submetidos aos processos de cura térmica, em especial observando eventual formação de etringita tardia na estrutura interna do material. Para analisar a influência da temperatura máxima de cura, foram adotados dois níveis: 65°C e 85°C. Já para analisar a influência do período isotérmico, foram adotados três níveis: 3 horas, 4 horas e 5 horas. Por sua vez, a idade de análise foi avaliada em cinco níveis: 1, 3, 7, 28 e 91 dias. Os resultados mostraram que a pior condição de cura, ou seja, o tipo de cura que conduziu ao menor resultado médio global de resistência à compressão dos blocos de concreto, foi a cura ao ar livre (tempo de cura aproximadamente igual a 24 horas), comprovando a importância da adoção de procedimentos de cura para a otimização do desempenho mecânico dos blocos de concreto. Além disso, foi verificado ainda que as melhores condições de cura foram a cura por envelopamento com lona plástica (tempo de cura aproximadamente igual a 24 horas) e a cura térmica a vapor sob pressão atmosférica com isoterma de 65°C e período isotérmico de 3 horas (tempo de cura aproximadamente igual a 8,5 horas), os quais foram considerados, estatisticamente, iguais. Também foi verificado que não houve influência significativa da temperatura máxima de cura em relação à resistência à compressão dos blocos. Já o período isotérmico foi significativo, de modo que a melhor condição foi 3 horas, enquanto 4 horas e 5 horas foram considerados, estatisticamente, iguais. Quanto à formação de etringita tardia, esta só foi verificada para as condições de cura térmica com isoterma de 85°C e períodos isotérmicos de 4 horas e 5 horas.

La temperatura de curado, utilizada por la industria de fibrocemento como catalizador en la hidratación del cemento para lograr alta resistencia inicial en sus productos, genera la hipótesis de expansión causada por formación tardía de etringita. Para el planteamiento de dicha hipótesis, el siguiente estudio contempló dos grupos de muestras con el objetivo de estudiar aisladamente su comportamiento químico y físico: uno de los grupos conformado por pasta cementicia con 35% de sustitución de calcario, y el otro conformado por fibrocemento con incorporación de PVA y celulosa. Por calorimetría isotérmica se observó la cinética de reacción de dos pastas, una de ellas curada a 23 ºC y la otra a 85 ºC donde esta última experimentó un efecto catalizador en sus reacciones. Ensayos de DRX y DTG/TG mostraron la presencia de etringita a 23 °C durante todas las edades de hidratación exhibiendo una tendencia de aumento en la intensidad de los picos y en la cantidad, respectivamente, en función del tiempo; sin embargo, a 85 ºC la presencia de esta fase fue inconstante durante las primeras edades de hidratación, a partir de mes de hidratación mostró aumento en cada una de las siguientes medidas realizadas mensualmente; a 85 °C hubo una menor formación de portlandita en comparación con los resultados obtenidos a 23 °C. Utilizando el método Rietveld y análisis químico fueron determinadas las cantidades y las composiciones químicas, respectivamente, de las fases del cemento anhidro y del calcario para alimentar el programa GEMS y simular la hidratación de la pasta estudiada en laboratorio para un tiempo infinito bajo la influencia de la temperatura. Simulaciones de la hidratación de la pasta se aproximaron a los resultados reportados por la revisión bibliográfica permitiendo predecir la presencia de fases en función de la estabilidad termodinámica. Resultados de porosimetría por inyección de mercurio mostraron una mayor concentración de poros, asociados a defectos, en muestras curadas a 85°C con respecto a las de 23 °C. Finalmente, medidas de variación longitudinal para muestras de fibrocemento elaboradas en laboratorio y en fábrica presentaron dispersión en los resultados de expansión para cada uno de los escenarios sin poder correlacionar todos los datos con el aumento de masa exhibido por cada una de las muestras; no obstante, se resalta el riesgo de deterioro del fibrocemento por el aumento de defectos causado por la temperatura así como la creación de todas las condiciones necesarias que favorecen la formación tardía de etringita a partir del proceso de fabricación empleado para este producto. / The curing temperature using by the fibercement industry as a catalyst in the hydration of cement, hypothesizes expansion caused by delayed ettringite formation. To approach this hypothesis, this study looked at two groups of samples in order to study its chemical and physical behavior individually. One group was composed of cement paste with 35% substitution of limestone filler, and the other by fibercement with addition of PVA and cellulose. In the isothermal calorimetry test was observed the kinetics reaction of both pastes, one cured at 23 °C and the other at 85 °C, the latter showed a catalytic effect in his reactions. XRD and DTG/TG showed the presence of ettringite at 23 °C for all hydration ages exhibiting a trend of increase in the peak intensity and amount, respectively in function of time, however, at 85 °C the presence of this phase was unstable during early ages of hydration until completing one month hydration where showed an increase in each of the following measurements monthly performed. At 85 °C there was a lower formation of portlandite when it was compared with results obtained at 23 °C. Using the Rietveld quantitative method analysis and chemical analysis were determined amounts and chemical compositions, respectively, of the phases of cement anhydrous and limestone filler to feed the GEMS software and simulate the hydration of paste studied in the laboratory for an infinite time under the influence of temperature. Simulations of paste hydration approached the reported data in literature allowing predicting the presence of phases according to thermodynamic stability. Additionally, results of mercury intrusion porosimetry showed a higher concentration of pores associated with defects, in samples cured at 85 ° C compared to 23 °. Finally, were taken measures of length offset of fibercement samples produced in laboratory and factory which had incorporation of PVA fibers and cellulose; they presented dispersion in the results of expansion for each of the scenarios unable to correlate with the mass gain exhibited by each sample. However, the risk of deterioration of the fibercement is highlighted by the increase of defects caused by temperature and the creation of all necessary conditions to favor the late formation of ettringite from the manufacturing process used for this product.

Les réactions de gonflements internes (RGI) sont des pathologies qui dégradent le béton de certains ouvrages tels que des barrages en étant à l'origine de gonflements. Electricité De France (EDF) doit assurer la sûreté des personnes et des biens situés à l'aval de ses structures. Ce travail a pour but d'améliorer la modélisation des phénomènes physiques liés à ces dégradations notamment l'interaction entre le produit gonflant et le squelette solide. La contrainte intraporeuse générée par la pression du produit peut créer un endommagement. Dans le modèle proposé, la fissuration peut avoir lieu à deux échelles : - Au niveau microscopique c'est le produit gonflant qui comble le volume libre autour du site de réaction avant de mettre en traction le matériau jusqu'à fissuration, ce qui provoque une baisse des caractéristiques du matériau (résistances en traction et en compression, module d'Young). Un critère de plasticité anisotrope et une loi d'écrouissage positif sont proposés pour modéliser ce phénomène. - Au niveau macroscopique, des gradients de gonflements peuvent se développer à l'intérieur d'ouvrages dont les états de contraintes et les conditions d'humidité et de température ne sont pas homogènes. Une fissuration localisée peut alors s'initier. Elle est la source de risques de dislocations des structures pouvant mener à leur ruine, de concentrations de contraintes dans les aciers structuraux, de chemins facilités pour les agents agressifs extérieurs et pour l'eau, composé essentiel au développement rapide de ces pathologies. La prise en compte des couplages hydro-poro-mécaniques est nécessaire à la simulation de ces phénomènes. Cette fissuration est modélisée par un critère plastique de Rankine anisotrope dont l'énergie dissipée par la fissuration est régularisée par la méthode d'Hillerborg. Après avoir présenté ce modèle, il est validé et appliqué à différentes échelles pour des états de contraintes et des conditions environnementales diverses, sur : - des éprouvettes soumises à des contraintes et des blocages multi-axiaux, - des poutres de laboratoire plus ou moins armées subissant des gradients d'humidité importants, - des barrages dans des conditions réelles construits dans les années 1950. Les résultats obtenus avec le modèle permettent de retrouver les déplacements, les contraintes et les résistances de ces structures après vieillissement. Pour les poutres atteintes de réaction alcali-silice, les résultats numériques obtenus montrent notamment une prise en compte réaliste des contraintes induites par la présence d'armatures, ce qui est confirmé par la validation du comportement à rupture. Néanmoins, des difficultés de calage subsistent pour les poutres subissant la réaction sulfatique interne. Elles sont issues des grandes amplitudes d'expansion et des couplages diffusion-fissuration qui rendent spécifique le comportement de ces structures par rapport à des structures subissant des expansions moins importantes. Ce modèle est un outil pour la requalification des ouvrages dans le but d'assurer leur sûreté. Il peut également servir de base de calcul à la simulation de diverses réhabilitations et travaux de confortements. / Internal swelling reactions (ISR) damage the concrete of structures such as dams. Electricité De France (EDF) must ensure the safety of the people and goods located below the dam. This work aims to improve the modeling of the physical phenomena related to these damages, especially the interaction between the swelling product and the solid skeleton. The intraporous stress generated by the product pressure can create cracking, which, in the proposed model, can occur at two scales: - At a microscopic scale, the swelling product fills the gaps around the reaction site, which induces cracking and a decrease of the materials' characteristics (tensile and compression strengths, Young's modulus). Numerically, an anisotropic plastic criteria and a hardening law allow modelling this phenomenon. - At a macroscopic level, swelling gradients can develop inside concrete structures whose stress states and climatic conditions (humidity and temperature) are not homogeneous. A localized cracking can begin. It is a source of structure dislocation risks and stress concentrations in structural reinforcement. It may ease the way in for outside aggressive agents and water, an essential component for these reactions. Hydro-poro-mechanical couplings need to be taken into account when simulating these phenomena. The cracking is modelled by a Rankine plastic criterion whose dissipated energy is regularized by the Hillerborg's method. After presenting the model, several stress states and environmental conditions are applied to the model in order to validate it on: - samples under multi-axial stresses and restraints, - plain and reinforced beams submitted to moisture gradients in laboratory, - 1950s dams in real conditions. The results of the model make it possible to reproduce the displacements, the stresses and residual strength of these structures after aging. For the beams submitted to alkali aggregate reaction, the numerical results show a relevant prediction of the stress induced by reinforcement, confirmed by the failure behaviour. Nevertheless, the model encounters calibration difficulties for beams affected by delayed ettringite formation. They come from large swelling amplitude and diffusion-cracking coupling which are inherent to these structure behaviours. This model is a tool for concrete structure requalification in order to ensure their safety. It can also be a basis for diverse rehabilitations and reinforcement works simulation.

Ce travail de recherche a pour but d’étudier des matériaux cimentaires soumis à des attaques sulfatiques selon trois différentes conditions : attaques sulfatiques externes (ASE), formation différée d’ettringite communément appelée attaque sulfatique interne et l’effect couplé des deux réactions, ainsi que de proposer un mécanisme uniforme pour les dégradations causées par ces réactions. En se basant sur le mécanisme proposé, un model poro-mécanique est proposé pour simuler l’expansion induite par les produits expansifs néoformés pendant la dégradation. Cette étude inclut trois parties : dans la première partie, les dégradations des éprouvettes de pâte de ciment correspondant à deux types de ciments (CEM I et CEM III) et deux dimensions exposées à trois différents types d’attaques sulfatiques (ASE, ASI, et le couplage des deux réactions), sont étudiés et comparés : variations de longueur, de masse, ainsi que des observations visuelles. Les éprouvettes exposées au couplage ASE-ASI montrent la cinétique d’expansion la plus rapide et le degré de dégradation le plus important, comparé aux autres cas. Ensuite, la structure poreuse des pâtes de ciment avant et après les attaques sulfatiques est caractérisée en utilisant différentes techniques : porosimétrie à mercure (MIP), sorption dynamique de vapeur (DVS), porosité accessible à l’eau ou essais de dissolution par traitement thermique. En comparant les variations de la distribution de la taille des pores des pâtes de ciment exposées à différentes conditions, les cristaux néoformés se trouvent précipités à la fois dans les pores capillaires et les pores des C-S-H. En plus de l’évolution de la distribution de la taille des pores pendant l’ASI, un mécanisme de dégradation est proposé : les cristaux néoformés (l’ettringite) sont précipités dans les grands pores, sans provoquer une expansion manifeste, et ils sont ensuite précipités dans les pores capillaires et les pores des C-S-H, ce qui induit un gonflement. Par ailleurs, le volume des pores occupé par les produits de l’ASI sont libérés après des essais de dissolution par traitement thermique, ce qui confirme la formation de produits d’expansion dans cette gamme de pores. Enfin, en se basant sur les résultats expérimentaux montrant que l’ettringite se forme en allant des grands pores vers les plus petits, un model poro-mécanique est proposé pour simuler l’expansion des matériaux cimentaires soumis à des attaques sulfatiques. Le modèle est basé sur la croissance contrôlée en surface et les propriétés physicochimiques pour l’ASE et l’ASI, malgré les différences entre ces deux réactions. Deux constantes indépendantes : ai et ap sont proposées pour représenter la cinétique de l’invasion des cristaux et la déformation. De plus, le modèle peut être couplé avec toutes les théories mécaniques, par exemple : l’élasticité, la plasticité, la théorie de l’endommagement ou autres. Le modèle illustre bien le processus de cristallisation et il prédit l’expansion correspondante à la fois à l’ASE et l’ASI / This work aims to study cement-based materials subjected to sulfate attacks in three different conditions: External Sulfate Attack (ESA), Delayed Ettringite Formation (DEF) and the Coupling effect of both, and to propose the same damage mechanism for all of them. Based on the proposed mechanism, a poromechanical model is established to simulate the expansion induced by expansive crystals during the degradation. The study includes the following three parts. In the first part, the degradation of cement paste specimens with two kinds of cement type (CEM I and CEM III) and two dimensions (2 2 12 cm3 and 11 11 22 cm3) exposed to three sulfate attack conditions (ESA, DEF, and Coupling effect) are studied and compared, including the length variation, mass variations, and observations. The specimens exposed to the coupling effect show the fastest kinetics and the most serious degree of degradation compared to the other cases. Then, the pore structure of cement pastes before and after sulfate attacks is characterized via different techniques: MIP, DVS, water accessible tests and heat-based dissolution tests. By comparing the variation of pore size distribution of cement pastes exposed to different conditions, the generated crystals are found to be precipitated both in capillary and gel pores. In addition to the evolution of pore size distribution during DEF, a damage mechanism is proposed: the generated crystals (ettringite) precipitate in the big pores without inducing an obvious expansion, and then penetrate into capillary and gel pores, which leads to a swelling. Moreover, the pore volume occupied by DEF induced products is released after heat-based dissolutiontests, which further confirms the formation of expansive products in that porerange. Finally, based on the experimental conclusion that ettringite forms through the large to small pores in all cases, a poromechanical model is proposed to simulate the expansion of cement-based materials submitted to sulfate attacks. The model is based on the surface-controlled growth and physicochemical properties both for ESA and DEF, despite the different source of sulfate ions. Two independent constants, ai and ap, are proposed to represent the kinetics of crystal invasion and deformation. Moreover, the model could be coupled with any mechanical theories, e.g. elasticity, plasticity, damage theory or any other. The model well illustrates the crystallization process and well predicts the corresponding expansion both in ESA and DEF