Une fois déposé, un sédiment est affecté au cours de son enfouissement par un ensemble de processus, regroupé sous le terme diagenèse, le transformant parfois légèrement ou bien suffisamment pour le rendre méconnaissable. Ces modifications ont des conséquences sur les propriétés pétrophysiques qui peuvent être positives ou négatives, c'est-à-dire les améliorer ou bien les détériorer. Une voie alternative de représentation numérique des processus, affranchie de l'utilisation des réactions physico-chimiques, a été adoptée et développée en mimant le déplacement du ou des fluides diagénétiques. Cette méthode s'appuie sur le principe d'un automate cellulaire et permet de simplifier les phénomènes sans sacrifier le résultat et permet de représenter les phénomènes diagénétiques à une échelle fine. Les paramètres sont essentiellement numériques ou mathématiques et nécessitent d'être mieux compris et renseignés à partir de données réelles issues d'études d'affleurements et du travail analytique effectué. La représentation des phénomènes de dolomitisation de faible profondeur suivie d'une phase de dédolomitisation a été dans un premier temps effectuée. Le secteur concerne une portion de la série carbonatée de l'Urgonien (Barrémien-Aptien), localisée dans le massif du Vercors en France. Ce travail a été réalisé à l'échelle de la section afin de reproduire les géométries complexes associées aux phénomènes diagénétiques et de respecter les proportions mesurées en dolomite. De plus, la dolomitisation a été simulée selon trois modèles d'écoulement. En effet, la dédolomitisation étant omniprésente, plusieurs hypothèses sur le mécanisme de dolomitisation ont été énoncées et testées. Plusieurs phases de dolomitisation per ascensum ont été également simulées sur des séries du Lias appartenant aux formations du groupe des Calcaire Gris, localisées au nord-est de l'Italie. Ces fluides diagénétiques empruntent le réseau de fracturation comme vecteur et affectent préférentiellement les lithologies les plus micritisées. Cette étude a permis de mettre en évidence la propagation des phénomènes à l'échelle de l'affleurement. / Once deposited, sediment is affected by diagenetic processes during their burial history. These diagenetic processes are able to affect the petrophysical properties of the sedimentary rocks and also improve as such their reservoir capacity. The modelling of diagenetic processes in carbonate reservoirs is still a challenge as far as neither stochastic nor physicochemical simulations can correctly reproduce the complexity of features and the reservoir heterogeneity generated by these processes. An alternative way to reach this objective deals with process-like methods, which simplify the algorithms while preserving all geological concepts in the modelling process. The aim of the methodology is to conceive a consistent and realistic 3D model of diagenetic overprints on initial facies resulting in petrophysical properties at a reservoir scale. The principle of the method used here is related to a lattice gas automata used to mimic diagenetic fluid flows and to reproduce the diagenetic effects through the evolution of mineralogical composition and petrophysical properties. This method developed in a research group is well adapted to handle dolomite reservoirs through the propagation of dolomitising fluids and has been applied on two case studies. The first study concerns a mid-Cretaceous rudist and granular platform of carbonate succession (Urgonian Fm., Les Gorges du Nan, Vercors, SE France), in which several main diagenetic stages have been identified. The modelling in 2D is focused on dolomitisation followed by a dedolomitisation stage. For the second study, data collected from outcrops on the Venetian platform (Lias, Mont Compomolon NE Italy), in which several diagenetic stages have been identified. The main one is related to per ascensum dolomitisation along fractures. In both examples, the evolution of the effects of the mimetic diagenetic fluid on mineralogical composition can be followed through space and numerical time and help to understand the heterogeneity in reservoir properties.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2016-06-08T11:17:37Z No. of bitstreams: 1 ricardosilvacampos.pdf: 8528381 bytes, checksum: 29e3f07b2a4b4215d4e42d012e0f5df3 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-07-13T13:24:53Z (GMT) No. of bitstreams: 1 ricardosilvacampos.pdf: 8528381 bytes, checksum: 29e3f07b2a4b4215d4e42d012e0f5df3 (MD5) / Made available in DSpace on 2016-07-13T13:24:53Z (GMT). No. of bitstreams: 1 ricardosilvacampos.pdf: 8528381 bytes, checksum: 29e3f07b2a4b4215d4e42d012e0f5df3 (MD5) Previous issue date: 2016-02-15 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / CNPq - Conselho Nacional de Desenvolvimento Científico e Tecnológico / Este trabalho apresenta o simulador FisioPacer, que é um simulador que reproduz a propagação do pulso elétrico pelo tecido cardíaco e a sua deformação mecânica. Foi utilizado um autômato celular acoplado a um sistema massa-mola para que as simulações sejam realizadas rapidamente. Foi também utilizado um algoritmo genético para automaticamente determinar parâmetros do modelo de forma a reproduzir outros experimentos in silico e o comportamento de um ventrículo real. Com intuito de validar o modelo foram feitos setenta e dois experimentos e os resultados foram comparados com outro simulador robusto, baseado em equações diferenciais. As comparações mostraram que o FisioPacer reproduziu satisfatoriamente o comportamento do tecido, sendo até quinze mil vezes mais rápido. Além disto, foram simuladas as funcionalidades eletromecânicas de um ventrículo esquerdo a partir de dados de um paciente, obtidos via ressonância magnética. / This work proposes a computational heart model named FisioPacer, which aims to reproduce the electrical pulse propagation over the cardiac tissue and its mechanical deformation. In order to perform fast simulations, it was used a cellular automaton coupled with a mass-spring system. A genetic algorithm was also used to automatically adjust model parameters, in order to reproduce in silico experiments and a real left ventricle behavior. For the model validation, seventy two experiments were performed and the results were compared to another robust simulator, based on partial differential equations. The comparisons showed that the FisioPacer simulator could reproduce cardiac tissue electromechanics, with up to 15000-fold improvement in computational time. Furthermore, a real patient left ventricle was simulated, with data obtained via MRI.

Les transferts de chaleur et de masse sont deux phénomènes physiques à la base de nombreux systèmes thermiques employés dans des secteurs variés tels que l'industrie, le bâtiment ou encore les énergies renouvelables. Les présents travaux de recherche envisagent différentes méthodologies d'optimisation de configurations assurant le transfert de flux de chaleur, couplé ou non à un écoulement fluide, au sens topologique du terme. Les équations aux dérivées partielles décrivant les phénomènes physiques sont discrétisées avec la méthode des volumes finis. La première partie du manuscrit examine successivement trois classes différentes de méthodes: la théorie constructale, les automates cellulaires et les méthodes par pénalisation. Le même cas académique, portant sur le refroidissement d'un volume fini générant de la chaleur, est résolu au moyen de ces trois méthodes, ce qui permet ainsi de comparer les performances de chaque algorithme. Cette comparaison démontre l'ascendant des méthodes par pénalisation sur les deux premiers types, tant structurellement que quantitativement, et permet également d'établir des solutions basées sur des compromis dans le cadre d'optimisations multi-objectifs. Par conséquent, la seconde partie envisage l'application de cette approche à des configurations réalisant des transferts de chaleur conducto-convectifs en régime laminaire. L'utilisation de paramètres de pénalisation en conjonction avec les volumes finis requiert une régularisation de la dissipation visqueuse le long de l'interface fluide/solide. Une approche bi-objectif est développée visant à minimiser la puissance dissipée par le fluide, tout en maximisant l'énergie thermique récupérée sur le système. Les solutions obtenues adoptent des configurations non-triviales qui sont divisibles en quatre classes topologiques différentes. La thèse ouvre ainsi un nouveau champ d'investigation pour l'optimisation d'écoulements couplés à la problématique du transport de chaleur. / Heat and mass transfers are two physical phenomena at the base of many thermal systems involved in various fields, such as industries, buildings or renewable energies. The present researches tackle different optimization methodologies of structures subject to heat transfers, coupled with a fluid flow or not, in the topological sense of the term. The partial differential equations describing the physical phenomena are discretized thanks to the finite volume method. The first part of the thesis successively studies three different classes of approaches: constructal theory, cellular automaton, and the solid isotropic material with penalization method. The same academic case, aiming at the optimal cooling of a finite-size volume generating heat, is tackled by means of these three methods, allowing the comparison of the performances of each algorithm. This comparison shows that the method based on the material penalization performs better than the first two approaches, structurally and quantitatively, and also establishes solutions based on a trade-off in the frame multi-objective optimization. Consequently, the second part applies this method to configurations subject to heat and mass transfers with laminar flows. The use of penalization parameters in conjunction with the finite volume method requires a regularization of the viscous dissipation along the solid/fluid interfaces. A bi-objective approach is implemented in order to minimize the total power dissipated by the fluid, while maximizing the thermal energy recovered from the system. The solutions show non-trivial configurations that can be categorized in four different topological classes. The present researches open a new investigation field for fluid flows coupled with the problem of heat transport.

La gestion agroécologique durable des paysages agricoles constitue un très fort enjeu social, écologique et économique. Favoriser les différents services écosystémiques que fournissent ces systèmes complexes, caractérisés par une forte variabilité spatio-temporelle liée aux activités anthropiques, est désormais crucial. Le travail développé dans cette thèse s’est focalisé sur les services de régulation des populations de ravageurs, en développant des modèles spatialisés destinés à identifier les interactions entre paysage et populations dans l’optique d’une gestion intégrée du paysage agricole. L’approche conceptuelle utilisée s’est appuyée sur une réflexion sur les choix de niveaux d’organisation et d’échelles spatio-temporelles pertinentes qui a conduit au développement de deux modèles imbriqués : un simulateur de paysages agricoles (ATLAS) reproduisant la variabilité spatio-temporelle des habitats et un automate cellulaire décrivant les dynamiques pluriannuelles d’un puceron des céréales : Rhopalosiphum padi. Une analyse statistique des résultats a permis d’identifier l’effet des différents paramètres du paysage (composition et climat) sur les densités de pucerons à différentes échelles spatiales au cours des saisons. Ont également été évalués les effets potentiels de différents scénarios de gestion plausibles au sein du paysage agricole étudié. Parmi nos résultats, nous identifions que le remplacement du maïs par le sorgho dans le sud-ouest de la France pourrait conduire à une augmentation des densités de R.padi au sein du paysage au printemps, période particulièrement sensible aux infestations. Nous replaçons ce travail dans le cadre plus large de la favorisation des services écosystémiques et apportons des premiers éléments de réflexion pour une gestion durable du paysage agricole; / Sustainable management of agricultural landscapes is a strong social, ecological and economic challenge. Favouring the different ecosystem services provided by these complex systems, characterized by a strong spatio-temporal variability linked to human activity, is by now crucial. The work developed in this thesis has focused on pest population regulation services, by developing spatialized models in order to identify the interactions between the landscape and populations in the context of an integrated management of the agricultural landscape. The conceptual approach we used is based on careful selection of the levels of organization and spatio-temporal scales which lead to the development of two embedded models: an agricultural landscape simulator (ATLAS) reproducing the spatial-temporal variability of habitats and a cellular automaton describing thepluriannual dynamics of a cereal aphid: Rhopalosiphum padi. A statistical analysis of the results highlighted the effects of different landscape parameters (composition and climate) on aphid densities at different spatial scales throughout the seasons. The potential effects of plausible management scenarios were also evaluated. Amongst our results, we identify the replacement of corn by sorghum in the south-western part of France to potentially lead to higher aphid densities within the landscape in spring, which is a season specifically sensible to infestations. We replace this work in the larger context of favouring ecosystem services and propose new ideas for a sustainable management of agricultural landscapes;

Ant algorithms have been used for a variety of combinatorial optimization problems. One of these problems, where ant algorithms haven't been used, is the design of transition rules for cellular automata (CA). Which is a problem that this master's thesis is focused on. This work begins with an introduction into ant algorithms and a overview of its applications, followed by an introduction into CA. In the next part the author proposes a way how to encode rules of CA into a graph which is used in ant algorithms. The last part of this thesis contains an application of encoded graph on elitist ant system and MAX-MIN ant system. This is followed by experimental results of creating transition rules for CA problems by these algorithms.

Focus of this master's thesis is on evolutionary design of cellular automata and it's optimalization. There are described Evolutionary algorithms and Cellular automata in first part. Thereafter, one of the new ways of transition function representation and its possible evolutionary design is presented. Name of this method is Conditionally Matching Rules. This is followed by description of realized experiments with detailed results. Finally, success with optimalization for some tasks is presented in last chapter along with discussion.

<p>The design of vehicle components for crashworthiness is one of the most challenging problems in the automotive industry. The safety of the occupants during a crash event relies on the energy absorption capability of vehicle structures. Therefore, the body components of a vehicle are required to be lightweight and highly integrated structures. Moreover, reducing vehicle weight is another crucial design requirement since fuel economy is directly related to the mass of a vehicle. In order to address these requirements, various design concepts for vehicle bodies have been proposed using high-strength steel and different aluminum alloys. However, the price factor has always been an obstacle to completely replace regular body steels with more advanced alloys. To this end, the integration of numerical simulation and structural optimization techniques has been widely practiced addressing these requirements. Advancements in nonlinear structural design have shown the promising potential to generate innovative, safe, and lightweight vehicle structures. In addition, the implementation of structural optimization techniques has the capability to shorten the design cycle time for new models. A reduced design cycle time can provide the automakers with an opportunity to stay ahead of their competitors. During the last few decades, enormous structural optimization methods were proposed. A vast majority of these methods use mathematical programming for optimization, a method that relies on availability sensitivity analysis of objective functions. Thus, due to the necessity of sensitivity analyses, these methods remain limited to linear (or partially nonlinear) material models under static loading conditions. In other words, these methods are no able to capture all non-linearities involved in multi-body crash simulation. As an alternative solution, heuristic approaches, which do need sensitivity analyses, have been developed to address structural optimization problems for crashworthiness. The Hybrid Cellular Automaton (HCA), as a bio-inspired algorithm, is a well-practiced heuristic method that has shown promising capabilities in the structural design for vehicle components. The HCA has been continuously developed during the last two decades and designated to solve specific structural design applications. Despite all advancements, some fundamental aspects of the algorithm are still not adequately addressed in the literature. For instance, the HCA numerically implemented as a closed-loop control system. The local controllers, which dictate the design variable updates, need parameter tuning to efficiently solve different sets of problems. Previous studies suggest that one can identify some default values for the controllers. However, still, there is no well-organized strategy to tune these parameters, and proper tuning still relies on the designer’s experience.</p> <p> </p> <p> Moreover, structures with multiple materials have now become one of the perceived necessities for the automotive industry to address vehicle design requirements such as weight, safety, and cost. However, structural design methods for crashworthiness, including the HCA, are mainly applied to binary structural design problems. Furthermore, the conventional methods for the design of multi-material structures do not fully utilize the capabilities of premium materials. In other words, the development of a well-established method for the design of multi-material structures and capable of considering the cost of the materials, bonding between different materials (especially categorical materials), and manufacturing considering is still an open problem. Lastly, the HCA algorithm relies only on one hyper-parameter, the mass fraction, to synthesize structures. For a given problem, the HCA only provides one design option directed by the mass constraint. In other words, the HCA cannot tailor the dynamic response of the structure, namely, intrusion and deceleration profiles.</p> <p> </p> <p>The main objective of this dissertation is to develop new methodologies to design structures for crashworthiness applications. These methods are built upon the HCA algorithm. The first contribution is about introducing s self-tuning scheme for the controller of the algorithm. The proposed strategy eliminates the need to manually tune the controller for different problems and improve the computational performance and numerical stability. The second contribution of this dissertation is to develop a systematic approach to design multi-material crashworthy structures. To this end, the HCA algorithm is integrated with an ordered multi-material SIMP (Solid Isotropic Material with Penalization) interpolation. The proposed multi-material HCA (MMHCA) framework is a computationally efficient method since no additional design variables are introduced. The MMHCA can synthesize multi-material structures subjected to volume fraction constraints. In addition, an elemental bonding method is introduced to simulate the laser welding applied to multi-material structures. The effect of the bonding strength on the final topology designs is studied using numerical simulations. In the last step, after obtaining the multi-material designs, the HCA is implemented to remove the desired number of bonding elements and reduce the weld length.</p> <p> </p> <p>The third contribution of this dissertation is to introduce a new Cluster-based Structural Optimization method (CBSO) for the design of multi-material structures. This contribution introduces a new Cluster Validity Index with manufacturing considerations referred to as CVI_m. The proposed index can characterize the quality of the cluster in structural design considering volume fraction, size, interface as a measure of manufacturability. This multi-material structural design approach comprises three main steps: generating the conceptual design using adaptive HCA algorithm, clustering of the design domain using Multi-objective Genetic Algorithm (MOGA) optimization. In the third step, MOGA optimization is used to choose categorical materials in order to optimize the crash indicators (e.g., peak intrusion, peak contact force, load uniformity) or the cost of the raw materials. The effectiveness of the algorithm is investigated using numerical examples.</p>

The focus of this thesis is the numerical study of subcritical transition to turbulence in boundary-layer flows. For the most part, boundary layers with uniform suction are considered. Constant homogeneous suction counteracts the spatial growth of the boundary layer, rendering the flow parallel. This enables research approaches which are not feasible in the context of spatially developing flows. In the first part, the laminar–turbulent separatrix of the asymptotic suction boundary layer (ASBL) is investigated numerically by means of an edge-tracking algorithm. The obtained edge states experience recurrent dynamics, going through calm and bursting phases. The self-sustaining mechanism bears many similarities with the classical regeneration cycle of near-wall turbulence. The recurrent simple structure active during calm phases is compared to the nucleation of turbulence events in bypass transition originating from delocalised initial conditions. The implications on the understanding of the bypass-transition process and the edge state's role are discussed. Based on this understanding, a model is constructed which predicts the position of the nucleation of turbulent spots during free-stream turbulence induced transition in spatially developing boundary-layer flow. This model is used together with a probabilistic cellular automaton (PCA), which captures the spatial spreading of the spots, correctly reproducing the main statistical characteristics of the transition process. The last part of the thesis is concerned with the spatio-temporal aspects of turbulent ASBL in extended numerical domains near the onset of sustained turbulence. The different behaviour observed in ASBL, i.e. absence of sustained laminar–turbulent patterns, which have been reported in other wall-bounded flows, is associated with different character of the large-scale flow. In addition, an accurate quantitative estimate for the lowest Reynolds number with sustained turbulence is obtained / <p>QC 20160429</p>

Dans cette thèse, on s'intéresse aux automates cellulaires en tant que modèle de calcul permettant de reconnaître des langages. Dans un tel domaine, il est toujours difficile d'établir des résultats négatifs, typiquement de prouver qu'un langage donné n'est pas reconnu en une certaine fonction de temps par une certaine classe d'automates. On se focalisera en particulier sur les classes de faible complexité comme le temps réel, au sujet desquelles de nombreuses questions restent ouvertes.Dans une première partie, on propose plusieurs manières d'affaiblir encore les classes de langages étudiées, permettant ainsi d'obtenir des exemples de résultats négatifs. Dans une seconde partie, on montre un théorème d'accélération par automate cellulaire d'un modèle séquentiel, les automates finis oublieux. Ce modèle est une version a priori affaiblie, mais non triviale, des automates finis à plusieurs têtes de lecture. / This document deals with cellular automata as a model of computation used to recognise languages. In such a domain, it is always difficult to provide negative results, that is, typically, to prove that a given language is not recognised in some function of time by some class of automata. The document focuses in particular on the low-complexity classes such as real time, about which a lot of questions remain open since several decades.In a first part, several techniques to weaken further still these classes of languages are investigated, thereby bringing examples of negative results. A second part is dedicated to the comparison of cellular automata with another model language recognition, namely multi-head finite automata. This leads to speed-up theorem when finite automata are oblivious, which makes them a priori weaker than in the general case but leaves them a nontrivial power.

Investigações da dinâmica crítica em modelos de magnetismo, para tempos curtos, têm aparecido com grande freqüência na literatura. Essa técnica foi descoberta por Li, Schülke e Zheng que, inspirados em trabalhos anteriores de Huse e Janssen et al., mostraram que generalizações de grandezas como a magnetização e o cumulante de Binder exibem comportamento universal já no início da simulação. O estudo da criticalidade em tempos curtos proporciona um caminho alternativo para a estimativa do expoente z, além de permitir o cálculo de um novo expoente dinâmico &#952;, associado ao comportamento anômalo da magnetização. Da mesma forma, simulações dependentes do tempo tornaram-se ferramenta útil para estudar transições de fase em autômatos celulares e modelos de spin. Em particular, as melhores estimativas para o expoente z do Ising bidimensional foram obtidas por meio da técnica de propagação de danos, introduzida por Kauffman no estudo de autômatos e mais tarde generalizada para modelos de spin. Na primeira parte deste trabalho utilizamos o método Monte Carlo em tempos curtos para investigar o modelo de Baxter-Wu, definido em uma rede bidimensional triangular com variáveis do tipo Ising, acopladas por interações de três corpos. Obtivemos os expoentes críticos dinâmicos z e &#952; além dos índices críticos estáticos ß e Nû. Os resultados não corroboram aqueles recentemente obtidos por Santos e Figueiredo para o expoente z. Na segunda parte do trabalho, investigamos a propagação de danos no modelo de Ising unidimensional submetido a duas dinâmicas propostas por Hinrichsen e Domany (HD). Em particular, nós estudamos o efeito da atualização síncrona (paralela) e assíncrona (dinâmica contínua) sobre o espalhamento do dano. Mostramos que o dano não se propaga quando a segunda dinâmica é implementada de forma assíncrona. Também mostramos que as regras para atualização do dano produzidas por essa dinâmica, quando a temperatura vai a infinito e um certo parâmetro Lambda é igual a zero, são equivalentes àquelas do bem conhecido autômato celular (modelo A) de Grassberger. / Short-time simulations have been used with great frequency in the literature. That technique was discovered by Li, Shülke and Zheng that, inspired in previous works by Huse and Janssen et al., showed that generalizations of quantities like magnetization and the Binder´s cumulant exhibit universal behavior in the beginning of the simulation (early time behavior). The study of criticality in short-times provides an alternative way to estimate the dynamic critical exponent z, besides allowing the calculation of a new dynamic exponent &#952;, associated to the anomalous behavior of the magnetization. In the same way, time-dependent simulations became a useful tool to study phase transitions in cellular automata and also for spin models. In fact, the best estimates for the exponent z of the two-dimensional Ising model were obtained through the technique of damage spreading, introduced by Kauffman in the study of cellular automata, later widespread for spin models. In the first part of this work we used short-time Monte Carlo simulations to investigate the Baxter-Wu model, defined in a triangular lattice whose variables are Ising-like coupled by triplet interactions. We have obtained estimates for the dynamic critical exponents z and &#952; besides static exponents ß e Nû. Our results do not corroborate recent estimates by Santos and Figueiredo for the critical exponent z. In the second part of this work, we investigated the damage spreading in the one-dimensional Ising model under two dynamics introduced by Hinrichsen and Domany (HD). In particular, we study the effects of synchronous (parallel) and asynchronous (continuous dynamics) updating on the spreading properties. We showed that the damage does not spread when the second dynamic is implemented in an asynchronous way. We found that the rules for updating the damage produced by this dynamic, as the temperature goes to infinity and a certain parameter Lambda is zero, are equivalent to those of Grassberger’s well-known model A cellular automaton.