Rozprawy doktorskie na temat „Behaviour isotropy”

The morphological properties of rock fractures may have a significant influence on their hydromechanical behaviour. Fracture surface roughness could change the fluid flow regime from laminar to turbulent, while it causes the flow properties to deviate from cubic law for smooth channels due to a change in fracture equivalent hydraulic aperture. Different empirical (including the well known Joint Roughness Coefficient, JRC) and statistical methods have been proposed for surface roughness characterisation in an attempt to link them to the hydromechanical behaviour of fractures.This thesis aims to investigate the potential for assessment of fluid behaviour by studying its surface geometrical properties. D[subscript]R[subscript]1 and D[subscript]R[subscript]2, the 2D and 3D roughness parameters developed recently using Riemannian geometry, were used to correlate fracture geometry to its flow behaviour. Also, the 2D Riemannian isotropy parameter (I[subscript]R[subscript]2) was used to correlate surface roughness anisotropy with directionality in fluid flow behaviour along different directions.Numerical simulations in both 2D and 3D were performed assuming the laminar flow regime using FLUENT software. This assumption is, to a large extent, acceptable for situations where the height to length ratios of a fracture is very small. 2D analysis of synthetic profiles with different geometries demonstrated how a change in profile roughness can affect flow response, for example, the pressure drop. JRC flow channels developed in this work as combinations of pairs of JRC profiles were simulated numerically. The analysis results indicated that channels with a similar JRC average for the upper and lower walls but a different JRC profile number responded differently when they were subjected to fluid flow. Therefore, assuming special fluid properties, correlations developed using the pressure drop of a fracture can be estimated by its analogy to JRC flow channels.3D simulations of a corrugated plane were performed assuming different asperity height distributions, for fluid travelling along different directions with respect to surface geometry and at different shear displacements. No asperity contact and failure is assumed in the analysis performed in this work. D[subscript]R[subscript]2 analysis results of the corrugated plane indicated how fluid flow could be related to surface geometry. For instance, it was observed that the pressure drop was maximised along the direction of maximum roughness and reduced to its minimum along a perpendicular direction which shows anisotropy in fluid flow behaviour. Significant changes in pressure drop due to shear offset indicated the importance of fracture wall displacements with respect to each other. A detailed analysis of one synthetically generated surface, and also five surfaces with identical statistical parameters except their correlation distances being different, further confirmed the above concepts. This was followed by analysing a real rock like fracture which was studied elsewhere for fracture shear tests in the lab. Simulation of this surface was performed with particular interest in identifying the locations where the velocity magnitude reduced to nearly zero after the fracture was subjected to a shear offset corresponding to maximum shear stress. These areas were found to be very similar to the locations of asperity degradations as observed through lab experiments. The roughness analysis of the surface was in agreement with the correlation found between the mechanical and hydraulic behaviour of the surface.The results of this research demonstrate how detailed analysis of surface geometry could provide valuable information with respect to surface flow behaviour. Detailed discussions and interpretations of the results will be presented and various conclusions will be made.

Les méthodes d'apprentissage automatique basées sur les réseaux de neurones offrent de nouvelles perspectives pour la modélisation du comportement mécanique des matériaux. Grâce à leur capacité à servir d'interpolateurs universels de fonctions, ces réseaux sont capables de capturer une grande variété de comportements complexes. Cependant, leur mise en œuvre requiert des ensembles de données volumineux, souvent difficilement accessibles par des moyens expérimentaux. Afin de pallier cette limitation, ce manuscrit introduit différents biais physiques permettant de modéliser le comportement mécanique de matériau, et plus spécifiquement le comportement viscoélastique non linéaire, à partir de données expérimentales restreintes.Les deux principes fondamentaux de la thermodynamique constituent un cadre contraignant pour la formulation de lois de comportement. Il permet de réduire le nombre de données nécessaires à l'entraînement des modèles, tout en renforçant leur robustesse face aux erreurs de mesure.Les réseaux de neurones récurrents, quant à eux, sont particulièrement bien adaptés pour modéliser des comportements dépendant de l'histoire du chargement. Leurs mémoires cachées offrent une analogie intéressante avec les variables internes introduites par le principe de l'état local en mécanique. Cependant, ces réseaux posent des défis en matière d'entraînement et de généralisation. Pour surmonter ces difficultés, un modèle de réseau de neurones à encodage mécanique est proposé. Ce modèle tire parti des variables internes d'un modèle de viscoélasticité linéaire pour encoder l'histoire du matériau, ce qui s'avère suffisant pour modéliser le comportement mécanique non linéaire du matériau.Un des défis majeurs de la modélisation en trois dimensions à partir de données expérimentales réside dans la prise en compte des symétries matérielles pour éviter des essais redondants. Dans le cas des matériaux isotropes, une méthode d'augmentation de données par rotation aléatoire des essais, combinée à un apprentissage par transfert latéral, permet de développer une loi de comportement 3D en utilisant seulement deux types d'essais uniaxiaux. Une formulation thermodynamique respectant intrinsèquement l'isotropie du matériau est proposée, bien que des défis d'entraînement restent à surmonter pour optimiser cette approche
The application of machine learning techniques based on neural networks provides novel insights into the modelling of the mechanical behaviour of materials. These networks are capable of capturing a wide variety of complex behaviours due to their ability to act as universal function approximators. However, the deployment of these techniques requires large datasets, which are often difficult to obtain experimentally. This manuscript introduces various physical biases that enable the modelling of mechanical behaviour, specifically non-linear viscoelastic behaviour, using limited experimental data, thereby addressing this limitation.The two fundamental principles of thermodynamics provide a robust framework for constraining the formulation of constitutive laws. This approach reduces the quantity of data required for model training, while simultaneously improving the models' resilience to measurement errors.Recurrent neural networks, on the other hand, are particularly well-suited for modelling behaviour that depends on the loading history. Their hidden memories mirror the internal variables introduced in mechanics by the local state principle. However, these networks present challenges in terms of training and generalisation. To overcome these difficulties, a neural network model with mechanical encoding is proposed. This model employs the internal variables of a linear viscoelasticity model to encode the material's history, which proves to be sufficient for modelling its non-linear mechanical behaviour.One of the most significant challenges in three-dimensional modelling from experimental data is the incorporation of material symmetries in order to avoid the need for superfluous testing. For isotropic materials, a method of increasing the data set by randomly rotating the tests, combined with lateral transfer learning, enables the development of a three-dimensional constitutive law using only two types of uniaxial test. A thermodynamically consistent formulation that inherently preserves the material's isotropy is proposed; however, challenges related to training remain to be addressed in order to optimise this approach

The long term plastic deformation, (creep), behaviour in tension of both oriented and isotropic polyethylene has been studied using a variety of relatively simple experimental techniques, primarily dead loading creep tests. In an attempt to gain an understanding of the mechanisms controlling creep two different viscoelastic models have been applied to the data. The first of these was a two process model, involving two activated Eyring processes in parallel. The second was a Fotheringham and Cherry type model, with a co-operative jump based on a single activated process. In the oriented state the two process model has been successful in describing the behaviour of the material. The Fotheringham and Cherry type model was not successful in describing the behaviour of this material. A complete expression of the two process model has been shown to fully describe the behaviour of one grade at one draw ratio, and an approximation of the model for the other draw ratios and grades has indicated that creep of oriented polyethylene is controlled by a c-shear mechanism. Because of time constraints it was only possible to analyse the behaviour of one grade of the isotropic material with the Fotheringham and Cherry model. Whilst this was successful the activated parameters obtained from it seem very low and would seem to indicate a chain rotation mechanism. Whilst it appears that the two process model should also describe this data, it has not been possible to produce a conclusive fit. It is considered that this is due to limitations in the computing software used for fitting. Hence an approximation of the model was used to produce approximate activation parameters. In addition to the investigation into the rate controlling mechanism of creep the data obtained for the oriented state has been shown to be consistent with the existence of a unique relationship between three important mechanical properties, (strain, strain rate and stress). This relationship holds provided that the initial morphology of the material is equivalent prior to orientation taking place. It has also been shown that the creep data obtained for the isotropic state are consistent with the existence of double yield points in polyethylene. Finally the creep data have been used in a computer model of craze deformation in order to investigate the rate controlling mechanism of slow crack growth. This has shown that craze failure, and hence crack lifetime is controlled by tensile creep to failure of fibrils at the base of the craze.

As a preliminary starting point for the present study, physical and mechanical properties of polypropylene nanocomposites (PPNCs) for samples received from Queen's University Belfast have been evaluated. Subsequently, polymer/clay nanocomposite material has been produced at Bradford. Mixing and processing routes have been explored, and mechanical properties for the different compounded samples have been studied. Clay intercalation structure has received particular attention to support the ultimate objective of optimising tensile and fracture behaviour of isotropic and die-drawn PPNCs. Solid-state molecular orientation has been introduced to PPNCs by the die-drawing process. Tensile stress-strain measurements with video-extensometry and tensile fracture of double edge-notched tensile specimens have been used to evaluate the Young's modulus at three different strain rates and the total work of fracture toughness at three different notch lengths. The polymer composite was analyzed by differential scanning calorimetry, thermogravimetric analysis, polarizing optical microscopy, wide angle x-ray diffraction, and transmission electron microscopy. 3% and 5% clay systems at various compatibilizer (PPMA) loadings were prepared by three different mixing routes for the isotropic sheets, produced by compression moulding, and tensile bars, produced by injection moulding process. Die-drawn oriented tensile bars were drawn to draw ratio of 2, 3 and 4. The results from the Queen's University Belfast samples showed a decrement in tensile strength at yield. This might be explained by poor bonding, which refers to poor dispersion. Voids that can be supported by intercalated PP/clay phases might be responsible for improvement of elongation at break. The use of PPMA and an intensive mixing regime with a two-step master batch process overcame the compatibility issue and achieved around 40% and 50% increase in modulus for 3% and 5% clay systems respectively. This improvement of the two systems was reduced after drawing to around 15% and 25% compared with drawn PP. The work of fracture is increased either by adding nanoclay or by drawing to low draw ratio, or both. At moderate and high draw ratios, PPNCs may undergo either an increase in the size of microvoids at low clay loading or coalescence of microvoids at high clay loading, eventually leading to an earlier failure than with neat PP. The adoption of PPMA loading using an appropriate mixing route and clay loading can create a balance between the PPMA stiffness effect and the degree of bonding between clay particles and isotropic or oriented polymer molecules. Spherulites size, d-spacing of silicate layers, and nanoparticles distribution of intercalated microtactoids with possible semi-exfoliated particles have been suggested to optimize the final PPNCs property.

Pour déterminer la vulnérabilité et la ruine des structures en béton armé soumises à une onde de choc engendrée par une explosion, le centre d'études de Gramat (c. E. G. ) a développé un nouveau modèle de comportement pour le béton. Ce modèle isotrope, base sur la théorie de l'endommagement utilise deux variables scalaires d'endommagement afin de prendre en compte divers phénomènes tels que: la dissymétrie en tension et compression, le caractère unilatéral (ouverture-fermeture de fissures), l'existence de déformations anélastiques, les effets des vitesses de déformation et les effets de frottement. Couplé à une méthode de régularisation de la localisation, ce modèle permet d'assurer une indépendance des résultats vis à vis du maillage. La formulation explicite, sans processus itératif, de cette loi de comportement permet également d'obtenir une certaine rapidité (temps de calcul) et robustesse (instabilités lors de la rupture du béton) lors des simulations numériques. Afin de valider ce modèle de comportement du béton en dynamique, des essais de souffle sur des dalles circulaires en béton arme ont été réalises au c. E. G. A l'aide d'un tube a choc. Des comparaisons calculs/expériences montrent les capacités et possibilités de cette nouvelle formulation

Sedimentary rocks are characterized with very low permeability (in the order of 10-22 m2), low diffusivity, a possible self-healing of fractures, and a good capacity to retard radionuclide transport. In recent years, sedimentary rocks are investigated by many research groups for their suitability for the disposal of radioactive waste. Development of deep geologic repositories (DGRs) for the storage of radioactive waste within these formations causes progressive modification to the state of stress, to the groundwater regime, and to the chemistry of the rock mass. Thermal effects due to the ongoing nuclear activity can cause additional disturbances to the system. All these changes in the system are coupled and time-dependent processes. These coupled processes can result in the development of an excavation damaged zone (EDZ) around excavations. More permeable than the undisturbed rock, the EDZ is likely to be a preferential pathway for water and gas flow. Consequently, the EDZ could be a potential exit pathway for the radioactive waste to biosphere. An investigation of the Hydraulic-Mechanical (HM) and Thermal-Hydraulic-Mechanical-Chemical (THMC) behaviour of sedimentary rock formations is essential for the development of DGRs within such formations. This research work consists of (1) an experimental investigation of the mechanical behaviour of the anisotropic Tournemire argillite, (2) modeling of the mechanical behaviour of the Tournemire argillite, and (3) numerical simulations of the mechanical and hydromechanical behavior of two host sedimentary rocks, the Tournemire argillite and Cobourg limestone, for deep geological repository for nuclear wastes. The experimental program includes the measurements of the physical properties of the Tournemire argillite and its mechanical response to loading during uniaxial compression tests, triaxial compression tests with different confining pressures, unconfined and confined cyclic compression tests, Brazilian tests, and creep tests. Also, acoustic emission events are recorded to detect the initiation and propagation of microcracks within the rock during the uniaxial testing. The approach for modeling the mechanical behaviour of the Tournemire argillite consists of four components: elastic properties of the argillite, a damage model, the proposed concept of mobilized strength parameters, and the classical theory of elastoplasticity. The combination of the four components results in an elastoplastic-damage model for describing the mechanical behaviour of the Tournemire argillite. The capabilities of the model are evaluated by simulating laboratory experiments. Numerical simulations consist of: (1) a numerical simulation of a mine-by-test experiment at the Tournemire site (France), and (2) numerical simulations of the mechanical and hydromechanical behaviour of the Cobourg limestone within the EDZ (Canada). The parameters influencing the initiation and evolution of EDZ over time in sedimentary rocks are discussed.

Der Vortrag beschreibt die Grundlagen der Elasto-Plastizität sowie die softwaretechnische Anwendung mit dem FEM-Programm Creo Simulate bzw. Pro/MECHANICA von PTC. Der erste Teil des Vortrages beschreibt die Charakteristika plastischen Verhaltens, unterschiedliche plastische Materialgesetze, Fließkriterien bei mehrachsiger Beanspruchung und unterschiedliche Verfestigungsmodelle. Im zweiten Vortragsteil werden Möglichkeiten und Grenzen der Berechnung elasto-plastischer Probleme mit der Software dargestellt sowie Anwendungstipps gegeben. Im dritten Vortragsteil schließlich werden verschiedene Beispiele vorgestellt, davon besonders ausführlich das Verhalten einer einachsigen elasto-plastischen Zugprobe vor und nach dem Eintreten der Einschnürdehnung
This presentation describes the basics of elasto-plasticity and its application with the finite element software Creo Simulate (formerly Pro/MECHANICA) from PTC. The first part describes the characteristics of plastic behavior, different plastic material laws, yield criteria for multiaxial stress states and different hardening models. In the second part, the opportunities and limitations of analyzing elasto-plastic problems with the FEM-code are described and user information is provided. The last part finally presents different examples. Deeply treated is the behavior of a uniaxial tensile test specimen before and after elongation with necking appears

碩士
國立中央大學
土木工程研究所
94
Bimrock is a kind of rock which contains the composition of block and matrix, its mechanical properites can be influenced by heterogeneous and anisotropy. This research is in different proportion with cement and kaolin to make transversely isotropic specimen of bimrock, then evolute to probe on mechanical behavior. The content of rock is an important parameter to estimate bimrock, this research use rotary scanner to pick specimen of bimrock and develope the image on the surface, use image process method to estimate the raito of specimen of bimrock. The result shows that the increase of confining pressure cause he destructive force to raise,and heterogeneous strength to reduce. This research use Tien and Kuo(2001) failure criterion and FLAC program to analyze the strength、heterogeneous and failure mode.the result show that prediction and experimental data are fairly agreeable.

碩士
國立中央大學
土木工程研究所
92
The main purpose of this research is to study the mechanical properties and behaviors of bimrocks that composed of blocks in matrix. An uniaxial bi-directional compaction mold were designed to prepare the artificial bimrocks. By using different composite materials, the mechanical properties and behaviors with distinct block proportions and stiffnesses of the block and matrix are investigated. For the testing results, the mechanical properties of bimrock are controlled by matrix itself or the interface between block and matrix in uniaxial compression. With increasing confining pressures, the global mechanical properties and behaviors are controlled by both blocks and matrix in triaxial compression. Generally, increasing of block proportions decreased the cohesion and increased the internal friction angle and the material parameter m in Hoek-Brown criteria. Based on the experimental observation, the failure modes of the artificial bimrocks at different block proportions and confining pressures can be classified into four categories: axial splitting mode; shear fracture mode; conjugate shear mode with a main shear plane, and multiple conjugate shear planes mode. For theoretical prediction, the prediction models with assuming perfectly bonded interfaces in composite materials can fit the young’s modulus of the specimens at lower block proportions in uniaxial compression, but the prediction models can’t fit the data at higher block proportions. In triaxial compression, the micromechanics model can be used to predict the test data well.

碩士
長榮大學
土地管理與開發學系碩士班
96
In this research, isotropic and anisotropic consolidation triaxial undrained test was used to test Mai Liao sand. The mechanical characteristics of the saturated sand were investigated under various initial conditions and stress paths. Under the initial conditions: sand sample with three different relaqtive densities of 40%, 50%, 60%, cyclic undrained shear test was performed. In isotropic case, the initial effective normal stresses were 100, 200, 300kPa. In anisotropic case, the initial values of lateral earth pressure coefficient kc were 0.4, 0.5, 0.67 under various effective lateral and normal stresses. The results showed that for saturated sand with different relative density the larger the relative density, the higher shear resistance and damage was not easy to occur. When stress ratio was so great that the sand sample failed, phase transformation occurred in both compression and tension sides. Loose sand showed plastic contraction (corresponding to elastic dilatancy), while dense sand shows plastic dilatancy (corresponding to elastic contraction). That is, with different relative density, plastic contraction or dilatancy may occur. In the beginning of shear test, the pore pressure increased rapidly, and then slowly or even decreased with the increment of shear strain. Therefore, the changes of plastic contraction or dilatancy, and pore pressure depend on relative density of the sand and effective confined pressure. The results in anisotropic case showed that the mechanical behavior of plastic contraction or dilatancy were the same as that in isotropic case. When the value of kc was changed, the residual stress path always followed the stress path of damage curve. It also showed that with the same relative density and different kc, initial effective lateral stresses did not affect stress path of damage of the sample in both isotropic and anisotropic cases.

碩士
國立中央大學
土木工程研究所
96
This research establishes a methodology of stress transformation by the concept of Mohr''s stress circle as well as the formula of stress transformation. This method can overcome the problem in the past research, which the rock condition will not be selfsame after taking samples in transversely isotropic rock. The methodology of the stress transformation makes use of numerical program (FLAC) to carry on the numerical simulation, that only needs to change stress increment of normal stress and shear stress on boundaries, which can create a series of triaxial tests in the same numerical model to investigate the mechanical properties of those transversely isotropic rocks in different types of dip angle formation. Therefore, it is convinced to make use of this method to probe into the result of mechanical behavior of the transversely isotropic rock. This research utilizes the methodology of stress transformation, and observes the numerical results of different transversely isotropic rock materials. Furthermore, the rock material properties and volumetric fraction of block are important parameters to estimate mechanical behavior of transversely isotropic rock, and used to establish the prediction model of the mechanical behavior of transversely isotropic rock. The results show that the prediction agrees with numerical results, which can largely reduce the experiment as well as the numerical analysis, showing good improvement in prediction of mechanical behavior of transversely isotropic rock.

碩士
國立交通大學
光電科技學程
107
The purpose of this study is to adopt dielectric spectroscopy to observe the change of optical textures and dielectric behaviors of liquid crystals (LCs) injected into sandwich-type cells with distinct types of spacers (rod and ball) during the isotropic to LC phase transition. It is found that the spacer will lead to a textural formations and generate a pretransitional (PT) phase for both rod- and ball-spacer-contained LC cells without affecting their dielectric constants. In order to confirm the effect of spacer on observed results, we also fabricated a counterpart in which there is no spacer in the electrode area. The results indicated that the PT phase is absent in the no-spacer LC cell and the regularity of the dielectric constant is different from those of cells with spacers (rod type). On this basis, the dielectric properties and textural transitions of various nematic LCs, including CYLC31 and HDE with positive dielectric anisotropy (+∆) and MLC6608 and ZLI2806 with negative dielectric anisotropy (–∆) injected in rod-spacer-contained cells. The result shows that the regularity of dielectric spectrum changed as a function of the temperature is affected by the magnitude of ∆, In other words, the larger ∆ of positive LCs in rod cells will lead to the induction of PT phase easier. Finally, we investigate the formation of PT phase in spacer-contained CLC cells by doping various contents of chiral agent R5011 in nematic host E44. By dielectric spectrum and textural observation result, it shows that the dielectric signal for the determination of the PT phase in the ’(T) becomes intense and the clearing temperature point gets decreased with increasing dopant concentration of R5011 in E44. As a result, we conclude that the generation of PT state in spacer-contained LC cells is determined by the magnitude of ∆ and the texture formation is affected by the shape of spacer.