Cette thèse porte sur la modélisation numérique des écoulements ouverts de convection naturelle au sein d'un canal vertical asymétriquement chauffé à flux constant. Elle s'inscrit dans un contexte national de comparaison associant approches numériques (benchmark de la Société Française de Thermique SFT) et expérimentales. La particularité de ce type d'écoulement réside dans le fait que le moteur du mouvement est situé au sein même de l'écoulement, rendant alors difficile la modélisation des interfaces et par conséquent la définition des conditions aux limites à appliquer aux frontières ouvertes du domaine de calcul. Nous proposons ici deux approches numériques de modélisation de ce type d'écoulement. La première approche consiste à inclure à la fois le canal vertical et son environnement extérieur dans le domaine de calcul. Cette approche intègre les interactions canal - environnement extérieur de manière implicite et nous permet d'obtenir une description complète de l'écoulement et ainsi de caractériser les interfaces du canal. Sur la base de ces simulations, des solutions numériques de référence modélisant un écoulement de convection naturelle dans un canal vertical immergé dans un environnement infini sont établies. La deuxième approche consiste à restreindre le domaine de calcul aux limites géométriques du canal. Plusieurs méthodes de résolution et modélisation des interfaces sont alors proposées et comparées avec les solutions de référence précédentes. Une approche originale basée sur des conditions limites de type Robin pour la pression motrice montre ainsi de très bon accords avec les solutions de référence. / This thesis focuses on the numerical modeling of natural convection flows in a vertical channel asymmetrically heated at constant heat flux. It takes place in a national context of comparison of numerical approaches (benchmark of the French Thermic Society SFT) and experimental approaches. The main issues result in the fact that the driving forces lie within the computational domain so inlet and outlet flow cannot be a priori prescribed. Therefore it is difficult to model the interfaces and to prescribe boundary conditions at the open frontiers of the computational domain. We propose two numerical approaches for modeling this type of flow. In the first approach the numerical domain includes the vertical channel and its external surroundings in the computational domain. This approach implicitly integrates interactions between the channel and its external environment. This allows us to obtain a complete description of the flow and thus to characterize the channel interfaces. Based on these simulations, numerical reference solutions which are modeling a natural convection flow in a vertical channel immersed in a infinite environment are established. In the second approach the computational domain is restricted to the geometric limits of the channel. Several methods for the numerical resolution and for modeling of the interfaces are proposed and then compared with the previous reference solutions. An original modeling based on Robin boundary conditions for the driving pressure is described and shows very good agreement with the reference solutions.

[pt] Neste trabalho foi desenvolvido um programa para simulação numérica do fluxo bifásico em meios porosos com transferência de massa entre as duas fases e transporte dos solutos em cada fase. O programa calcula o aumento da mobilidade da fase não aquosa e da solubilidade de seus componentes na água como resultado da adição de um co- solvente. O objetivo foi simular problemas de contaminação no subsolo com líquidos imiscíveis com a água e a remediação através da técnica de injeção de co-solventes, assim como a contaminação a partir de vazamentos de gasolina acrescida de álcool em postos de abastecimento. Podem-se ainda resolver problemas de fluxo bifásico envolvendo uma fase gasosa. Foi desenvolvido e implementado um modelo para a transferência de massa durante o fluxo das duas fases. O cálculo das propriedades das duas fases ao longo do tempo é feito utilizando o diagrama ternário de fases. Para resolver as equações do fluxo e do transporte de solutos foi empregado o método de elementos finitos. / [en] A computer code was developed for numerical simulation of two-phase flow in porous media considering mass transfer between the phases involved and solute transport in each phase. Enhanced mobility of nonaqueous phase and enhanced aqueous solubility of its components as a result of cosolvent injection are calculated. The objective of this work was to simulate subsurface contamination by nonaqueous phase liquids and remediation utilizing cosolvent injection technique, and also contamination by gasoline with alcohol from fuel tanks. It is also possible to solve two-phase flow problems involving a gaseous phase. A model for mass transfer during two-phase flow was developed and implemented. Properties variation with time for both phases is calculated using ternary phase diagram. Finite element method was used to solve flow and solute transport equations.

The objective of this research is to evaluate a coastal prediction system under various real world scenarios to test the efficacy of the system as a management tool in Tampa Bay. The prediction system, comprised of a three-dimensional numerical circulation model and a Lagrangian based particle tracking model, simulates oceanographic scenarios in the bay for past (hindcast), present (nowcast) and future (forecast) time frames. Instantaneous velocity output from the numerical circulation model drives the movement of particles, each representing a fraction of the total material, within the model grid cells. This work introduces a probability calculation that allows for rapid analysis of bay-wide particle transport. At every internal time step a ratio between the number of particles in each individual model grid cell to the total number of particles in the entire model domain is calculated. These ratios, herein called transport quotients, are used to construct probability maps showing locations in Tampa Bay most likely to be impacted by the contaminant. The coastal prediction system is first evaluated using dimensionless particles during an anhydrous ammonia spill. In subsequent studies biological and chemical characteristics are incorporated into the transport quotient calculations when constructing probability maps. A salinity tolerance is placed on particles representing Karenia brevis during hindcast simulations of a harmful algal bloom in the bay. Photobleaching rates are incorporated into probability maps constructed from hindcast simulations of seasonal colored dissolved organic matter (CDOM) transport. The coastal prediction system is made more robust with the inclusion of biological parameters overlaid on top of the circulation dynamics. The system successfully describes the basic physical mechanisms underlying the transport of contaminants in the bay under various real world scenarios. The calculation of transport quotients during the simulations in order to develop probability maps is a novel concept when simulating particle transport but one which can be used in real-time to support the management decisions of environmental agencies in the bay area.

Reducing agricultural nutrient loading in Iowa is critical to achieving Gulf of Mexico hypoxia water quality goals. Iowa comprises 4.4% of the Mississippi-Atchafalaya River Basin but contributes an average of 29% of the annual nitrate (NO3-N) load to the Gulf of Mexico (Jones et al., 2018). The main goal of this research was to study nitrogen fate and transport in agricultural areas of Iowa at different spatial scales using a unique combination of water monitoring and numerical modeling. High-frequency, continuous water quality monitoring provided valuable insights into stream and wetland NO3-N dynamics. A biogeochemical model was written and coupled to a spatially distributed, surface-subsurface hydrologic model to perform continuous (multi-year) nitrogen fate and transport simulations at the field, wetland, and watershed scales. Field scale simulations of a tile-drained, corn-soybean rotation under conventional agricultural management over a 5-yr period illustrated strengths and weaknesses of the soil nitrogen model. Using a simplified approach to describe soil organic matter dynamics, the simulated annual nitrogen balance and NO3-N loss in tile drainage were comparable to observations and literature estimates. However, the model was not able to predict the correct response of NO3-N loss in tile drainage to fertilizer rate, which was attributed in part to limitations with the current plant uptake function which did not capture the nonlinear relationship expected between fertilizer rate and crop nitrogen uptake. NO3-N removal was quantified at one of Iowa’s largest constructed wetlands using high-frequency (15-min), continuous water quality monitoring and hydrologic modeling. The wetland reduced incoming NO3-N concentrations 49% and loads by an estimated 61 kg day-1 from May-Nov over a 3-yr period. Wetland removal was influenced by both hydrologic and biological conditions; mass removal was greatest in Jun when discharge and NO3-N loading were highest, while percent removal was greatest in Aug when discharge was low, water residence times in the wetland were high, and warm water temperatures enhanced processing. The high-frequency monitoring captured NO3-N dynamics not possible with traditional lower frequency grab sampling, including concentration dynamics connected to storm events telling of sources and pathways of NO3-N delivery, diurnal variations in concentration indicative of biological processes, and the marked variability in wetland removal performance during low and high flow conditions. Over 5600 wetlands of similar removal performance treating over 60% of Iowa’s area and costing $1.5 billion would be required to reduce the state’s baseline NO3-N load by 45%. The high-frequency monitoring guided and informed numerical simulations of nitrogen fate and transport at the wetland and watershed scales. Wetland simulations using imposed discharge and water quality conditions upstream of the wetland (inlet) and first order, temperature dependent kinetics produced satisfactory daily and monthly predictions of NO3-N concentration and water temperature downstream of the wetland (outlet) from May-Nov in 3/4 and 4/4 study years, respectively. NO3-N predictions were most sensitive to the denitrification first order rate constant and temperature during low discharge periods and least sensitive to both during storm events. Temperature dependent kinetics were necessary to accurately predict wetland NO3-N removal in late summer. The continuous watershed simulations produced satisfactory monthly predictions of inlet and outlet NO3-N concentration and outlet water temperature. Consistent with findings from other modeling studies, annual nitrogen components and NO3-N dynamics were simulated reasonably well under average hydrologic conditions, while simulated NO3-N dynamics weakened under extreme (wet) hydrologic conditions. Temperature was important for predicting the seasonality of wetland NO3-N removal during the growing season, while other factors such as organic carbon and dissolved oxygen may be more influential outside the growing season when removal can still occur despite cold conditions. A preliminary evaluation of six recently constructed wetlands that detain and process agricultural runoff from 12% of a 45 km2 watershed in north central Iowa estimated sizable flood and NO3-N reductions locally which diminished moving downstream. Continuous watershed simulations over a 13 month period following wetland implementation estimated peak flow reductions of 3-43% at the wetlands that dissipated with drainage area; similarly, the wetlands reduced NO3-N loads by an estimated 7-25% locally and 2% at the watershed outlet. Further refinements to the biogeochemical-hydrologic model are needed to improve simulated NO3-N dynamics in order to more reliably assess downstream flow and NO3-N reduction benefits. This work identified limitations with the current modeling approach, areas of future work, and offers recommendations to guide future conservation design. Sensible hydrologic predictions are imperative to the success and dependability of the water quality simulations, which may seem obvious but can be difficult to ascertain in ungauged catchments. Future work aspires to couple a complete agricultural systems model with a physically-based hydrologic model to simulate the nitrogen cycle in a more comprehensive manner to assess which field scale nitrogen processes are most important to accurately predict stream nutrient loading at the watershed scale. Constructed wetlands could provide greater flood and nutrient reduction benefits if the normal pool hydraulics were designed with smaller hydraulic structures that more effectively throttle down incoming flows and provide the opportunity for active rather than passive pool management. As the ultimate goal of this research and other like work is to quantify progress of water quality goals set forth by the Gulf Hypoxia Task Force and help guide future conservation practice implementation, continued investment in science-based water research, water monitoring, and water modeling is necessary.

Selective laser melting (SLM) is a new and rapidly developing manufacturing method for producing full-density, geometrically complex metal parts. The SLM process is time and cost effective for small-scale production; however, wide-spread adoption of this technique is severely limited by residual stresses that can cause large deformations and in-process build failures. The issues associated with residual stress accumulation are most apparent in parts with overhanging features. Due to the complexity of the SLM process, the accumulation of residual stresses is difficult to assess a priori. The deformations and in-process failures caused by residual stress accumulation often lead to an expensive and time consuming iterative manufacturing process. To aid in the development of general SLM design guidelines for overhanging features, the effect of varying two support structure design parameters on residual stress accumulation were investigated. A part-scale thermo-mechanical finite element model was implemented using Diablo, a multi-physics finite element code developed by Lawrence Livermore National Laboratory (LLNL), and trends observed in the model were validated experimentally. By comparing the distribution and magnitude of residual stresses, it was determined that reducing cooling rate gradients in overhanging features reduces the resulting residual stresses. Additionally, it was shown that volume effective material properties can be used to reduce computational costs in computational models of the SLM process.

Práce se zabývá studiem procesů probíhajících při zhášení silnoproudého oblouku ve zhášecí komoře jističe. Je zaměřena na výpočet dynamiky tekutin a teplotního pole v okolí elektrického oblouku. V práci je dále popsán vliv vzdálenosti plechů v komoře a vliv tvarů plechů z hlediska aerodynamických podmínek uvnitř komory. Dalším cílem dosaženým touto prací je poskytnutí informací o vlivu polohy elektrického oblouku na termodynamické vlastnosti uvnitř komory. Toto je důležité, zejména pokud je oblouk do komory vtahován jinými silami, např. elektromagnetickými a během tohoto vtahovacího procesu mění svůj tvar i polohu. Za účelem co nejjednoduššího, ale zároveň co nejefektivnějšího řešení úkolu, byl vyvinut software určen speciálně pro výpočet dynamiky tekutin numerickou metodou konečných objemů (FVM). Tato metoda je, v porovnání s rozšířenější metodou konečných prvků (FEM), vhodnější pro výpočet dynamiky tekutin (CFD) zejména proto, že režie na výpočet jedné iterace jsou menší v porovnání s ostatními numerickými metodami. Další výhodou tohoto softwarového řešení je jeho modularita a rozšiřitelnost. Cely koncept softwaru je postaven na tzv. zásuvných modulech. Díky tomuto řešení můžeme využít výpočtové jádro pro další numerické analýzy, např. strukturální, elektromagnetickou apod. Jediná potřeba pro úspěšné používání těchto analýz je napsáni solveru pro konečné prvky (FEM). Jelikož je software koncipován jako multi–thread aplikace, využívá výkon současných vícejádrových procesorů naplno. Tato vlastnost se ještě více projeví, pokud se výpočet přesune z CPU na GPU. Jelikož současné grafické karty vyšších tříd mají několik desítek až stovek výpočetních jader a pracují s mnohem rychlejšími pamětmi, než CPU, je výpočetní výkon několikanásobně vyšší.

The aim of this thesis is to compare the results of water infiltration from field tests, from laboratories and from numerical modeling at two selected locations. The first objective is the search procedure and the description of both sites and its adjacent areas. Further work continues with infiltration field tests using ring infiltrometer, at first theoretically for each method, and then practically with own personal measurements in the field. Further from the collected soil samples the measurements are performed in the laboratory, first in the permeameter, and followed by the grain size distribution test, from which the hydraulic conductivity is derived by using empirical formulas. At last the numerical modeling is used and all the results are compared. In the final phase of thesis the recommendations are given for infiltration at various locations and comparsion of the infiltration methods.

L'application de champ électrique est connue pour avoir la capacité d'améliorer significativement la stabilité des flammes. A ce sujet, un modèle mathématique permettant de modéliser la combustion en présence d'un champ électrique a été développé. Les équations de l'aérothermochimie sont couplées à des équations de bilan pour les densités des espèces chargées, et une équation de Poisson pour le potentiel électrique est résolue. Une situation principale est étudiée pendant la thèse ; elle concerne la stabilisation de flammes de diffusion par application d’un champ électrique continu ou alternatif.Les résultats obtenus montrent que la présence du champ électrique améliore significativement la stabilisation de la flamme. L’ampleur de cette amélioration dépend de l’intensité et de la polarité de la tension appliquée. Si la tension appliquée est alternative, un facteur supplémentaire s’ajoute pour influencer la stabilisation ; il s’agit de la fréquence. Une interprétation des mécanismes permettant la stabilisation est proposée. / The application of electric field is known to have the ability to improve significantly the flame stability. In this regard, a mathematical approach to model combustion in the presence of an electric field was developed. The Navier-Stokes equations along with transport equations for charged species and the electric potential Poisson’s equation are solved. A main situation, that concerns the stabilization of diffusion flames by applying a direct or alternating electric field, is studied. The results show that the presence of the electric field improves the flame stabilization. The magnitude of this improvement depends on the intensity and polarity of the applied voltage. If the applied voltage is alternating, an additional factor, which is the frequency of the electric current, influences also the extent of the flame stabilization improvement. An interpretation of the stabilization mechanisms is proposed.

La subduction est une des principales expressions superficielles de la convection mantellique et représente un ingrédient crucial de la géodynamique globale. Cela affecte différents processus de la Terre comme la génération des méga-tremblements de terre et des volcans explosifs sur la surface ou le recyclage des espèces volatiles dans l’intérieur profond. Malgré son importance, plusieurs aspects de la subduction restent à clarifier.Dans ce travail, nous avons étudié la mécanique et l’énergétique du phénomène en adoptant un modèle numérique 2-D de “subduction libre”, basé sur la méthode des éléments frontière. En interprétant systématiquement nos solutions numériques utilisant la théorie des couches minces visqueuses, nous avons déterminé diverses lois d’échelle décrivant les mécanismes physiques sous-jacents aux différents aspects du phénomène. Deux paramètres adimensionnels se distinguent par leur récurrence dans ces lois d’échelle: i) la résistance (adimensionelle) de l’interface de subduction, qui contrôle la contrainte de cisaillement agissant à l’interface entre les deux plaques et ii) la rigidité de la plaque en subduction, qui décrit la résistance mécanique opposée par cette plaque à la flexion. Ce dernier paramètre est particulièrement important, car il met en évidence l’échelle de longueur qui décrit correctement la déformation en flexion de la plaque en subduction (bending length).En ce qui concerne les aspects énergétiques de la subduction, nous avons également étudié l’effet de la dissipation de l’énergie produite dans les zones de subduction sur la convection du manteau à grande échelle. Nos résultats semblent suggérer que la loi d’échelle classique trouvée dans l’étude de la convection de Rayleigh-Bénard en régime permanent d’une couche de fluide isovisqueux reste généralement valable aussi pour la convection du manteau terrestre.Pour conclure, nous avons mis en place une expérience de convection basée sur le séchage d’une suspension colloidale de nanoparticules de silice. Comme les résultats préliminaires ont montré, grâce à sa rhéologie particulière, ce matériau semble être un candidat prometteur pour la modélisation de la convection mantellique en laboratoire. / Subduction is one of the principal surface expressions of mantle convection and it represents a key ingredient of global geodynamics. It affects Earth processes ranging from the generation of mega-earthquakes and explosive volcanoes at thesurface to the recycling of volatile species back into the deep interior. Yet despite its obvious importance, several aspects of subduction remain to be clarified.In this work we endeavored to shed light on the mechanics and the energetics of the phenomenon adopting of a 2-D numerical model of “free subduction” based on the Boundary-Element Method. Systematically interpreting our numerical solutions in the light of thin viscous-sheet theory, we determined various scaling laws describing the physical mechanisms underlying different aspects of the phenomenon. Two dimensionless parameters stand out for their recurrence in suchscaling laws: i) the (dimensionless) strength of the subduction interface, which controls the shear stress acting at the interface between the two plates and ii) the flexural stiffnes of the subducting plate, which describes the mechanical resistance opposed by such plate to bending. This latter parameter is particularly important as it highlights the length scale that properly describes the bending deformation of the subducting plate (bending length).For what concerns the energetics of subduction, we also investigated the effect of the dissipation of energy occurring at subduction zones on large-scale mantleconvection. Our results seem to suggest that the classical scaling law found in the study of the steady-state Rayleigh-Bénard convection of an isoviscous fluid layer remains generally valid also for Earth’s mantle convection.To conclude, we ran a convection experiment based on the drying of a colloidal suspension of silica nanoparticles. As preliminary results have shown, thanksto its particular rheology, this material seems to be a promising candidate for effective laboratory modeling of mantle convection.

De nombreuses incertitudes liées aux mécanismes d'interaction entre les blocs rocheux et le versant naturel lors des chutes de blocs persistent ; la prévision de tels événements reste de ce fait encore incertaine. Néanmoins, les outils numériques et la puissance de calcul ne cessent d'évoluer. Si, auparavant, les calculs trajectographiques étaient restreints à des géométries simplifiées et à des mouvements balistiques en deux dimensions, il devient désormais possible d'y intégrer des raffinements tels que la forme complexe des blocs, des modèles numériques tridimensionnels de terrain d’une grande définition ou une prise en compte fine des mécanismes dissipatifs au niveau du point d'impact entre le bloc et le versant de propagation.L’objectif principal de la thèse est d’analyser, avec un code numérique discret en trois dimensions, l’influence des paramètres de forme et d’interaction sur la nature du rebond dans un contexte d’ingénierie. Nous présentons tout d’abord une méthodologie d'identification et d'étude de sensibilité des paramètres de contact, élaborée et validée à partir d’expérimentations de laboratoire. Cette méthodologie a été appliquée par la suite à deux expérimentations de chute de blocs menées sur sites réels à moyenne et à grande échelle.L’étude réalisée à moyenne échelle a permis de confronter le modèle numérique à des données obtenues lors d'une campagne expérimentale sur voies ferroviaires commanditée par la SNCF et menée en collaboration avec IRSTEA. Les analyses qui ont été réalisées ont porté sur les vitesses d’impact des blocs avec le ballast et les distances de propagation.L’étude menée à grande échelle s’appuie sur plusieurs séries de lâchés de blocs réalisées sur le site expérimental de la carrière d'Authume dans le cadre d’un Benchmark proposé dans le cadre du Projet National C2ROP. L’objectif principal du Benchmark est de tester et de comparer entre eux des logiciels trajectographiques, des codes de calculs numériques et les pratiques d’ingénierie pour en définir la pertinence et les domaines de validité. Dans le cadre de la thèse, ce travail a été conduit en plusieurs phases (à l’aveugle puis avec des données partielles mesurées lors de la campagne d'essais) et nous présentons l'évolution de ces analyses à l'issue de chacune des phases. L’étude a porté principalement sur les vitesses, les hauteurs et les énergies de passage des blocs en certains points du profil de propagation, ainsi que sur les positions d’arrêt des blocs. Une étude sur l'influence de la forme des blocs sur les distances de propagations est également présentée.Enfin, un Benchmark interne réalisé au sein de l'entreprise IMSRN montre l’importance, sur les analyses, de l'expertise de l'opérateur, et des conséquences de l'utilisation de différents outils trajectographiques (en 2D et en 3D). Ces travaux mettent en lumière les problématiques actuelles auxquelles sont souvent confrontés les bureaux d'études et les ingénieurs en charge des études de risques. / Numerous uncertainties related to the machanical interaction between rock boulders and the natural slope during block falls persist; and the forecast of such events is therefore still uncertain. Nevertheless, digital tools and computing power are constantly evolving. Previously, trajectory calculations were restricted to simplified geometries and two-dimensional ballistic movements, but it is now possible to incorporate refinements such as the complex shape of the blocks, three-dimensional numerical models of terrain of large sizes, as well as a better accounting of the dissipative mechanisms at the point of impact between the block and the run-out slope.The main objective of this work is to analyze, with a discrete elements code in three dimensions, the influence of the shape and interaction parameters on the nature of the rebound in an engineering context. We first present a methodology for identifying and studying the sensitivity of contact parameters, developed and validated from laboratory experiments. This methodology was subsequently applied to two block fall experiments conducted on medium and large real-scale scenarios.The study conducted on a medium scale allowed the numerical model to be compared with data obtained during an experimental rockfall tests campaign commissioned by the SNCF and conducted in collaboration with IRSTEA in a railway. The analyzes that were carried out mainly focused on the impact velocities of the blocks with ballast and propagation distances.The large-scale study is based on a series of block releases performed at the experimental site (Authume quarry, France) as part of a Benchmark proposed inside the National Project C2ROP. The mainly goal of this Benchmark is to access and compare trajectory softwares, numerical computation codes and engineering practices to define their relevance and validity domains. As part of the thesis, this work was conducted in several phases (blind phase and then conducted taking in account partial data measured during the experimental tests) and we present the evolution of these analyzes at the end of each one of these phases. The study focused on the velocities, heights and energies of the blocks at certain points of the propagation profile, as well as on the stopping positions of the blocks. The influence of block shapes on run-out distances is also presented.Lastly, an internal Benchmark performed within the IMSRN company shows the importance, on the analyzes, of the expertise of the operator, and the consequences derived from the application of different trajectography tools (in 2D and in 3D). This work highlights the current issues that are often faced by engineering offices and engineers in charge of risk quantification.