Improving the design of flow channels in a liquid-cooled heat sink is critical for boosting the capabilities of electronic components as well as reducing energy usage by the pump. This work explores the use of topology optimization to minimize the pressure difference across a heat sink and consequently, the energy used to supply the liquid. Topology optimization involves solving mathematical equations to obtain the optimal design for a defined cost function, here the total pressure difference between the inlet and outlet. A design variable called the porosity is defined for each cell in the mesh. The porosity features in a sink term in the momentum equation, which 'solidifies' cells by velocity suppression when deemed to be counterproductive to the cost function. The adjoint method of topology optimization, in particular, is a well-established tool for use in flow network problems and includes non-physical parameters such as the adjoint velocity and pressure. The method isn't without its drawbacks, such as the numerical instability of the adjoint equations, and the absence of boundary layers or wall functions at the interface of high and low porosity. The strength of the adjoint method lies in the ease with which it calculates the gradient of the cost function with respect to the porosity. When applied to the geometries in this work, it is observed that the problem is non-convex and results in multiple optimums with similar cost values. Thus the objective becomes seeking solutions with the simplest shape and at the same time having a minimized pressure difference. Interesting techniques are tested, namely an interpolation function, a velocity tolerance, and a volume constraint. The work is accomplished by modifying an existing adjoint optimization solver in the open-source CFD software, OpenFOAM.

A comprehensive survey of available numerical methods and models was performed on the open source computational fluid dynamics solver OpenFOAM version 2.0 for incompressible turbulent bluff body flows. Numerical methods are illuminated using source code for side-by-side comparison. For validation, the accuracy of flow predictions over a sphere in the subcritical regime and delta wing with sharp leading edge is assessed. Solutions show mostly good agreement with experimental data and data obtained from commercial software. A demonstration of the numerical implementation of a dynamic hybrid RANS/LES framework is also presented, including results from test studies.

In this thesis a Large Eddy Simulation (LES) of a hybrid rocket engine burning ethylene (C2H4) in nitrous oxide (N2O) is explored. This is done primarily using a solver and solution scheme provided by the Swedish Defence Research Agency (FOI) and an (at this date) unpublished chemistry model. This sheds light on some transiet behaviour of a prior experiment conducted with a model engine that the simulation was based on. Due to time constraints the simulation did not cover the full test of the engine. The results confirm predictions from the experiment that the propellant was fuel rich. Some insight on how oxidizer swirl propagates throughout the engine was discovered as well.

No description available.

No description available.

Evaluation of the flow field imparted by a marine vehicle reveals the underlying efficiency and performance. However, the relationship between precise design features and their impact on the flow field is not well characterized. The goal of this work is first, to investigate the thermally-stratified near field of a self-propelled marine vehicle to identify the significance of propulsion and hull-form design decisions, and second, to develop a functional mapping between an arbitrary vehicle design and its associated flow field to accelerate the design analysis process. The unsteady Reynolds-Averaged Navier-Stokes equations are solved to compute near-field wake profiles, showing good agreement to experimental data and providing a balance between simulation fidelity and numerical cost, given the database of cases considered. Machine learning through convolutional networks is employed to discover the relationship between vehicle geometries and their associated flow fields with two distinct deep-learning networks. The first network directly maps explicitly-specified geometric design parameters to their corresponding flow fields. The second network considers the vehicle geometries themselves as tensors of geometric volume fractions to implicitly-learn the underlying parameter space. Once trained, both networks effectively generate realistic flow fields, accelerating the design analysis from a process that takes days to one that takes a fraction of a second. The implicit-parameter network successfully learns the underlying parameter space for geometries within the scope of the training data, showing comparable performance to the explicit-parameter network. With additions to the size and variability of the training database, this network has the potential to abstractly generalize the design space for arbitrary geometric inputs, even those beyond the scope of the training data. / Doctor of Philosophy / Evaluation of the flow field of a marine vehicle reveals the underlying performance, however, the exact relationship between design features and their impact on the flow field is not well established. The goal of this work is first, to investigate the flow surrounding a self–propelled marine vehicle to identify the significance of various design decisions, and second, to develop a functional relationship between an arbitrary vehicle design and its flow field, thereby accelerating the design analysis process. Near–field wake profiles are computed through simulation, showing good agreement to experimental data. Machine learning is employed to discover the relationship between vehicle geometries and their associated flow fields with two distinct approaches. The first approach directly maps explicitly–specified geometric design parameters to their corresponding flow fields. The second approach considers the vehicle geometries themselves to implicitly–learn the underlying relationships. Once trained, both approaches generate a realistic flow field corresponding to a user–provided vehicle geometry, accelerating the design analysis from a multi–day process to one that takes a fraction of a second. The implicit–parameter approach successfully learns from the underlying geometric features, showing comparable performance to the explicit–parameter approach. With a larger and more–diverse training database, this network has the potential to abstractly learn the design space relationships for arbitrary marine vehicle geometries, even those beyond the scope of the training database.

A lack of complex analysis for a thermally buoyant jet in a stratified crossflow has motivated the studies presented. A computational approach using the incompressible Navier--Stokes equations (NSE) under the Boussinesq approximation is utilized. Temperature and salinity scalar transport equations are utilized in conjunction with a linear equation of state (EOS) to obtain the density field and thus the buoyancy forcing. Comparing simulation data to experimental data of a point heat source in a stratified environment provides general agreement between the aforementioned computational model and the physics studied. From the literature surveyed, no unified agreement was presented on the selection of turbulence models for the jet--in--crossflow (JICF) problem. For this reason, a comparison is presented for a standard Reynolds--Averaged Navier--Stokes (RANS) and a hybrid Reynolds--Averaged Navier--Stokes/large eddy simulation (HRLES) turbulence model. The mathematical differences are outlined as well as the implications each model has on solving a buoyant jet in stratified crossflow. The RANS model provides a general over prediction of all flow quantities when comparing to the HRLES models. Studies involving the removal of the thermal component inside the jet as well as varying the environmental stratification strength have largely determined that these affects do not alter the near-field in any significant way, at least for a high Reynolds number JICF. The velocity ratio of the jet being the ratio of the jet velocity to the free--stream flow velocity. Deviating from a velocity ratio of one has provided information on the variability of the forcing on the plate the jet exits from, as well as in the integrated energy quantities far downstream of the jet's exit. The departures presented here show that any deviation from the unity value provides an increase in the overall forces seen by the plate. It was also found that the change in the integrated potential and turbulent kinetic energies is proportional to the deviation from a unity velocity ratio. / Master of Science / A lack of complex analysis for a heated jet in a non-uniform crossflow has motivated the studies presented. A computational approach for the fluid dynamics governing equations under specific assumptions is implemented. Additional equations are solved for temperature and salinity in conjunction with a linear equation of state to obtain the density field. Comparing simulations to experimental data of a point heat source in a non-uniform, fluid tank provides general agreement between the aforementioned computational model and the physics studied. Studying the literature yields no unified agreement on the selection of turbulence treatment for the jet-in-crossflow problem. For this reason, a comparison is presented for two various techniques with differing complexity. The mathematical differences as well as the implications each model are outlined, specifically pertaining to a heated jet in a non-uniform crossflow. The simpler model provides a general over prediction when compared to the more complex model. Studies involving the removal of the heat from inside the jet as well as varying the environmental forcing have largely determined that these affects do not alter the flow field near the jet's origin point in any significant way. Changing the jet's velocity has provided information on the variability of the forcing on the plate the jet exits from, as well as in the energy released into the environment far downstream of the jet's exit. The ratios presented show that any deviation from a notional value provides an increase in the overall forces seen by the plate. It was also found that the change in the released energies is proportional to the deviation from the notional jet velocity.

Une étude de l'impact des mascarets sur le transport des sédiments à l'aide de la simulation numérique a été réalisée dans ce travail. En utilisant le logiciel OpenFOAM CFD, nous avons généré 17 simulations numériques de mascaret avec divers nombres de Froude Fr, allant de 0,99 à 1,66. Deux types de mascarets, ondulant et déferlant, ont été couverts dans ces 17 simulations numériques. Pour les particules sédimentaires non cohésives, nous avons utilisé les équations de Maxey et Riley pour déterminer la trajectoire des particules sédimentaires non cohésives sous l’influence d’un mascaret ondulant. En utilisant le schéma Runge-Kutta du quatrième ordre, une méthode tracker résout les équations de Maxey et Riley qui nécessitent l’information des champs de vitesse au temps t. Pour les particules sédimentaires cohésives, nous avons calculé la distribution des particules sédimentaires cohésives en utilisant un modèle de transport de flocs, présenté par Winterwerp (2001). Dans ce modèle, la concentration volumique solide des sédiments et le diamètre des flocs D sont estimés. Les équations de transport de et D sont résolues en utilisant la méthode des moments présentée par Beaudoin et al. (2002 et 2004). La méthode des moments permet de réduire le temps CPU rendant possible une étude paramétrique. De ce travail, nous avons trouvé une classification du mascaret en fonction du nombre de Froude Fr. Cette classification est également basée sur l’étude menée par Furgerot (2014). Pour un nombre de Froude 1,04 < Fr < 1,43, le mascaret est ondulant. Pour un nombre de Froude 1,43 < Fr < 1,57, le mascaret est partiellement déferlant, similaire à la transition de mascaret définie par Furgerot (2014). Pour un nombre de Froude Fr > 1,57, le mascaret est totalement déferlant. Une analyse de la distribution de la pression a été effectuée par Baddour et Song (1990). Nous avons trouvé que les pressions totales et hydrostatiques d’un mascaret ondulant ont de grandes valeurs sous la crête et le creux. Dans le cas d’un mascaret ondulant, les pressions totales ne sont pas égales à les pressions hydrostatiques. Cela provoque la présence de pressions dynamiques. Dans le cas de mascaret déferlant, les pressions totales deviennent égales aux pressions hydrostatiques. La turbulence réduit les pressions dynamiques. L'impact des mascarets sur le transport de particules sédimentaires non cohésives et cohésives a été étudié dans ce travail. Pour les particules sédimentaires non cohésives, nous avons observé que la trajectoire utilisant l’écoulement généré par OpenFOAM est similaire à la trajectoire de type e proposée par Chen et al. (2010). Des modifications du modèle de Chen ont été faites en incluant les effets de la gravité, l’élévation et l’atténuation pour reproduire des trajectoires de particules non cohésives sous un mascaret ondulant. Nous avons obtenu des relations linéaires entre les paramètres du modèle de Chen modifié (β1 , β2 et β3) et le nombre de Froude Fr. C’est parce que le niveau de la turbulence du mascaret ondulant est faible. L’écoulement induit par mascaret ondulant n’est pas complexe. Ce phénomène physique est quasi linéaire. Le paramètre β1, lié à la célérité avant du mascaret ondulant, diminue lorsque le nombre de Froude Fr augmente. Les paramètres β2 et β3, liés respectivement à l’élévation et à l’atténuation du mascaret déferlant, augmentent lorsque le nombre de Froude Fr augmente. Enfin, pour les particules sédimentaires cohésives, nous avons calculé la distribution de la taille des flocs D sous deux types de mascaret, ondulant et déferlant. Nous avons utilisé le diamètre initial de la particule sédimentaire cohésive d = 4 μm. La taille du floc initial D est égale à 10 μm avec la concentration du floc c = 0,5 kg/m3 . Et nous avons limité la taille maximale du floc à 2000 μm. Nous avons observé que la valeur maximal de la taille des flocs Dmax augmente de façon exponentielle par rapport au nombre de Froude Fr. / A study of the impact of tidal bores on sediment transport by using the numerical simulation has been done in this work. Using OpenFOAM CFD software, we have generated 17 numerical simulations of tidal bores with various values of Froude number Fr, ranging from 0.99 to 1.66. Two types of tidal bores, undular and breaking, have been covered in these 17 numerical simulations. We have studied the behavior of two types of sediment particles, non-cohesive and cohesive sediment particles. For the non-cohesive sediment particles, we have resolved the Maxey and Riley equations to study the influence of undular tidal bores on the trajectory of non cohesive sediment particles. Using the fourth order Runge-Kutta scheme, the method tracker can solve the Maxey and Riley equations that requires the information of velocity fields at time t. For the cohesive sediment particles, we have calculated the distribution of cohesive sediment particles using a floc model that allows to estimate the sediment solid volume concentrationand the diameter of flocs D, presented by Winterwerp (2001). The transport equations of and D are solved using the moment method presented by Beaudoin et al. (2002 and 2004). The moment method has been used because it allows to reduce the CPU time, making feasible a parametric study. From this work, we have found a classification of tidal bores as a function of Froude number Fr. This classification is also based on the study conducted by Furgerot (2014). We have obtained that for a Froude number 1.04 < Fr < 1.43, the tidal bore is undular. For 1.43 < Fr < 1.57, the tidal bore is partially breaking that is similar with the tidal bore transtition defined by Furgerot (2014). For Fr > 1.57, the tidal bore is totally breaking. An analysis of pressure distributions has been performed by Baddour and Song (1990). We found that the total and hydrostatic pressures of undular tidal bores have great values under the crest and the trough wave. In the case of undular tidal bores, the total pressures are not equal to the hydrostatic pressures. In the case of breaking tidal bores, the total pressures become equal to the hydrostatic pressures when the tidal bores are totally breaking. The turbulence reduces the dynamic pressures. The impact of tidal bores on the transport of non-cohesive and cohesive sediment particles have been studied in this work. For the non-cohesive sediment particles, we have observed that the trajectory using the flow generated by OpenFOAM is similar with the type e trajectory proposed by Chen et al. (2010). The modifications of Chen’s model have been done by including the effects of gravity, elevation and attenuation to reproduce non-cohesive particle trajectories under an undular tidal bore. We have obtained that the relationship between the Chen’s parameters (β1 , β2 and β3) and the Froude number Fr are linear. This is because the level of turbulence for undular tidal bores is low. The flow induced by an undular tidal bore is not complex. This physical phenomenon is quasi linear. The parameter β1 , related to the front celerity of tidal bores, decreases when the Froude number Fr increases. The parameters, β2 and β3, related to the elevation and attenuation of tidal bores respectively, increase when the Froude number Fr increases. Finally, for the cohesive sediment particles, we have calculated the distribution of floc size D under two types of tidal bore, undular and breaking. We have used the initial diameter of cohesive sediment particles d = 4 μm. The initial floc size D is equal to 10 μm with the consentration of floc c = 0.5 kg/m3. And we have limited the maximum floc size equal to 2000 μm. We have obtained that the maximum value of floc size Dmax increases exponentially with the Froude number Fr.

Les émissions polluantes à l’échappement des véhicules automobiles sont l'une des principales sources de pollution de l'air dans le monde d'aujourd'hui. Par conséquent, la législation a évolué afin de limiter ces émissions. L'un des aspects clés pour répondre consiste à bien maîtriser les échanges gazeux au sein du moteur à combustion interne. Cette amélioration est possible par l'optimisation de répartiteurs d'admission d'air. Dans ces répartiteurs d'admission d'air, la maitrise de l’écoulement de type tumble est une piste de progrès. Des volets sont installés à la sortie du répartiteur afin d'améliorer le rapport de tumble et donc le mélange air-carburant (VTS-Variable Tumble System). Une autre caractéristique de l'écoulement à l'intérieur des répartiteurs est l'effet des écoulements pulsés qui engendrent des fluctuations de pression assez importante. Par conséquent, le but de cette étude consiste à simuler le flux d'air pulsé à l'intérieur des répartiteurs d'admission et à identifier l'effet des pulsations de pression sur les composants actifs tels que les volets. Le travail de simulation dans la présente thèse a été effectué à partir du code open source CFD OpenFOAM. Dans un premier temps, l'effet des pulsations de pression est simulé à l'intérieur d'un tube d'acier et une méthodologie de simulation est développée. Les résultats de la simulation sont validés à partir de résultats expérimentaux obtenus sur un dispositif spécifique, le banc dynamique. Ensuite, des simulations ont été effectuées sur le répartiteur d'admission principal avec des volets. Tout d’abord, les simulations sont effectuées en régime permanent avec cinq positions d'ouverture différentes du clapet. Les forces et les moments agissant sur le volet en régime permanent sont obtenus et analysés. Puis, des simulations en régime transitoire avec des effets de pulsation de pression sont effectuées. Les résultats de la simulation instationnaire sont comparés aux résultats expérimentaux en termes de fluctuations de pression relative. Les effets des pulsations de pression sur les forces aérodynamiques et les moments agissant sur les volets sont analysés et commentés. / The exhaust emissions from automobiles are one of the major sources of air pollution in today’s world. Thence,research and development is the key feature of the modern automotive industries to meet strict emission legislation. One of the key aspects to meet these requirements is to improve the gas exchange process within internal combustion engines. It is possible by the design optimization of the air intake manifolds for internal combustion engines. One of such advancement in air intake manifolds is variable tumble systems (VTS). In VTS system, tumble flaps are installed at the exit of the manifold runner in order to improve tumble ratio and hence air-fuel mixing. Another feature of the flow inside the intake manifolds is pressure pulsation effect. Therefore, the aim of the Ph.D. work is to simulate the pulsating air flow inside the air intake manifolds and to identify the effect of the pressure pulsations on the active components like tumble flaps. The simulation work in the present thesis has been carried out on open source CFD code OpenFOAM. In a first step, the effect of pressure pulsations is simulated inside a steel tube and a simulation methodology is developed. The results of the simulation are validated on a specific experimental device, the dynamic flow bench. Then,simulations have been carried out on the main intake manifold with tumble flaps. Firstly, the simulations are performed with five different opening positions of the tumble flap in a steady state configuration. The forces and moments acting on the flap in steady state are obtained and analyzed. Then, unsteady simulations with pressure pulsation effects are performed. The results of obtained from unsteady simulation are compared with the experimental results in terms of relative pressure fluctuations. The effect of the pressure pulsation on the aerodynamic forces and moments acting on the tumble flaps are analyzed and explained.

Les barrages sur les rivières sont des obstacles à la migration piscicole. Les passes à poissons sont des ouvrages permettant aux espèces piscicoles de migrer, autorisant le rétablissement de la continuité écologique des cours d'eau. Le but de ce travail est de mieux comprendre les phénomènes physiques présents dans les passes à poissons naturelles. Ces passes sont des canaux à forte pente, équipé de rangées de plots en quinconce. Pour valider la pertinence de l'utilisation d'un modèle numérique, l'étude s'est limitée à l'écoulement autour d'un cylindre émergent placé au centre d'un canal. Le travail est décomposé en deux parties, une expérimentale et une numérique. La partie expérimentale est conduite dans un canal transparent de 4m de long, 0,4m de large et 0,4m de hauteur. Le diamètre du cylindre est 4cm et sa hauteur 20cm (toujours émergent). Les cas étudiés sont des débits de 5, 10, 15 et 20 l/s pour une pente nulle. Les nombres de Froude sont supérieurs à 0,5 et les nombres de Reynolds, basés sur le diamètre, sont compris entre 15000 et 50000. Les écoulements ont été filmés et un algorithme de suivi de particules (PTV) a été développé. Des zones de faibles vitesses existent, même pour Fr=1, pouvant assurer des zones de refuge pour le poisson. Les forces de trainée ont été mesurées sur le plot. Les évolutions des coefficients de trainée avec le nombre de Froude et des rapports de forme de l'écoulement autour du plot ont ainsi été évaluées. La partie numérique est réalisée avec OpenFOAM pour 4 cas d'étude (Q=10 et 20 l/s, S=0 et 2%) et 2 modèles de turbulence URANS, le RNG k-epsilon et le k-omega SST. Des modélisations en 2D ont également été faites avec Telemac 2D. Les résultats obtenus ont été comparés aux résultats expérimentaux. La modélisation 2D (shallow water) est exploitable seulement pour des nombres de Froude faibles, d'où la nécessité des modélisations en 3D. Le modèle komega SST semble le mieux adapté pour reproduire les écoulements étudiés. Les vitesses locales et les structures en 3D, non quantifiables expérimentalement, ont ensuite été décrites. Les influences du fond et de la surface libre sur le sillage apparaissent clairement en provoquant des vitesses verticales et des tourbillons à grandes échelles. Enfin, une simulation en LES a été conduite. Les structures tourbillonnaires sont mieux représentées que pour les modèles URANS, mais les temps de calcul sont grands. / The dams on rivers are fishes migration obstacles. The fishways are devices allowing the fishes to migrate, permitting the restauration of the ecological continuity. The aim of this work is to better comprehend the physical phenomena existing in the nature-like fishways. This kind of fishway is a high slope channel with staggered rows of blocks. To validate the relevance of the use of a numerical model, the study is limited to the flow around a single free surface piercing cylinder placed in the center of a flume. The work is in two parts, experimental and numérical. The experimental part is conducted in a transparent flume of 4m length, 0.4m width and 0.4m height. The cylinder diameter is 4cm and its height 20cm (always emerged). The studied cases are flow rates of 5 to 20 l/s for a flat bed. The Froude numbers are over 0.5 et the Reynolds numbers, based on the diameter, are in between 15000 and 50000. The flows were filmed and a particules tracking velocimetry (PTV) algorithm was developped. Slow velocities areas exist, even for Fr=1, allowing shelter zones for a fish. The drag forces were also measured. The drag coefficients evolutions with the Froude number and with the flow aspect ratio were estimated. The numerical part is done with OpenFOAM for 4 cases (Q=10 et 20 l/s, S=0 et 2%) and 2 URANS turbulence models, RNG k-epsilon and k-omega SST. 2D simulations are also carried out with Telemac2D. The results are compared with the experimental ones. The 2D modelisation (shallow water) is workable only for small Froude numbers, which justifies the 3D modelisation. The k-omega SST seems the most relevant to reproduce the studied flows. The local velocities and 3D structures, unquantifiable experimentally, were described. The bed and free surface influences on the wake are clearly shown leading to vertical velocities and big scale vorticies. Finally, a LES simulation was conducted. The vortex structures are better reproducted than the URANS simulations, but the computation times are significant.