This dissertation presents the development of a novel hybrid wave model, comprised of the irrotational, 1-D horizontal Boussinesq and 2-D vertical turbulence-closed Reynolds Averaged Navier-Stokes (RANS) wave models. The two constituents are two-way coupled with the interface placed at a location where turbulence is relatively small. Boundary conditions on the interfacing side of each model is provided by its counterpart model through data exchange, requiring certain transformation due to the difference in physical variables employed in both models. The model is intended for large-scale wave simulation, accurate in both the nonbreaking and breaking zones with relatively coarser grid in the former and finer in latter, and yet efficient. Hybrid model tests against idealized solitary and standing wave motions and wave-overtopping on structure exhibit satisfactory to very good agreement. Compared with pure RANS simulations, the hybrid model saves computational time by a factor proportional to the reduction in the size of the RANS domain. Also, a large-scale tsunami simulation is provided for a numerical setup that is practically unapproachable using RANS alone; not only does the hybrid model offer more rapid simulation of relatively small-scale problems, it provides an opportunity to examine very large total domains with the fine resolution typical of RANS simulations. To allow for implementation on even larger domain with affordable CPU time, the hybrid model is parallelized to run on distributed memory machine. This is done by parallelizing the RANS model while leaving the Boussinesq model serial. One of the processors is responsible for both the sub-RANS and Boussinesq calculations. ICCG(0) for solving the pressure equation is parallelized using the nonoverlappingdecomposition technique, requiring more iterations than the serial one. Standing wave and hypothetical tsunami simulations with 960×66 and 1000×100 grids, and using 8 processors confirm model validity and computational efficiency of 82% and 65%. Finally, the 2-D Boussinesq model is parallelized using domain decomposition technique. The solution to the tridiagonal system arising in the model is calculated as the sum of the homogeneous and particular solutions. Parallel model tests using up to 32 processors exhibit model accuracy and efficiency of 80% for simulation with 500×500–2000×2000 grids.

A computational investigation was performed to assess the effectiveness of Detached-Eddy Simulation (DES) as a tool for predicting icing effects. The AVUS code was employed to compute solutions for an iced wing configuration using DES and steady-state Reynolds Averaged Navier-Stokes (RANS) equation methodologies. The model wing was an extruded GLC305/944-ice shape section with a rectangular planform. Unstructured grids were generated using VGRID/GRIDTOOL. The one-equation Spalart-Allmaras turbulence model was used for all steady state RANS and DES computations. The numerical results were evaluated by comparison with experimental data. RANS solutions significantly under-predicted the lift and drag even after mesh refinement. The time-averaged DES computations showed some improvement in lift and drag coefficients, when compared to experimental data near stall at a 6 deg angle of attack. No significant improvement was observed at lower angles of attack. The DES computations were determined to be valid, since significant changes in the flow field were not observed after both mesh refinement and time step refinement.

L'atomisation d'un jet circulaire d'eau typique des applications agricoles est présentée dans cette étude. Maîtriser la dispersion de l'eau à des fins d'irrigation ou de traitements phytosanitaires implique de réduire la consommation d’eau et la pollution de l'environnement. Un cas d'étude simplifié est construit : une buse ronde dn=1.2 mm et d'une longueur Ln=50dn y est considérée. La vitesse d'injection est fixée à UJ=35 m/s et alignée avec la gravité, plaçant le jet liquide dans un régime d'atomisation turbulent. L'écoulement est statistiquement axisymétrique. L'approche est à la fois expérimentale et numérique.Un modèle multiphasique Eulérien de mélange décrit l'écoulement constitué de deux phases. Plusieurs modèles de turbulence U-RANS sont utilisés: k-ε et RSM. Une attention particulière est alors portée à la modélisation des effets de masse volumique variable issus de la formulation du fluide de mélange. Un solveur numérique spécifique est développé à l'aide du code CFD OpenFOAM. Une série de cas d'étude est construite pour tester l'influence de la modélisation de la turbulence et des fermetures de premier/second-ordre des flux massiques turbulents.Les techniques optiques (LDV et DTV) sont déployées pour recueillir des informations statistiques des phases liquide et gazeuse du spray. La campagne expérimentale est réalisée de x/dn=0 jusqu'à x/dn=800. En ce qui concerne la LDV, des gouttelettes d'huile d'olive (~1 µm) permettent d'analyser la phase gazeuse. Une distinction entre les gouttes de liquide et ces traceurs est obtenue par une configuration spécifique de la source laser et le paramétrage de la détection des bouffées Doppler (Filtre-BP et le SNR). Dans la zone dispersée, les mesures par DTV permettent d'estimer les vitesses et les tailles des gouttes. Une attention particulière est portée à l'estimation de la profondeur de champ (DOF) afin d'obtenir une corrélation taille-vitesse des gouttes moins biaisée.Les résultats numériques et expérimentaux concordent pour le champ de vitesse moyenne. Une forte dépendance au modèle de turbulence est trouvée. Cependant, le modèle RSM ne simule pas le comportement du tenseur de Reynolds. En effet, ni l'anisotropie trouvée expérimentalement (R22/R11≈0.05), ni la vitesse de glissement liquide-gaz ne peuvent être reproduites; même avec une fermeture au 2nd-ordre des flux massiques turbulents. Le fort rapport de masse volumique (eau/air), la directionnalité de l'écoulement et la production d'énergie cinétique turbulente peuvent être à l'origine d'une faible dispersion et d'un faible mélange entre les deux fluides. Ce mécanisme n'est pas encore clarifié du point de vue de la modélisation RSM. / A typical water round-nozzle jet for agricultural applications is presented in this study. The dispersion of a liquid for irrigation or pesticides spraying is a key subject to both reduce water consumption and air pollution. A simplified study case is constructed to tackle both scenarios, where a round dn=1.2 mm nozzle of a length Ln=50dn is considered. The injection velocity is chosen to be UJ=35 m/s, aligned with gravity, placing the liquid jet in a turbulent atomization regime. The flow is considered statistically axisymmetric. Experimental and numerical approaches are considered.An Eulerian mixture multiphase model describes the original two-phase flow. Several U-RANS turbulence models are used: k-ε and RSM; where special attention is taken to the modelling of variable density effects from the mixture formulation. A custom numerical solver is implemented using the OpenFOAM CFD code. A series of study cases are constructed to test the influence of the turbulence modeling and first/second-order closures of the turbulent mass fluxes. LDV and DTV optical techniques are used to gather statistical information from both the liquid and the gas phases of the spray. The experimental campaign is carried out from x/dn=0 to x/dn=800. Concerning the LDV, small (~1 µm) olive-oil tracers are used to capture the gas phase, where a distinction between the liquid droplets and tracers is achieved by a specific set-up of the laser power source and the burst Doppler setting (BP-Filter and SNR). On the dispersed zone, DTV measurements are carried out to measure velocities and sizes of droplets. Special attention to the depth-of-field (DOF) estimation is taken in order to obtain a less biased droplet’s size-velocity correlation.Numerical and experimental results show good agreement on the mean velocity field. A strong dependence on the turbulence model is found. However, the RSM does not capture the same behaviour on the calculated Reynolds stresses. Indeed, neither the experimental anisotropy (R22/R11≈0.05), nor the liquid-gas slip-velocity can be reproduced, even with a second-order closure for the turbulent mass fluxes. The strong density ratio (water/air), flow’s directionality and production of turbulent kinetic energy may be the cause of a weak dispersion and mixing between the two fluids. This mechanism is not yet clarified from a RSM modeling point-of-view.

In this thesis, the flow and heat transfer in dimpled channels have been investigated using the Reynolds-averaged Navier-Stokes (RANS) approach. The primary objective of this investigation is to identify the capabilities of RANS models to reproduce the characteristics of the flow and heat transfer in dimples. The flow in dimpled channels has been chosen as the test case due to their relevance to gas turbine cooling applications, as well as the fairly complex flow features over dimples, which poses a challenge to turbulence modelling. Five turbulence models have been tested in the present work. These include: the Launder and Sharma k-epsilon model, both the Craft et al. (1996) and (2000) cubic k-epsilon models, the Hanjalic and Jakirlic Reynolds stress model (RSM), as well as the Craft (1998) two-component limit (TCL) RSM. The models have been chosen such that all three classes of RANS closure were tested. The tested models have been applied to two dimpled channel configurations with increasing complexity. In the first, the flow over a single dimple in a channel has been considered, while in the second, the case of a staggered array of dimples has been examined. Moreover, across these two configurations, the effect of the dimple depth, the channel height and the Reynolds number have also been investigated. The results show that all models produce a physically viable solution for the problem of the flow in dimpled channels. Nevertheless, the Craft et al. (1996) and (2000) cubic k-ε models, as well as the Craft (1998) TCL RSM, predicted dimple flow structures that deviate from the expected state. In general, the main flow characteristics are reproduced by the RANS models, and the predicted mean velocity profiles are in good agreement with the data. All models report an overall enhancement in heat transfer levels when using dimples in comparison to those of a plane channel.

Cette étude a pour objectif d’étudier différentes approches de modélisation de la turbulence sur un compresseur centrifuge industriel à fort taux de pression afin d’essayer d’élargir notre compréhension des différents phénomènes physiques mis en jeu et leur interaction avec la turbulence. D’abord, la sensibilité au maillage et au modèle turbulence est évalué sur des calculs RANS. Une analyse de simulations LES est ensuite effectuée. En particulier, une étude de l’effet de la turbulence sur l’écoulement et une comparaison aux résultats RANS et expérimentaux est réalisée. Enfin, deux approches hybrides DES sont étudiées afin d’exposer les problèmes rencontrés par ces modèles sur cette configuration. Il en résulte une évaluation des différentes méthodes et de leur applicabilité future dans un contexte industriel. / This study aims to assess the abilities of existing numerical simulation methods to predict the complex physical phenomena occurring in an industrial centrifugal compressor and especialy the effect of turbulence on the different flow features. RANS simulations are first performed using various turbulence model, then LES simulations and finally, two simulations using RANS/LES hybrid models of DES type are carried out. The whole compressor operating range is simulated using RANS, but because of LES and DES high computational costs, attention is focussed on the nominal operating point. Particular care is devoted to determine the impact of grid refinement on the simulation results. To this end, simulations are performed on three grids, respectively composed of over 8, 26 and 165 million cells. Even though the grids used do not fulfill the mesh refinement criteria recommended in the litterature for an accurate wall-resolved LES simulation, the simulation performed on the denser grid provides interesting conclusions on the turbulence generation and its interaction with the mean flow.The hybrid DES approches used involve a shield function to prevent the boundary layers to be computed in LES. However, this function is found to be unsuited to this centrifugal compressor flow. Indeed, the RANS and LES regions are not correctly located and most of the tip leakage flow is resolved using a RANS approach, preventing the development of turbulence.This work allowed us to evalute the various approches and highlight some of the problems and advantages of each for the simulation of this centrifugal compressor.

he primary objective of this study is to validate and/or identify issues for available numerical methods and turbulence models in OpenFOAM 2.0.0. Such a study will provide a guideline for users, will aid acceptance of OpenFOAM as one of the research solvers at institutions and also guide future multidisciplinary research using OpenFOAM. In addition, a problem of aerospace interest such as the flow features and vortex breakdown around a VFE-II model is obtained for SA, SST RANS and SA-DDES models and compared with DLR experiment. The available numerical methods such as time schemes, convection schemes, P-V couplings and turbulence models are tested as available for a fundamental case of a backward facing step for RANS and Hybrid RANSLES prediction of fully turbulent flow at a Reynolds number of 32000 and the OpenFOAM predictions are validated against experimental data by Driver et.al and compared with Fluent predictions.

The ingestion of sand-particles into turbomachinery decreases their longevity and perfor- mance and can even lead to failure. To address these problems, studying particle-laden flows is of high interest in the field. Due to their low computational cost, RANS simulations remain the preferred method in the design phase of engineering solutions. The reduction in computational cost stems from the fact that RANS simulations neglect velocity fluctuations and solve for the time averaged velocity field. This leads to non-physical results since these velocity fluctuations affect the particle trajectories. To improve the accuracy, the simula- tions of the particles are adapted with particle dispersion models that calculate the velocity fluctuations. This work investigates the dispersion of sand particles due to different types of particle dispersion models. A converging pipe with subsequent impact plate is used for this study. Spherical sand-particles ranging in size from 5 to 100 microns in diameter are injected against the main fluid flow into the pipe. The commercial CFD software Ansys Fluent is used and the RANS simulations are run with the k-omega SST turbulence model. Variants of both the Discrete Random Walk (DRW) and Continuous Random Walk (CRW) dispersion models are looked at. Tracer particles stayed close to the center axis inside the pipe and showed the largest dispersion through the models. The dispersion was higher for the CRW compared to the DRW model for all particle sizes. Inertial particles reached the inner walls of the pipe and showed a reduced effect by the dispersion models. The dispersion for larger particles mainly depends on the particle-wall interactions. / Master of Science / The movement of particles by a fluid is found in many natural and industrial processes. A prominent example from the aerospace sector is the ingestion of sand into gas turbine engines. Anybody who has seen videos or in person landings and takeoffs of helicopters and fixed-wing aircraft in dusty environments can see the large amounts of ingestion into the engines. These sand particles damage the engines and lead to a reduced lifetime, higher maintenance requirements and possible failure of the engine. To successfully predict the movement of those particles into and through the engine, simulations are an often used tool. Since the simulation of turbulent flows requires high computational effort, the most common approach is to only solve for the average of the fluid velocities over time in each location which greatly reduces the computational cost. As a result when including particles with the fluid flow, the effects that fluctuating velocities have on the movement of the particles are neglected. Models using random numbers can imitate the missing velocity fluctuations and include these effects in the simulations. There are different types of models with so-called Continuous Random Walk models calculating a continually changing fluctuating velocity as the particle travels through the domain. Discrete Random Walk models on the other hand have a constant fluctuating velocity for some time until it switches to another randomized fluctuating velocity as the particle travels through the domain. This work compares the effects of the different models on the motion of the particles. An increase in dispersion for larger and smaller particles is found for the Continuous Random Walk model.

An interface-preserving level set method is incorporated into the Reynolds- Averaged Navier-Stokes (RANS) numerical method for the time-domain simulation of green water effects. This generalized method can be used to evaluate two- and three-dimensional, laminar and turbulent, free surface flows in moving non-orthogonal grids. In the method, free surface flows are modeled as immiscible two-phase (air and water) flows. A level set function is used to mark the individual fluids and the free surface itself is represented by the zero level set function. The level set evolution equation is coupled with the conservation equations for mass and momentum, and solved in the transformed plane. Chimera domain decomposition technique is employed to handle embedding, overlapping, or matching grids. To demonstrate the feasibility of the method, calculations are performed in several bench mark free surface flows including dam break flows, free jets, solitary wave propagations and the impingement of dam break flow on a fixed structure. The comparisons between the simulations and the experimental data provide a thorough validation of the present method. The results also show the potential capability of level-set RANS method in much more complicated free surface flow simulations. After validations, the method is applied to simulate sloshing flows in LNG tank and green water over the platform. In sloshing flows, the level-set RANS method captures the large impact pressure accurately on LNG tank walls. It also generates a plunging breaker successfully in front of a platform in the numerical wave tank. The good agreements between numerical and experimental results prove the level set RANS method is a powerful and accurate CFD methodology in free surface flow simulations.

An interface-preserving level set method is incorporated into the Reynolds- Averaged Navier-Stokes (RANS) numerical method for the time-domain simulation of green water effects. This generalized method can be used to evaluate two- and three-dimensional, laminar and turbulent, free surface flows in moving non-orthogonal grids. In the method, free surface flows are modeled as immiscible two-phase (air and water) flows. A level set function is used to mark the individual fluids and the free surface itself is represented by the zero level set function. The level set evolution equation is coupled with the conservation equations for mass and momentum, and solved in the transformed plane. Chimera domain decomposition technique is employed to handle embedding, overlapping, or matching grids. To demonstrate the feasibility of the method, calculations are performed in several bench mark free surface flows including dam break flows, free jets, solitary wave propagations and the impingement of dam break flow on a fixed structure. The comparisons between the simulations and the experimental data provide a thorough validation of the present method. The results also show the potential capability of level-set RANS method in much more complicated free surface flow simulations. After validations, the method is applied to simulate sloshing flows in LNG tank and green water over the platform. In sloshing flows, the level-set RANS method captures the large impact pressure accurately on LNG tank walls. It also generates a plunging breaker successfully in front of a platform in the numerical wave tank. The good agreements between numerical and experimental results prove the level set RANS method is a powerful and accurate CFD methodology in free surface flow simulations.

This study focused on the improvement of film cooling for gas turbine vanes using both computational and experimental techniques. The experimental component used a matched Biot number model to measure scaled surface temperature (overall effectiveness) distributions representative of engine conditions for two new configurations. One configuration consisted of a single row of holes on the pressure surface while the other used numerous film cooling holes over the entire vane including a showerhead. Both configurations used internal impingement cooling representative of a 1st vane. Adiabatic effectiveness was also measured. No previous studies had shown the effect of injection on the mean and fluctuating velocity profiles for the suction surface, so measurements were made at two locations immediately upstream of film cooling holes from the fully cooled cooling configuration. Different blowing conditions were evaluated. Computational tools are increasingly important in the design of advanced gas turbine engines and validation of these tools is required prior to integration into the design process. Two film cooling configurations were simulated and compared to past experimental work. Data from matched Biot number experiments was used to validate the overall effectiveness from conjugate simulations in addition to adiabatic effectiveness. A simulation of a single row of cooling holes on the suction side also gave additional insight into the interaction of film cooling jets with the thermal boundary layer. A showerhead configuration was also simulated. The final portion of this study sought to evaluate the performance of six RANS models (standard, realizable, and renormalization group k-ε; standard k-ω; k-ω SST; and Transition SST) with respect to the prediction of thermal boundary layers. The turbulent Prandtl number was varied to test a simple method for improvement of the thermal boundary layer predictions. / text