Particle Image Velocimetry (PlY) is a technique for measuring instantaneous twodimensional fluid velocity fields from a chosen plane of interest within a flow field. This thesis presents new developments and applications of the technique which have been used to study both the non-reacting and reacting flow fields within cylindrical combustion chambers. Non-reacting, swirling laminar flow fields near the transitional flow regime have been investigated by both Computational Fluid Dynamics (CFD) modelling and PIV experiments. Direct comparisons between CFD, PlY and other published results are made and close agreements are found. Additionally, the PlY technique has been optimised by careful use of a thin laser illumination sheet and correct choice of laser pulse separation. This has enabled successful PlY measurements in the boundary layers of the flow field where high velocity gradients exist. The PlY technique has been applied to measure the flame development and propagation process within the chamber under both quiescent and swirling flow conditions. Representative sequences of PIV results at different flame propagation stages are obtained and the accuracy in the extraction of the flame location is discussed. They clearly reveal the instantaneous flame front position and the unburned gas velocity field simultaneously. These features provide further insight into the combustion process itself and also the interaction between the combustion and flow field. A new application of PIV, combined with a flame speed detection technique, has been proposed and developed to obtain direct measurements of the laminar burning velocity of combustible mixtures. The laminar burning velocity is determined as the difference between the flame speed and the unburned gas velocity immediately ahead of the flame front. PIV is used to measure the unburned gas velocity field and either a pair of ionisation probes or a laser beam refraction technique is used to measure the local flame speed simultaneously. The relative merits of each technique are compared. The laminar burning velocities of propane-air mixtures initially at atmospheric conditions for equivalence ratios ranging from 0.7 - 1.4 were measured. The measured values show close agreement with previously published results based on other techniques. The advantages and limitations of the PIV techniques used in this work are examined and the prospects of their improvement and further application are discussed.

A novel technique to study the two-phase flow field around an asymmetric diaphragm in a vertical pipe is presented, that enables producing data for CFD code validation in complex geometries. Main feature is a translocation of the diaphragm to scan the 3D void field with a stationary wire-mesh sensor. Besides the measurement of time-averaged void fraction fields, a novel data evaluation method was developed to extract estimated liquid velocity profiles from the wire-mesh sensor data. The flow around an obstacle of the chosen geometry has many topological similarities with complex flow situations in bends, T-junctions, valves, safety valves and other components of power plant equipment and flow phenomena like curved stream lines, which form significant angles with the gravity vector, flow separation at sharp edges and recirculation zones in their wake are present. In order to assess the quality of the CFD code and their underlying multiphase flow and turbulence models pre-test calculations by ANSYS CFX 10.0 were carried out. A comparison between the calculation results and the experimental data shows a good agreement in term of all significant qualitative details of the void fraction and liquid velocity distributions. Furthermore, the report contains a method to assess the lateral components of bubble velocities in the form of a basic theoretical description and visualisation examples. The plots show the deviation of the flow around the obstacle in term of vectors represented the average velocities of the instantaneous cross-sections of all bubbles in the time interval when they pass the measuring plane. A detailed uncertainty analyse of the velocity assessments concludes the presented report. It includes remarks about the comparison with a second method for calculating bubble velocity profiles - the cross-correlation. In addition, this chapter gives an overview about the influence of acceleration and deceleration effects on the velocity estimation.

Two-phase flow experiments at vertical pipes are much suitable for studying the action of different constitutive relations characterizing the momentum exchange at the gas/liquid interface as well as the dynamic behaviour of the gas/liquid interface itself. The flow can be observed in its movement along the pipe and, in particular, within the shear field close to the pipe wall over a considerable vertical distance and, consequently, over a comparatively long time without the immediate separation of gas and liquid characteristic for horizontal flows. Wire-mesh sensors, which were the working horse in the described experiments, supplied sequences of instantaneous two-dimensional gas fraction distributions with a high-resolution in space and time. This allows to derive from the data not only void fraction and bubble velocity profiles, but also bubble size distributions, bubble-size resolved radial gas fraction profiles as well as the axial evolution of these distributions. An interfacial surface reconstruction algorithm was developed in order to extract the extension of interfacial area from the wire-mesh sensor data. The sensors were upgraded to withstand parameters that are close to nuclear reactor conditions. Most of the experiments were performed for both air/water flow at ambient pressure and steam/water flow of up to 6.5 MPa at identical combinations of the gas and liquid superficial velocities. This offers excellent conditions for studying the influence of the fluid properties.

This study focused on experimentally measuring the performance of a fully cooled, scaled up C3X turbine vane. Experimental measurements focused on investigating row-to-row interactions of coolant jets and the contributions of external film cooling and internal impingement cooling to overall cooling effectiveness. Overall effectiveness was experimentally measured using a thermally scaled, matched Biot number vane model featuring a realistic internal impingement scheme and had normalized surface temperatures that were representative of those found on engine components. A geometrically identical vane was also constructed out of low conductivity polystyrene foam to measure the normalized adiabatic wall temperature, or adiabatic effectiveness of the film cooling configuration. The vanes featured a full coverage film-cooling scheme with a five-row showerhead and 13 total rows of holes containing 149 total coolant holes. This study was the first study to make highly detailed measurements of overall effectiveness on a fully-cooled vane model and expands on previous studies of adiabatic and overall effectiveness on the showerhead and single rows of holes on a matched Biot vane by considering a fully cooled configuration to determine if the results from these previous studies also hold for a fully cooled configuration. Additionally, velocity and thermal fields were measured just upstream of two different suction side rows of holes in order to study the effect of introducing upstream coolant injection. The effects of mainstream turbulence and span-wise location were examined and at the downstream row of holes, the contributions of different rows of holes to the approach flow were compared. This study was the first to measure mean and fluctuating velocity data on the suction side of a turbine vane with upstream coolant injection. Understanding the effects of how upstream injection affects the performance of downstream rows of holes is critical to understanding the film cooling performance on a fully cooled turbine airfoil. / text

A novel technique to study the two-phase flow field around an asymmetric diaphragm in a vertical pipe is presented, that enables producing data for CFD code validation in complex geometries. Main feature is a translocation of the diaphragm to scan the 3D void field with a stationary wire-mesh sensor. Besides the measurement of time-averaged void fraction fields, a novel data evaluation method was developed to extract estimated liquid velocity profiles from the wire-mesh sensor data. The flow around an obstacle of the chosen geometry has many topological similarities with complex flow situations in bends, T-junctions, valves, safety valves and other components of power plant equipment and flow phenomena like curved stream lines, which form significant angles with the gravity vector, flow separation at sharp edges and recirculation zones in their wake are present. In order to assess the quality of the CFD code and their underlying multiphase flow and turbulence models pre-test calculations by ANSYS CFX 10.0 were carried out. A comparison between the calculation results and the experimental data shows a good agreement in term of all significant qualitative details of the void fraction and liquid velocity distributions. Furthermore, the report contains a method to assess the lateral components of bubble velocities in the form of a basic theoretical description and visualisation examples. The plots show the deviation of the flow around the obstacle in term of vectors represented the average velocities of the instantaneous cross-sections of all bubbles in the time interval when they pass the measuring plane. A detailed uncertainty analyse of the velocity assessments concludes the presented report. It includes remarks about the comparison with a second method for calculating bubble velocity profiles - the cross-correlation. In addition, this chapter gives an overview about the influence of acceleration and deceleration effects on the velocity estimation.

Two-phase flow experiments at vertical pipes are much suitable for studying the action of different constitutive relations characterizing the momentum exchange at the gas/liquid interface as well as the dynamic behaviour of the gas/liquid interface itself. The flow can be observed in its movement along the pipe and, in particular, within the shear field close to the pipe wall over a considerable vertical distance and, consequently, over a comparatively long time without the immediate separation of gas and liquid characteristic for horizontal flows. Wire-mesh sensors, which were the working horse in the described experiments, supplied sequences of instantaneous two-dimensional gas fraction distributions with a high-resolution in space and time. This allows to derive from the data not only void fraction and bubble velocity profiles, but also bubble size distributions, bubble-size resolved radial gas fraction profiles as well as the axial evolution of these distributions. An interfacial surface reconstruction algorithm was developed in order to extract the extension of interfacial area from the wire-mesh sensor data. The sensors were upgraded to withstand parameters that are close to nuclear reactor conditions. Most of the experiments were performed for both air/water flow at ambient pressure and steam/water flow of up to 6.5 MPa at identical combinations of the gas and liquid superficial velocities. This offers excellent conditions for studying the influence of the fluid properties.

À ce jour, les différentes méthodes de reconstruction des mouvements du plasma à la surface du Soleil qui ont été proposées présupposent une MHD idéale (Welsch et al., 2007). Cependant, Chae & Sakurai (2008) ont montré l’existence d’une diffusivité magnétique turbulente à la photosphère. Nous introduisons une généralisation de la méthode du Minimum Energy Fit (MEF ; Longcope, 2004) pour les plasmas résistifs. Le Resistive Minimum Energy Fit (MEF-R ; Tremblay & Vincent, 2014) reconstruit les champs de vitesse du plasma et la diffusivité magnétique turbulente qui satisfont à l’équation d’induction magnétique résistive et qui minimisent une fonctionnelle analogue à l’énergie cinétique totale. Une séquence de magnétogrammes et de Dopplergrammes sur les régions actives AR 9077 et AR 12158 ayant chacune produit une éruption de classe X a été utilisée dans MEF-R pour reconstruire les mouvements du plasma à la surface du Soleil. Les séquences temporelles des vitesses et des diffusivités magnétiques turbulentes calculées par MEF-R sont comparées au flux en rayons X mous enregistré par le satellite GOES-15 avant, pendant et après l’éruption. Pour AR 12158, nous observons une corrélation entre les valeurs significatives de la diffusivité magnétique turbulente et de la vitesse microturbulente pour les champs magnétiques faibles. / To this day, the various methods proposed for the reconstruction of plasma motions at the Sun’s surface are all based on ideal MHD (Welsch et al., 2007). However, Chae & Sakurai (2008) have shown the existence of an eddy magnetic diffusivity at the photosphere. We introduce a generalization of the Minimum Energy Fit (MEF; Longcope, 2004) for resistive plasmas. The Resistive Minimum Energy Fit (MEF-R; Tremblay & Vincent, 2014) infers velocity fields and an eddy magnetic diffusivity which solve the resistive magnetic induction equation and minimize an energy-like functional. A sequence of magnetograms and Dopplergrams documenting the active regions AR 9077 and AR 12158 are used as input in MEF-R to reconstruct plasma motions at the Sun’s surface. Time series of the inferred velocities and eddy magnetic diffusivities are compared to the soft X-ray flux observed by GOES-15. We find a positive correlation between significant eddy magnetic diffusivities and microturbulent velocities for weak magnetic fields in AR 12158.

Nearfield Acoustical Holography (NAH) is an acoustic field visualization technique that can be used to reconstruct three-dimensional (3-D) acoustic fields by projecting two-dimensional (2-D) data measured on a hologram surface. However, linear NAH algorithms developed and improved by many researchers can result in significant reconstruction errors when they are applied to reconstruct 3-D acoustic fields that are radiated from a high-level noise source and include significant nonlinear components. Here, planar, nonlinear acoustical holography procedures are developed that can be used to reconstruct 3-D, nonlinear acoustic fields radiated from a high-level noise source based on 2-D acoustic pressure data measured on a hologram surface. The first nonlinear acoustic holography procedure is derived for reconstructing steady-state acoustic pressure fields by applying perturbation and renormalization methods to nonlinear, dissipative, pressure-based Westervelt Wave Equation (WWE). The nonlinear acoustic pressure fields radiated from a high-level pulsating sphere and an infinite-size, vibrating panel are used to validate this procedure. Although the WWE-based algorithm is successfully validated by those two numerical simulations, it still has several limitations: (1) Only the fundamental frequency and its second harmonic nonlinear components can be reconstructed; (2) the application of this algorithm is limited to mono-frequency source cases; (3) the effects of bent wave rays caused by transverse particle velocities are not included; (4) only acoustic pressure fields can be reconstructed. In order to address the limitations of the steady-state, WWE-based procedure, a transient, planar, nonlinear acoustic holography algorithm is developed that can be used to reconstruct 3-D nonlinear acoustic pressure and particle velocity fields. This procedure is based on Kuznetsov Wave Equation (KWE) that is directly solved by using temporal and spatial Fourier Transforms. When compared to the WWE-based procedure, the KWE-based procedure can be applied to multi-frequency source cases where each frequency component can contain both linear and nonlinear components. The effects of nonlinear bent wave rays can be also considered by using this algorithm. The KWE-based procedure is validated by conducting an experiment with a compression driver and four numerical simulations. The numerical and experimental results show that holographically-projected acoustic fields match well with directly-calculated and directly-measured fields.

Les phénomènes éruptifs du Soleil sont souvent accompagnés par l'accélération de particules chargées qui peuvent avoir des impacts significatifs sur la Terre. Toutefois, le mécanisme responsable de ces phénomènes n'est pas suffisamment bien compris pour qu’on puisse en prédire l'occurence. Les satellites et les observatoires terrestres sondent la photosphère, la chromosphère et la couronne du Soleil et sont essentiels pour l'étude de l'activité solaire. Les simulations numériques tentent de faire le pont entre la physique décrivant l'intérieur de l'étoile et de telles observations. La prochaine étape pour des simulations réalistes serait la prévision à court terme des structures à la surface du Soleil. Les travaux présentés dans cette thèse explorent comment des notions empruntées de la météorologie (e.g., l'assimilation des données) et de l'intelligence artificielle (e.g., les réseaux de neurones) pourraient être utilisées pour la prédiction à court terme de l'activité solaire dans le contexte de la météorologie spatiale. En particulier, nous présentons notre implémentation de l'assimilation des données dans un modèle magnétohydrodynamique (MHD) radiatif du Soleil calme (i.e., en l'absence d'activité magnétique) afin de prédire l'évolution de la granulation solaire durant une courte période de temps. Toutefois, ce ne sont pas toutes les variables du modèle qui peuvent être observées ou mesurées à l'aide d'instruments. Par exemple, les mesures directes des mouvements du plasma à la surface du Soleil sont limitées à la composante le long de la ligne de visée. Plusieurs algorithmes ont donc été développés afin de reconstruire la composante transverse à partir de mesures de l'intensité de la lumière ou du champ magnétique. Nous comparons les champs de vitesse inférés par différentes méthodes, dont un réseau de neurones, afin d'identifier la méthode la mieux adaptée pour générer des observations synthétiques dans une chaîne de réduction des données qui pourraient ensuite être introduites dans notre système pour l'assimilation des données. / Eruptive events of the Sun, which often occur in the context of flares, convert large amounts of magnetic energy into emission and particle acceleration that can have significant impacts on Earth's environment. However, the mechanism responsible for such phenomena is not sufficiently well understood to be able to predict their occurrence. Satellites and ground-based observatories probe the Sun's photosphere, chromosphere and corona and are key in studying solar activity. Numerical models have attempted to bridge the gap between the physics of the solar interior and such observations. The next step for realistic simulations would be to forecast the short term evolution of the Sun's photosphere. The following work explores how notions borrowed from meteorology (e.g., data assimilation) and artificial intelligence (e.g., neural networks) could be used to forecast short term solar activity for space-weather modelling purposes. More specifically, we present an implementation of data assimilation in a radiative MHD model of the Quiet Sun (i.e., in the absence of significant magnetic activity) to forecast its evolution over a short period of time. However, not all model variables are directly observable. For example, direct measurements of plasma motions at the photosphere are limited to the line-of-sight component. Multiple algorithms were consequently developed to reconstruct the transverse component from observed continuum images or magnetograms. We compare velocity fields inferred by different methods, including a neural network, to identify the method best suited to generate instantaneous synthetic observations in a data reduction pipeline that would included in our data assimilation framework.