An experimental study was conducted in order to investigate the two-phase-flow phenomena in a mini-size, horizontal impacting tee junction. The test section was machined in an acrylic block with a rectangular cross-section of 1.87-mm height × 20-mm width on the inlet and outlet sides. Air-water mixtures at 200 kPa (abs) and room temperature were used as the test fluids. Four flow regimes were identified visually: bubbly, plug, churn, and annular over the ranges of gas and liquid superficial velocities of 0.04 ≤ JG ≤ 10 m/s and 0.02 ≤ JL ≤ 0.7 m/s, respectively, and a flow regime map was developed. The present flow-regime map was compared with several experimental maps. It is thought from those comparisons that the channel height has a more significant role in determining the flow-regime boundaries than the hydraulic diameter. The two-phase fully-developed pressure gradient was measured in the inlet and the outlet sides of the junction for six different inlet conditions and various mass splits at the junction. Comparisons were conducted between the present data and former correlations. The correlations that agreed best with the present data were identified. Five single-phase test sets were performed. In each set of experiments, the pressure distribution was measured for the whole range of the mass split ratio, Wi/W1. The pressure drop at the junction at each value of Wi/W1 was calculated. Values of the pressure-loss coefficient, , were calculated at various Wi/W1 and inlet Reynolds number. The pressure-loss coefficient was strongly dependent on the inlet Reynolds number in the laminar region, while the results for the turbulent region were almost coincident. Numerical simulations of single-phase flow in an impacting tee junction of identical dimensions to that of the present test-section were performed to confirm the results of the experiments. Phase-redistribution experiments were conducted covering all four inlet flow regimes and models were proposed for predicting the experimental data. Good agreement in terms of magnitude and trend was obtained between the present experimental data and the proposed model. New correlations were developed for the single- and two-phase pressure drop in the junction.

An experimental study was conducted in order to investigate the two-phase-flow phenomena in a mini-size, horizontal impacting tee junction. The test section was machined in an acrylic block with a rectangular cross-section of 1.87-mm height × 20-mm width on the inlet and outlet sides. Air-water mixtures at 200 kPa (abs) and room temperature were used as the test fluids. Four flow regimes were identified visually: bubbly, plug, churn, and annular over the ranges of gas and liquid superficial velocities of 0.04 ≤ JG ≤ 10 m/s and 0.02 ≤ JL ≤ 0.7 m/s, respectively, and a flow regime map was developed. The present flow-regime map was compared with several experimental maps. It is thought from those comparisons that the channel height has a more significant role in determining the flow-regime boundaries than the hydraulic diameter. The two-phase fully-developed pressure gradient was measured in the inlet and the outlet sides of the junction for six different inlet conditions and various mass splits at the junction. Comparisons were conducted between the present data and former correlations. The correlations that agreed best with the present data were identified. Five single-phase test sets were performed. In each set of experiments, the pressure distribution was measured for the whole range of the mass split ratio, Wi/W1. The pressure drop at the junction at each value of Wi/W1 was calculated. Values of the pressure-loss coefficient, , were calculated at various Wi/W1 and inlet Reynolds number. The pressure-loss coefficient was strongly dependent on the inlet Reynolds number in the laminar region, while the results for the turbulent region were almost coincident. Numerical simulations of single-phase flow in an impacting tee junction of identical dimensions to that of the present test-section were performed to confirm the results of the experiments. Phase-redistribution experiments were conducted covering all four inlet flow regimes and models were proposed for predicting the experimental data. Good agreement in terms of magnitude and trend was obtained between the present experimental data and the proposed model. New correlations were developed for the single- and two-phase pressure drop in the junction.

A study on high-viscosity oil-water flow in horizontal pipes has been conducted applying experimental, mechanism analysis and empirical modelling, and CFD simulation approaches. A horizontal 1 inch flow loop was modified by adding a designed sampling section to achieve water holdup measurement. Experiments on high-viscosity oil-water flow were conducted. Apart from the data obtained in the present experiments, raw data from previous experiments conducted in the same research group was collated. From the experimental investigation, it is found that that the relationship between the water holdup of water-lubricated flow and input water volume fraction is closely related to the oil core concentricity and oil fouling on the pipe wall. The water holdup is higher than the input water volume fraction only when the oil core is about concentric. The pressure gradient of water-lubricated flow can be one to two orders of magnitude higher than that of single water flow. This increased frictional loss is closely related to oil fouling on the pipe wall. Mechanism analysis and empirical modelling of oil-water flow were conducted. The ratio of the gravitational force to viscous force was proposed to characterise liquid-liquid flows in horizontal pipes into gravitational force dominant, viscous force dominant and gravitational force and viscous force comparable flow featured with different basic flow regimes. For viscous force dominant flow, an empirical criterion on the formation of stable water-lubricated flow was proposed. Existing empirical and mechanistic models for the prediction of water holdup and/or pressure gradient were evaluated with the experimental data; the applicability of different models is demonstrated. Three-dimensional CFD modelling of oil-water flow was performed using the commercial CFD code Fluent. The phase configurations calculated from the CFD model show a fair agreement with those from experiments and mechanism analysis. The velocity distribution of core annular flow is characterised with nearly constant velocity across the oil core when the oil viscosity is significantly higher than the water viscosity, indicating that the high-viscosity oil core flows inside the water as a solid body. The velocity profile becomes similar to that of single phase flow as the oil viscosity becomes close to the water viscosity.

Escoamentos sobre superfícies côncavas, como os que ocorrem no intradorso de uma pá de turbina, estão sujeitos à instabilidade centrífuga. A esse tipo de configuração atribui-se possibilidade de transição à turbulência devido a formação dos vórtices de Görtler. Estudos são propostos no sentido de identificar possível influência do gradiente de pressão nos mecanismos de desenvolvimento desses vórtices e sua interação com outras perturbações na transição. O processo de investigação dá-se numericamente por meio do desenvolvimento e uso de um código numérico paralelizado e de alta ordem de precisão. Resultados obtidos caracterizam o gradiente de pressão adverso como mais instável se comparado ao caso neutro ou favorável. Variações no gradiente de pressão não se mostram eficientes no processo de controle da instabilidade. Ao gradiente adverso atribui-se antecipação da região de saturação dos vórtices. Ressalta-se ainda a natureza desestabilizadora do gradiente adverso quanto aos mecanismos de amplificação dos modos varicoso e sinuoso associados à instabilidade secundária. / Flows over concave surfaces are subjected to centrifugal instability and may transition to turbulence. Studies are conducted to identify the role of the external pressure gradient on the development of the Görtler vortices and their interaction with other flow disturbances. Numerical simulations are carried out by the development and use of an in-house parallel code with highorder of accuracy. Adverse pressure gradient configurations are observed to be more unstable than the neutral and favourable ones. Pressure gradient variations do not prove to be an efficient way to control the centrifugal instability. The destabilizing behaviour that is observed by the adverse pressure gradient justifies its influence on the anticipation of the saturation of the primary vortices and growth of the sinuous and varicose secondary modes.

No description available.

Boundary layer separation is usually an unwanted phenomenon in most technical applications as for instance on airplane wings, on ground vehicles and in internal flows such as diffusers. If separation occurs it leads to loss of lift, higher drag and results in energy losses. It is therefore important to be able to find methods to control and if possible avoid separation altogether without introducing a too heavy penalty such as increased drag, energy consuming suction etc. In the present work we study one such control method, namely the use of vortex generators (VGs), which are known to be able to hinder turbulent boundary layer separation. We first study the downstream development of streamwise vortices behind pairs and arrays of vortex generators and how the strength of the vortices is coupled to the relative size of the vortex generators in comparison to the boundary layer size. Both the amplitude and the trajectory of the vortices are tracked in the downstream direction. Also the influences of yaw and free stream turbulence on the vortices are investigated. This part of the study is made with hot-wire anemometry where all three velocity components of the vortex structure are measured. The generation of circulation by the VGs scales excellently with the VG blade height and the velocity at the blade edge. The magnitude of circulation was found to be independent of yaw angle. The second part of the study deals with the control effect of vortex generators on three different cases where the strength of the adverse pressure gradient (APG) in a turbulent boundary layer has been varied. In this case the measurements have been made with particle image velocimetry. It was found that the streamwise position where the VGs are placed is not critical for the control effect. For the three different APG cases approximately the same level of circulation was needed to inhibit separation. In contrast to some previous studies we find no evidence of a universal detachment shape factor H12, that is independent of pressure gradient. / QC 20101119

Endwall film cooling is a necessity in modern gas turbines for safe and reliable operation. Performance of endwall film cooling is strongly influenced by the hot gas flow field, among other factors. For example, aerodynamic design determines secondary flow vortices such as passage vortices and corner vortices in the endwall region. Moreover blockage presented by the leading edge of the airfoil subjects the incoming flow to a stagnating pressure gradient leading to roll-up of the approaching boundary layer and horseshoe vortices. In addition, for a number of heavy frame power generation gas turbines that use cannular combustors, the hot and turbulent gases exiting from the combustor are delivered to the first stage vane through transition ducts. Wakes induced by walls separating adjacent transition ducts located upstream of first row vanes also influence the entering main gas flow field. Furthermore, as hot gas enters vane passages, it accelerates around the vane airfoils. This flow acceleration causes significant streamline curvature and impacts lateral spreading endwall coolant films. Thus endwall flow field, especially those in utility gas turbines with cannular combustors, is quite complicated in the presence of vortices, wakes and strong favorable pressure gradient with resulting flow acceleration. These flow features can seriously impact film cooling performance and make difficult the prediction of film cooling in endwall. This study investigates endwall film cooling under the influence of pressure gradient effects due to stagnation region of an axisymmetric airfoil and in mainstream favorable pressure gradient. It also investigates the impact of wake on endwall film cooling near the stagnation region of an airfoil. The investigation consists of experimental testing and numerical simulation. Endwall film cooling effectiveness is investigated near the stagnation region on an airfoil by placing an axisymmetric airfoil downstream of a single row of inclined cylindrical holes. The holes are inclined at 35° with a length-to-diameter ratio of 7.5 and pitch-to-diameter ratio of 3. The ratio of leading edge radius to hole diameter and the ratio of maximum airfoil thickness to hole diameter are 6 and 20 respectively. The distance of the leading edge of the airfoil is varied along the streamwise direction to simulate the different film cooling rows preceding the leading edge of the airfoil. Wake effects are induced by placing a rectangular plate upstream of the injection point where the ratio of plate thickness to hole diameter is 6.4, and its distance is also varied to investigate the impact of strong and mild wake on endwall film cooling effectiveness. Blowing ratio ranged from 0.5 to 1.5. Film cooling effectiveness is also investigated under the presence of mainstream pressure gradient with converging main flow streamlines. The streamwise pressure distribution is attained by placing side inserts into the mainstream. The results are presented for five holes of staggered inclined cylindrical holes. The inclination angle is 30° and the tests were conducted at two Reynolds number, 5000 and 8000. Numerical analysis is employed to aid the understanding of the mainstream and coolant flow interaction. The solution of the computational domain is performed using FLUENT software package from Fluent, Inc. The use of second order schemes were used in this study to provide the highest accuracy available. This study employed the Realizable º-µ model with enhance wall treatment for all its cases. Endwall temperature distribution is measured using Temperature Sensitive Paint (TSP) technique and film cooling effectiveness is calculated from the measurements and compared against numerical predictions. Results show that the characteristics of average film effectiveness near the stagnation region do not change drastically. However, as the blowing ratio is increased jet to jet interaction is enhanced due to higher jet spreading resulting in higher jet coverage. In the presence of wake, mixing of the jet with the mainstream is enhanced particularly for low M. The velocity deficit created by the wake forms a pair of vortices offset from the wake centerline. These vortices lift the jet off the wall promoting the interaction of the jet with the mainstream resulting in a lower effectiveness. The jet interaction with the mainstream causes the jet to lose its cooling capabilities more rapidly which leads to a more sudden decay in film effectiveness. When film is discharged into accelerating main flow with converging streamlines, row-to-row coolant flow rate is not uniform leading to varying blowing ratios and cooling performance. Jet to jet interaction is reduced and jet lift off is observed for rows with high blowing ratio resulting in lower effectiveness.

The velocity fields of high Reynolds number, turbulent, wall boundary layers in non-equilibrium pressure gradients are experimentally investigated. Experiments in two wall configurations were performed; one with a hydrodynamically smooth test wall composed of flat aluminum panels, and the other with a rough surface consisting of 2 mm tall, staggered, circular cylindrical elements. A representative set of pressure gradient distributions were generated on the research wall by a systematically rotated NACA 0012 airfoil placed in a wind tunnel section to determine the functional dependence of the boundary layer formation on pressure gradient. Particle image velocimetry (PIV) was the primary measurement technique used to determine time-resolved features of the velocity flow field. newline{}newline{} It is shown that regardless of wall condition and Reynolds number, the non-equilibrium turbulent boundary layers exhibit increasingly non-local behavior with streamwise development. This is apparent as a lag to the pressure gradient distribution observed in the streamwise developing integrated boundary layer parameters. These ``history effects" are also prevalent in mean velocity profiles which are exhibited as a cross-over of the favorable and adverse pressure gradient profiles in the logarithmic layer. Similar cross-over points are observed in the Reynolds shear and normal stresses, particularly at the streamwise station downstream of the pressure gradient switch. The primary effect of the rough wall is to increase the magnitude of flow scales, and, while they exhibit the same qualitative history effects as the smooth wall, the rough wall flows show an earlier relaxation to equilibrium. Despite inherent uncertainties of indirect skin friction methods for the rough wall, the effective sandgrain roughness parameter k_s does not show a functional dependency to pressure gradient history. An evaluation of the wall-similarity hypothesis solely based on boundary layer thickness to roughness parameter ratios delta/k_s is insufficient and additional parameters such as pressure gradient histories, local roughness Reynolds numbers, and bias uncertainties due to instrument spatial resolution must be considered. / Doctor of Philosophy / In the interface between a surface and a moving fluid is the boundary layer where high shear and viscous stresses cause the bulk velocity to decrease to zero. When turbulent, this region of fluid is characterized by random, chaotic, and fluctuating motions of varying sizes. Parameters such as pressure gradients and geometric irregularities of the surface, referred to as roughness, can increase fluctuating pressures and velocities within the boundary layer and cause unwanted noise, vibration, and increased drag. Although many studies have evaluated boundary layers with either roughness or pressure gradient independently, most surfaces in practical application are susceptible to the compounding influences of both of these parameters. Thus, it is necessary to expand the current knowledge database to include complex flow fields necessary to improve data driven modeling and vehicle design.newline{}newline{} This study focuses on experimental observations of the turbulent velocity field developing in both a rough and smooth wall boundary layer that is induced to a family of bi-directional pressure gradients generated by the pressure field of a rotating airfoil inside in a wind tunnel. Through statistical observations of the velocity field it was found that the varying pressure gradients caused the flow to develop non-local dependencies such that the response of the downstream boundary layer was dependent on the upstream flow history. The principal effect of roughness was to increase the magnitude of turbulent scales, but to show the same qualitative response to the pressure gradient history as seen in a smooth wall flow. However, direct comparison of rough and smooth wall turbulence statistics by means of the ``wall-similarity hypothesis" requires careful consideration of multiple parameters including these flow histories, scales prescribed by roughness parameters, and bias errors from experiment under-resolution of the velocity field.

The pressure fluctuations under a high Reynolds Number, rough-wall, turbulent, boundary layer have been studied in the Virginia Tech Stability Wind Tunnel. Rough surfaces of varying element height (1-mm, 3-mm), shape (hemispheres, cylinders) and spacing (5.5-mm, 10.4-mm, 16.5-mm) were investigated in order to ascertain how the turbulent pressure fluctuations change with changes in roughness geometry. Rough surfaces which contain two types of elements are investigated and relationships between the combination surface and the individual surfaces have been uncovered. Measurements of the wall pressure fluctuations were made using pinhole microphones and hotwire measurements were made to obtain the velocity and turbulence field. Among the principal findings is the development of two scaling laws for the low frequency pressure fluctuations. Both of these are based on the idea that the defect between the edge velocity and some local boundary layer velocity sustains the pressure fluctuations in the outer regions of the flow. The first scaling uses the broadband convection velocity as the local velocity of the large scale pressure fluctuations. The second scaling uses the mean boundary layer velocity. Both these scalings appear more robust than the previously proposed scalings for the low frequency region and are able to scale the pressure spectra of all the data to within 3.5-dB. In addition, it was proven that the high frequency shear friction velocity scaling of Meyers et al. (2015) is universal to rough surfaces of different element shape and density. Physical insights into the shear friction velocity, on which this scaling is based, have been revealed. This includes an empirical formula which estimates the element pressure drag coefficient from the roughness density and the Reynolds number. The slopes in the mid-frequency region were found to vary with element density and microphone location such that a useful scaling could not be determined for this region. The possibility of an overlap region is explored and the expectation of a -1 slope is disproved. It is hypothesised that an evanescent decay of the mid-frequency pressure fluctuations occurs between their actual location and the wall where they are measured. A method for accounting for this decay is presented in order to scale the pressure fluctuations in this region. Lastly, a piecewise interpolation function for the pressure spectrum of rough wall turbulent boundary layers was proposed. This analytical function is based on the low frequency scaling on mean velocity and the high frequency scaling of Meyers et al. (2015) The mid-frequency is estimated by a spline interpolation between these two regions. / Ph. D.

Many flow cases in fluid dynamics face undesirable flow separation due to ad-verse pressure gradients on wall boundaries. This occurs, for example, due togeometrical reasons as in a highly curved turbine-inlet duct or on flow-controlsurfaces such as wing trailing-edge flaps within a certain angle-of-attack range.Here, flow-control devices are often used in order to enhance the flow and delayor even totally eliminate flow separation. Flow control can e.g. be achieved byusing passive or active vortex generators (VGs) for momentum mixing in theboundary layer of such flows. This thesis focusses on such passive and activeVGs and their modelling for computational fluid dynamics investigations. First, a statistical VG model approach for passive vane vortex genera-tors (VVGs), developed at the Royal Institute of Technology Stockholm andthe Swedish Defence Research Agency, was evaluated and further improvedby means of experimental data and three-dimensional fully-resolved computa-tions. This statistical VVG model approach models those statistical vortexstresses that are generated at the VG by the detaching streamwise vortices.This is established by means of the Lamb-Oseen vortex model and the Prandtllifting-line theory for the determination of the vortex strength. Moreover, thisansatz adds the additional vortex stresses to the turbulence of a Reynolds-stresstransport model. Therefore, it removes the need to build fully-resolved three-dimensional geometries of VVGs in a computational fluid dynamics mesh. Usu-ally, the generation of these fully-resolved geometries is rather costly in termsof preprocessing and computations. By applying VVG models, the costs arereduced to that of computations without VVGs. The original and an improvedcalibrated passive VVG model show sensitivity for parameter variations suchas the modelled VVG geometry and the VVG model location on a flat plate inzero- and adverse-pressure-gradient flows, in a diffuser, and on an airfoil withits high-lift system extracted. It could be shown that the passive VG modelqualitatively and partly quantitatively describes correct trends and tendenciesfor these different applications. In a second step, active vortex-generator jets (VGJs) are considered. They were experimentally investigated in a zero-pressure-gradient flat-plate flow atTechnische Universitä̈t Braunschweig, Germany, and have been re-evaluated for our purposes and a parameterization of the generated vortices was conducted. Dependencies of the generated vortices and their characteristics on the VGJsetup parameters could be identified and quantified. These dependencies wereused as a basis for the development of a new statistical VGJ model. This modeluses the ansatz of the passive VVG model in terms of the vortex model, theadditional vortex-stress tensor, and its summation to the Reynolds stress ten-sor. Yet, it does not use the Prandtl lifting-line theory for the determinationof the circulation but an ansatz for the balance of the momentum impact thatthe VGJ has on the mean flow. This model is currently under developmentand first results have been evaluated against experimental and fully-resolvedcomputational results of a flat plate without pressure gradient. / <p>QC 20120511</p>