The main aim of this research was to understand the influence of vibration of machine components on lubricating films formed in high-pressure contacts. In the current investigation Spacer Layer Imaging Method has been used to monitor the response of elastohydrodynamic films subjected to lateral and vertical vibrations. For both cases the EHL contact was produced by steel or tungsten carbide ball and a transparent disc which was made of glass or sapphire, loaded against each other. The contacting side of the disc was sputtered with a thin chromium layer and a silica spacer layer. White light was shown onto the contact through a specially built microscope. The interferometric fringes formed by the rays reflected by the chromium layer and by the ball's surface are captured by a high speed CCD camera. The images were subsequently analyzed and converted to film thickness maps according to calibration curves. During the tests conducted under lateral vibrations the effect of several parameters including the frequency of lateral motion, Hertzian pressure, temperature variation and the main entrain speed were investigated. Results have shown that lateral oscillations create ripples through the lubricant film only at highest lateral frequency and low entraining speeds. The parameter which influences the formation of the perturbations in the film is the ratio between the main rolling speed and the lateral speed of the contact. The smaller the ratio the larger the transient phenomena can be identified. It has also been found that temperature change has no significant influence upon the film behaviour. The effect of vertical vibrations on the film thickness was also investigated. The analysis of the film thickness has shown that a sudden increase of load had an effect of increasing the contact diameter and at the same time modified the convergence in the inlet, an enhanced film thickness was produced at the inlet periphery of the initial contact zone and travel through the contact at a velocity equal to the average speed of the contacting surfaces.

The torque associated with overcoming the losses on a rotating disc is of particular importance to the designers of gas turbine engines. Not only does this represent a reduction in useful work, but it also gives rise to unwanted heating of metal surfaces and the adjacent fluid. This research presents a numerical study on the effect of rotor-mounted bolts on the moment coefficient and flow structure within a rotor–stator cavity under conditions representative of modern gas turbine engine design. Steady-state, two-dimensional and three-dimensional, computational fluid dynamics simulations are obtained using the FLUENT commercial code with a standard (k–ω) turbulence model. The model is firstly validated against experimental data and then used to study the effects of presence of rotormounted hexagonal bolts in the rotor-stator cavity under investigation using different dimensionless flow parameters. Also investigated were the effects of changing the number and size of rotor-mounted bolts on the flow structure and amount of losses for two test cases; one corresponding a throughflow dominated condition and the other corresponding a rotationally dominated one. The simulation results showed that decreasing the throughflow rate reduces the area of the wake region causing the wakes to become more circumferential in their path around the bolts. Also it was found that increasing the number and diameter of bolts respectively reduces and increases the area of the wake region. For N>18 a separation bubble forms above the bolt which its length increases with increasing the number of bolts. The total moment coefficient of all bolts in the system increases with increasing the number of bolts. However, the rate of this increase reduces by mounting more bolts. While increasing the diameter of the bolts consistently increases the moment and drag coefficients for the rotationally dominated condition, for the throughflow dominated case an increase and a reduction was observed for respectively the moment and drag coefficients.

This study was motivated by a need to better understand the sprays that can develop when oil leaks occur in gas turbine engines. Current gas turbine engines incorporate an extensive network of oil distribution pipes which deliver lubrication oil to bearings and seals at various locations across the engine. Parts of the oil pipe network are situated in hot, high pressure engine cavities where an oil leak, from a fractured pipe or leaking seal, could ignite and lead to an engine fire. Oil leaks in gas turbine engines create liquid injection in cross-airstream situations, a subject which has been widely studied for combustion systems. However, previous studies are almost exclusively based on circular nozzle geometries. For a fractured oil pipe, the geometry through which the oil leaks approximates to a slot shape rather than a circular nozzle. Sprays which develop in cross-airstreams are most sensitive to the parameters of Weber number (Weg eq) and momentum flux ratio (q). A wide range for these parameters are considered to be possible in engine oil leak scenarios because of the variety of crack dimensions possible and range of airflow conditions across the different sections of the engine; from zero to in excess of We g eq = 4000 and q = 300 could be possible in extreme cases. The aim of this study was to generate and then characterise sprays in representative conditions. The main focus was the characterisation of the droplets which formed in the sprays, with the key objective of providing validation data for CFD codes. Droplet characterisation was performed using a phase Doppler particle analysis system. High speed video as well as pulsed laser sheet digital imaging were also used in the study to provide insight into upstream features of the spray field. A 0.5 x 5.38 mm slot shaped nozzle geometry was used in two orientations; perpendicular alignment ↓↓⦶ and parallel alignment ↓↓⦶ . Water was injected into a cross-airstream over a twelve point test matrix with momentum flux ratios (q) values within the range of 4 ≳ q ≳ 32 and Weber number (We g eq) values within the range of 300 ≳ We g eq ≳ 1600. The position of the spray was highly dependent on slot nozzle orientation. The spray was considerably further offset from the nozzle injection wall in parallel alignment ↓↓⦶ , compared to the perpendicular alignment ↓↓⦶. However, the centre-line distribution of Arithmetic Mean Diameter (AMD) was similar for both orientations, albeit offset further from the injection wall for the parallel slot nozzle. The underlying structure of droplet size distribution was consistent with results for sprays from circular nozzles. At low liquid injection pressures the sprays produced by the perpendicular aligned slot ↓↓⦶ exhibited impingement, producing large droplets in the near wall region. Where impingement was not present, the data showed that AMD was not significantly influenced by the orientation of the slot nozzle; with all tests generating results in the range of 16 μm ≳ AMD ≳ 80 μm.

<p>This thesis deals with fluid dynamic mechanisms related to papermaking, specif- ically: the initial dewatering mechanisms during roll-forming and fibre motion in sedimentation and in shear flow. </p><p>Pressure and wire position measurements have been conducted in a model resembling the forming zone and the measured pressure distributions are shown to have more complex patterns than the simple model p / T/R (where T is the wire tension and R is the roll radius). It is shown that an increase in wire tension has a similar effect as a decrease in flow-rate on the shape of the pressure distribution. In addition, it is shown that the drainage has a stabilizing effect on the dewatering pressure. </p><p>The flow around the forming roll has also been modelled with the assump- tion that the wire is impermeable. A non-linear equation for the position of the wire is derived that clearly shows that the Weber number, We, is an im- portant parameter. The equation is linearized around the trivial solution and has a standing wave solution with a specific wavelength that scales with the We-number. </p><p>Motion of non-Brownian fibre settling in a Newtonian fluid at a small but finite Reynolds number has been studied experimentally. Two different regimes of sedimentation were identified. For dilute suspensions, fibres gener- ally fall without flipping and may travel at velocities larger than that of an isolated particle. In the semi-dilute regime we found the settling process to be dominated by large-scale fluctuations. The velocity fluctuations scale with the suspension volume concentrationφ according toφ1/3, which is similar to the findings for settling spheres. </p><p>The influence of shear on fibre orientation in the near wall region was studied in cellulose acetate fibre suspensions. At low concentration and low aspect ratio fibres were observed to orient perpendicular to the streamwise direction (named rollers) in the near wall region whereas the orientation further into the suspension was unchanged. As the concentration and aspect ratio increased the fraction of rollers decreased. </p><p>Finally, an evaluation of a commercial Ultra Velocity Profiler unit in fibre suspensions are presented. The idea was to determine the velocity and characterise the turbulence from ultra sound echoes from particles in the fluid. However, the spatial and/or temporal resolution of the measurements did not permit turbulence characterisation. These limitations might be possible to overcome and some procedures are proposed and evaluated.</p>

This thesis deals with fluid dynamic mechanisms related to papermaking, specif- ically: the initial dewatering mechanisms during roll-forming and fibre motion in sedimentation and in shear flow. Pressure and wire position measurements have been conducted in a model resembling the forming zone and the measured pressure distributions are shown to have more complex patterns than the simple model p / T/R (where T is the wire tension and R is the roll radius). It is shown that an increase in wire tension has a similar effect as a decrease in flow-rate on the shape of the pressure distribution. In addition, it is shown that the drainage has a stabilizing effect on the dewatering pressure. The flow around the forming roll has also been modelled with the assump- tion that the wire is impermeable. A non-linear equation for the position of the wire is derived that clearly shows that the Weber number, We, is an im- portant parameter. The equation is linearized around the trivial solution and has a standing wave solution with a specific wavelength that scales with the We-number. Motion of non-Brownian fibre settling in a Newtonian fluid at a small but finite Reynolds number has been studied experimentally. Two different regimes of sedimentation were identified. For dilute suspensions, fibres gener- ally fall without flipping and may travel at velocities larger than that of an isolated particle. In the semi-dilute regime we found the settling process to be dominated by large-scale fluctuations. The velocity fluctuations scale with the suspension volume concentrationφ according toφ1/3, which is similar to the findings for settling spheres. The influence of shear on fibre orientation in the near wall region was studied in cellulose acetate fibre suspensions. At low concentration and low aspect ratio fibres were observed to orient perpendicular to the streamwise direction (named rollers) in the near wall region whereas the orientation further into the suspension was unchanged. As the concentration and aspect ratio increased the fraction of rollers decreased. Finally, an evaluation of a commercial Ultra Velocity Profiler unit in fibre suspensions are presented. The idea was to determine the velocity and characterise the turbulence from ultra sound echoes from particles in the fluid. However, the spatial and/or temporal resolution of the measurements did not permit turbulence characterisation. These limitations might be possible to overcome and some procedures are proposed and evaluated. / QC 20101021

This work is concerned with the application of finite difference simulation to modelling the pressure response in partially penetrating oil wells. This has relevance to the oil and hydrology industries where pressure behaviour is used to infer the nature of aquifer or reservoir properties, particularly permeability. In the case of partially penetrating wells, the pressure response carries information regarding the magnitude of permeability in the vertical direction, a parameter that can be difficult to measure by other means and one that has a direct influence on both the total volumes of oil that can be recovered and on the rate of recovery. The derivation of the non-linear differential equations that form the basis for multiphase fluid flow in porous media is reviewed and it is shown how they can be converted into a set of finite difference equations. Techniques used to solve these equations are explained, with particular emphasis on the approach followed by the commercial simulation package used in this study. This involves use of Newton's method to linearize the equations followed by application of a pre-conditioned successive minimization technique to solve the resulting linear equations. Finite difference simulation is applied to a hypothetical problem of solving pressure response in a partially penetrating well in an homogenous but anisotropic medium and the results compared with those from analytical solutions. Differences between the results are resolved, demonstrating that the required level of accuracy can be achieved through selective use of sufficiently small grid blocks and time-steps. Residual discrepancies with some of the analytical methods can be traced to differences in the boundary conditions used in their derivation. The simulation method is applied to matching a complex real-life well test with vertical and lateral variation in properties (including fluid saturation). An accurate match can be achieved through judicious adjustment of the problem parameters with the proviso that the vertical permeability needs to be high. This suggests that the recovery mechanism in the oil field concerned can be expected to be highly efficient, something that has recently been confirmed by production results. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 2002.

Le but du travail présenté est la compréhension puis la simulation numérique des mécanismes d'érosion des augets de turbine Pelton par impacts répétés de gouttes d'eau dans le but de prédire la durée de vie des composants. Tout d'abord, les phénomènes de propagation d'ondes dans les milieux fluide et solide sont étudiés. Cela permet de mettre en lumière l'évolution temporelle et la distribution spatiale des pressions de contact, et l'apparition de microjets par éjection supersonique du fluide au contact. Les études expérimentales de l'érosion par gouttes d'eau traduisent un dommage basé sur la fissuration par fatigue. Des simulations numériques en dynamique rapide couplées fluide-structure sont alors effectuées. Le domaine solide est discrétisé par la Méthode des éléments Finis (MEF), et le domaine fluide par la méthode Smoothed Particle Hydrodynamics (SPH), qui est une méthode particulaire (sans maillage) particulièrement adaptée aux grandes distorsions et au suivi des surfaces libres. L'analyse des états de contraintes vient corroborer la nature cyclique de l'endommagement. La simulation d'érosion est alors réalisée à l'aide de critères de fatigue multiaxiaux. Le choix se porte vers un premier critère général de l'American Society of Mechanical Engineers (ASME), utilisant les valeurs principales des différences de contraintes au cours du temps. Le second choix concerne un critère à plan critique : le critère de Dang Van 2. Il traite séparément la contrainte hydrostatique et le cisaillement alterné maximal local. Ces critères permettent de définir les régions érodées du solide au bout d'un nombre d'impact donné, ce qui fait de cette démarche une méthode prédictive. Une étude paramétrique pour différentes tailles de gouttes et vitesses d'impact est ensuite réalisée, puis on évalue l'influence de la présence d'une couche de coating. / The goal of this work is the comprehension and the numerical simulation of erosion caused by repeated droplet impact on Pelton turbine buckets, to predict the lifetime of these components. First, waves propagation phenomenon inside fluid and solid domains are presented, which allows determining the time evolution and spatial distribution of contact pressure, and the birth of lateral microjets by supersonic ejection of the fluid on the contact. Experimental studies of erosion by droplet impact highlight a fatigue cracking-based erosion mechanism. Then, coupled FSI computation are performed. The solid subdomain is discretized by the Finite Element Method (FEM), and the fluid subdomain by the Smoothed Particle Hydrodynamics (SPH), which is a particle method (meshless) effectively recommended for large distortions and free surface tracking. Stress analysis confirms the cyclic nature of the damage mechanism, and erosion simulation is performed using multiaxial fatigue criteria. The first selected criterion is a general one from the American Society of Mechanical Engineers (ASME) using principal values of stress differences over time. The second one is the Dang van 2 criterion, belonging to the family of critical plane criteria. This criterion considers separately the effects due to hydrostatic stress on one hand, and the ones induced by maximum local shear on the other. These two criteria are used to building the equivalent eroded zones of the solid subdomain for a given number of impacts, which allows to qualify this procedure as a predictive predictive. Finally, a parametric study for different droplet sizes and velocites is computed, and the effects of a coating layer are investigated.