No description available.

The objective is to estimate the surface pressure distributions and corresponding spectra induced by fully developed hairpin vortices inclined at an angle of 45 degree to the wall in turbulent boundary layers. On the assumption that fully developed hairpin vortices are governed by inviscid dynamics, we obtain an exact formulation for the stagnation pressure, in terms of a Green function integral along the vortex lines. We then evaluate the surface static pressure by subtracting the dynamic pressure from the results of this formulation applied to our vortex geometry. On the basis of the attached eddy model, which implies that the form of the wave number spectrum can be deduced from the properties of a single eddy, we develop the expressions needed for the surface pressure spectra in terms of eddy number-density. This approach draws on flow visualization evidence, which indicates that the number of eddies observed in both streamwise and spanwise directions of the flow is inversely proportional to their size. The overall wavenumber spectrum consists of contributions from eddies of all sizes, weighted by the number-density. From a parameter study we investigate the influence of varying the largest and smallest eddy scales for different eddy aspect ratios. We then validate our model against existing descriptions and measurements of turbulent boundary layer pressures. We examine the predicted wave number spectra and compare them with those of existing empirical models, available direct numerical simulation and also with the results of flight-test measurements. The present model clearly predicts a theoretically-expected characteristic of the wavenumber spectrum (the -1 slope), whereas this is absent from the empirical models. This discrepancy apart, good quantitative agreement is achieved, particularly for the spanwise (cross-flow) spectrum. The agreement is best for the novel formulation for off-axis correlations proposed by Smol'yakov and Tkachenko (1991), rather than the traditional form (e.g. Corcos). The comparison with flight-test data is made via predicted and measured spatial correlations, and again shows good agreement. The present hairpin vortex model shows that it is capable of predicting the properties of the wall pressure field, and is therefore a promising candidate for use in exploring features less readily obtainable by other methods, e.g. off-axis correlations.

No description available.

No description available.

No description available.

No description available.

Moving beds are devices where a slow flow of particulate solids moving under the influence of gravity is contacted with a flow of gas. Heat transfer rates, mass transfer rates, and consequently reaction rates, for these devices are often modelled using information from well-packed fixed beds. It has been found, however, that moving beds behave differently from well-packed fixed beds (Paterson <I>et al</I>. 1991). The cause of this different behaviour is a bed structure, which is a consequence of the movement of the particles, that yields gas flow maldistribution (Crawshaw <I>et al</I>., 1993). Gas flow maldistribution in moving beds was extensively studied by Moppett (1996) through determinations of gas residence time distributions (RTDs) and radial profiles of axial interstitial gas velocity. These results were obtained in beds of uniform sized particles. The present work extends the study of moving beds to cases where the bed is formed with mixtures of particles of different sizes. The observation of particle segregation by means of particle distribution determinations was carried out with the implementation of two novel techniques, the particle sampling device and the setting of the bed with gelatine techniques. The moving bed structures were further probed using gas flows, specially by the determination of gas RTDs and radial profiles of axial interstitial gas velocity. The significance of the mass fraction of the different sized particles in the mixture forming the beds and the conditions of the particle inlet were both analysed. It was found that the particles of different sizes segregate and the amount of the segregation is strongly dependent on the conditions used for the inlet of the particles into the bed, namely the diameter of the particle inlet tube and the free-fall gap between the top of the column and the top surface of the bed. Particle segregation causes moving bed structures which yield extreme gas flow maldistribution.

Experiments were conducted at low speed in a flat plate rig and a compressor cascade. The flat plate rig was used to optimise flow control by blowing with vortex generator jets. This study showed that a boundary layer separation could be prevented with the application of steady blowing at a jet velocity achievable in a real gas turbine compressor. Pulsed blowing was also studied, yielding a reduction in the injected mass flow rate of 40%, but required an increase in the injected jet velocity. The recommendations of the flat plate study were applied to a cascade in order to establish the performance impact of flow control by blowing with vortex generator jets in an environment more representative of a real compressor. Together with endwall control, flow control was found to remove a separation occurring at high incidence and reduce the profile loss coefficient at midspan by 67%. The skewed jets were, however, found to aggravate a corner separation occurring on the endwall of the cascade to which the jets were skewed, requiring additional endwall suction to be applied to yield a uniformly thin wake across the full blade span. For a blade with a representative aspect ratio, flow control by steady blowing with vortex generator jets was found to reduce the loss coefficient averaged over the full span by 20% relative to the uncontrolled case. In addition to blowing, flow control was also applied in the cascade rig in the form of boundary layer suction. Suction was found to reduce the loss coefficient averaged over the full span by 33% relative to the uncontrolled case. The optimal configurations identified were studied in terms of their impact on the engine cycle, when applied to reduce the solidity of a stator row. Flow control was found to enable a 35% reduction in the number of blades at no cost to compressor efficiency.

In agglomeration using liquid binders, powders are agitated in the presence of liquid to form structured products but the detailed flow fields during processing have not been investigated. Systematic experiments have been performed in a mechanical mixer (diameter 250mm) with ploughshare impellers to examine the effect of processing variables on the velocity fields and torque during agglomeration. Liquid to powder ratio, impeller rotation speed and initial powder volume fill were varied and agglomerate properties were measured. Coincident transitions of the mechanism of agglomeration and type of flow were induced by varying the impeller speed. At around 4Hz the mode of the agglomerate size distribution increased rapidly together with an increase in tap bulk density, a weakening of the agglomerates, and a change in the relationship between torque and time. Tracers were used to show that there was an associated change in the size enlargement mechanism from crushing and layering to coalescence. Detailed velocity information, obtained using a positron camera, showed that at around 4Hz the particle angular velocity increased rapidly while mean axial velocity and the axial dispersion coefficient were greatest. Some particles left the bed as the transition from gravity-dominated to centrifuging flow occurred. For all conditions around 90% of the agglomerates were moving at less than 30% of the impeller tip speed and it is likely that the impeller Froude number (Fr) is not the appropriate criterion to designate the transition from gravity-dominated to centrifuging flow. For instance at 4Hz (Fr=8), around 5% of the agglomerates and none of the sand had a particle Froude number (Fr_p) greater than 1. At 6Hz (Fr=18), 20% of the agglomerates and 5% of the sand had Fr_p>l. Hence the particle Froude number is a better criterion to determine the onset of centrifuging flow. Taken together, the agglomerate properties and velocity distributions were consistent with the current binary collision model for agglomeration. In order to further improve knowledge of agglomeration it is essential to conduct integrated studies of agglomerate properties and flow fields using well-defined systems together with physically-based approaches to modelling.

No description available.