Search

A new approach to the deconvolution of acoustic sources has been developed and is presented in this work. The goal of this postprocessing is to simplify the beamforming output by suppressing the side lobes and reducing the sources' main lobes to single (or a small number of) points that accurately identify the noise sources' positions and their actual levels. In this work, a new modeling technique for the beamforming output is proposed. The idea is to use an image-processing-like approach to identify the noise sources from the beamforming maps: that is, recognizing lobe patterns in the beamformed output and relating them to the noise sources that would produce that map. For incoherent sources, the beamforming output is modeled as a superposition of point-spread functions and a linear system is posted. For coherent sources, the beamforming output is modeled as a superposition of complex point-spread functions and a nonlinear system of equations in terms of the sources' amplitudes is posted. As a first approach to solving this difficult problem, the system is solved using a new two-step procedure. In the first step, an approximated linear problem is solved. In the second step, an optimization is performed over the nonzero values obtained in the previous step. The solution to this system of equations renders the sources' positions and amplitudes. In the case of coherent sources, their relative phase can also be recovered. The technique is referred as noise source localization and optimization of phased-array results. A detailed analytical formulation, numerical simulations, and sample experimental results are shown for the proposed postprocessing. DOI: 10.2514/1.38073

Experiments were conducted on a high-fidelity, model-scale, main landing gear in the Virginia Polytechnic Institute and State University Stability Wind Tunnel. This wind tunnel was recently upgraded to an anechoic facility, which allowed aeroacoustic measurements to be carried out in the far field and in an environment with significantly less reflections. The model was a very faithful replica of the full-scale landing gear, designed to address the issues associated with low-fidelity models. A 63-element microphone phased array was used to locate the noise-source components of the landing gear from different streamwise positions, both in the near and far fields. The same landing-gear model was previously tested in the original hard-walled configuration of the tunnel with the same phased array mounted on the wall of the test section (i.e., near-field position). The new anechoic configuration of the Virginia Polytechnic Institute and State University wind tunnel offered a unique opportunity to directly compare data collected in hard-walled and semi-anechoic test sections, using the same landing-gear model and phased-array instrumentation. Through these tests, some of the limitations associated with testing in hard-walled wind tunnels were addressed.

Linear cascade testing was performed to simulate the flow conditions experienced by stator blades in an axial compressor with supersonic relative Mach numbers at the inlet to the downstream embedded rotors. Experiments were conducted in a transonic blowdown wind tunnel with a nominal inlet Mach number of 0.65. A single moving normal shock introduced at the exit of the stator cascade simulated the bow shock from a downstream rotor. The shock was generated using a shock tube external to the wind tunnel. Pressure measurements indicated that the stator matched its design intent loading, turning, and loss under steady flow conditions. Effects of the passing shock on the stator flowfield were investigated using shadowgraph photography and digital particle image velocimetry (DPIV). Measurements were taken with three different shock strengths. In each case, the passing shock induced a vortex around the trailing edge of the stator. The size and strength of these vortices were directly related to the shock strength. A suction side separation on the trailing edge of the stator was observed and found to correlate with the vortex blockage. [DOI: 10.1115/1.2370745]

Three sets of low solidity steam turbine nozzle cascades were designed and tested. The objective was to reduce cost through a reduction in parts count while maintaining or improving performance. The primary application is for steam turbine high pressure sections where Mach numbers are subsonic and high levels of unguided turning can be tolerated. The base line design A has a ratio of pitch to axial chord of 1.2. This is the pitch diameter section of a 50% reaction stage that has been verified by multistage testing on steam to have a high level of efficiency. Designs B and C have ratios of pitch to axial chord of 1.5 and 1.8, respectively. All three designs satisfy the same inlet and exit vector diagrams. Analytical surface Mach number distributions and boundary layer transition predictions are presented. Extensive cascade test measurements were carried out for a broad incidence range from -60 to +35 deg. At each incidence, four outlet Mach numbers were tested, ranging from 0.2 to 0.8, with the corresponding Reynolds number variation from 1.8 x [10.sup.5] to 9.0 x [10.sup.5]. Experimental results of loss coefficient and blade surface Mach number are presented and compared for the three cascades. The experimental results have demonstrated low losses over the tested Mach number range for a wide range of incidence from -45 to 15 deg. Designs B and C have lower profile losses than design A. The associated flow physics is interpreted using the results of wake profile, blade surface Mach number distribution, and blade surface oil flow visualization, with the emphasis placed on the loss mechanisms for different flow conditions and the loss reduction mechanism with lower solidity. The effect of the higher profile loading of the lower solidity designs on increased end wall losses induced by increased secondary flow, especially on low aspect ratio designs, is the subject of ongoing studies. [DOI: 10.1115/1.2372774]

An experimental and numerical study was performed on an optimized compressor stator cascade designed to operate efficiently at high inlet Mach numbers ([M.sub.1]) ranging from 0.83 to 0.93 (higher supercritical flow conditions). Linear cascade tests confirmed that low losses and high turning were achieved at normal supercritical flow conditions (0.7 < [M.sub.1] < 0.8), as well as higher supercritical flow conditions (0.83 < [M.sub.1] < 0.93), both at design and off-design incidences. The performance of this optimized stator cascade is better than those reported in the literature based on Double Circular Arc (DCA) and Controlled Diffusion Airfoil (CDA) blades, where losses increase rapidly for [M.sub.1] > 0.83. A two-dimensional (2D) Navier-Stokes solver was applied to the cascade to characterize the performance and flow behavior. Good agreement was obtained between the CFD and the experiment. Experimental loss characteristics, blade surface Mach numbers, shadow-graphs, along with CFD flowfield simulations, were presented to elucidate the flow physics. It is found that low losses are due to the well-controlled boundary layer, which is attributed to an optimum flow structure associated with the blade profile. The multishock pattern and the advantageous pressure gradient distribution on the blade are the key reasons of keeping the boundary layer from separating, which in turn accounts for the low losses at the higher supercritical flow conditions.

An experimental and computational study to determine the effects of riblets on the performance of the Supersonic Throughflow Fan (STF) cascade blades was performed. The cascade was tested in the Virginia Tech intermittent wind tunnel facility, where the Mach and Reynolds (based on chord) numbers were 2.36 and 4.8 x [10.sup.6], respectively. The riblet sheets were symmetric v-grooved type and were applied onto the blade surfaces. Three different riblet heights were tested: 0.023, 0.033, and 0.051 mm. Riblet testing was conducted at design incidence as well as at off-design conditions (incidence angles: +5, -10 deg). Loss coefficients were measured and compared with a control test case where an equivalent thickness of smooth material was applied to the blade. Results show that at the design incidence, the riblet sheet with a height of 0.033 mm provides the optimal benefit, with a reduction of 8.5% in loss coefficient compared to the control case. Smaller effects were measured at the off-design conditions. In addition to the experimental study, a numerical investigation of the riblet effect on the STF cascade was conducted at design incidence. A simple method was developed to model riblet effects due to decrease in turbulent viscous drag and the delay of turbulent transition on the blades. Conclusions from numerical study indicate the 2/3 of the total decrease in losses are the result of delaying the transition location. The final 1/3 decrease in loss coefficient comes from the decrease in turbulent viscous losses. [DOI: 10.1115/1.1667883]

An investigation has been performed of the flow in a supersonic through-flow fan blade cascade. The blade shapes are those of the baseline supersonic through-flow fan (STF). Measurements were made at an inlet Mach number of 2.36 over a 15 deg range of incidence. Flowfield wave patterns were recorded using spark shadowgraph photography and steady-state instrumentation was used to measure blade surface pressure distributions and downstream flowfield. From these measurements, the integrated loss coefficients are presented as a function of incidence angle along with analysis indicating the source of losses in the STF cascade. The results are compared with calculations made using a two-dimensional, cell-centered, finite-volume, Navier-Stokes code with upwind options. Good general agreement is found at design conditions, with lesser agreement at off-design conditions. Analysis of the leading edge shock shows that the leading edge radius is a major source of losses in STF blades. Losses from the leading edge bluntness are convected downstream into the blade wake, and are difficult to distinguish from viscous losses. Shock losses are estimated to account for 70% to 80% of the losses in the STF cascade. [DOI: 10.1115/1.1601257]

Aerosolized fluorescent particles of Kiton Red 620 dye in a water/glycol fluid are generated using a Venturi-type atomizer and shown to provide effective flow seeding for fluorescent particle image velocimetry (PIV), which can mitigate the detrimental effects of laser reflections from surfaces. Ninety two percent of particles by number concentration were found to be <1 ÎĽm in diameter, an acceptable size threshold for gas-flow PIV purposes. A PIV application was conducted in a wind tunnel (freestream velocity U ∞ = 27 m sâˆ'1), using the particles for measurement of the boundary layer flow approaching a forward-facing step (approach boundary layer momentum thickness Reynolds number of R e θ = 5930). Particles were generated from solutions with dye molar concentrations of 2.5 Ă— 10 âˆ' 3 and 1.0 Ă— 10 âˆ' 2 mol lâˆ'1, and PIV images were obtained for both elastic Mie scattering and filtered, Stokes-shifted fluorescent light. Raw images indicate that the fluorescence yield of the 1.0 Ă— 10 âˆ' 2 mol lâˆ'1 solution provides PIV images with high contrast, even in the near-surface regions where Mie scattering image contrast is highly affected by surface reflections. Boundary layer profiles are processed in the region of adverse pressure gradient leading up to the forward-facing step, where the fluorescent PIV was found to perform comparably to the most optimized Mie scattering PIV; both approaches obtained data as near to the wall as 30 ÎĽm, or two viscous wall units in our flow of interest. These results indicate that the new seeding method holds promise for near-surface measurement applications with more complicated three-dimensional geometries, where it is impossible to arrange PIV cameras to reject surface-scattered light. [ABSTRACT FROM AUTHOR]

There is generally good agreement between the results of a three-dimensional viscous Navier-Stokes calculation and experimental measurements of endwall secondary flow in a linear supersonic compressor cascade. This indicates that a full three-dimensional calculation is useful in accurately predicting the structure and details of the secondary flowfield. It is also possible to use a three-dimensional viscous computational fluid dynamics calculation to elucidate the flow physics.