A new class of cost-optimized prefactored high-order compact schemes is developed for shockfree error-bounded aeroacoustic applications. The cost-optimization theory of Pirozzoli (2007), based on the minimization of the computational cost for a given level of error, is applied to a class of prefactored compact sixth-order schemes. They are extended to obtain a new class of time-explicit cost-optimized schemes. Appropriate high-order prefactored boundary closures are coupled with the new interior schemes. Their effect on the stability and accuracy of the interior schemes and their wave propagation characteristics in Fourier space are investigated. More conventional non-reflecting boundary conditions are shown to display an impedance mis-match, reducing the order of accuracy of the overall scheme. An 11-point stencil with double precision accuracy is used as the prefactored interior boundary stencil. It shows a better performance in spectral sense compared to the equivalent ones available in literature. An eigenvalue analysis is performed, to verify the stability of the prefactored cost-optimized schemes coupled with the boundary closures. Characteristics based boundary conditions and absorbing layers are evaluated. A parallelization strategy, based on a finite-sized overlapping interface, is presented and weak scalability tests results are shown. The theoretical roll-off error of the new schemes agree well with the computed norm error roll-off between the analytical prediction and the numerical experiments, for a monochromatic sinusoidal test-case. There is a good agreement between the predicted percentage cost-saving of the one-dimensional cost function and the savings in computational time from the numerical tests. A 22% computational cost-saving at the design level of error is achieved. Sample applications to broadband and two-dimensional space benchmark problems demonstrate the low error-bounded and high-order accuracy characteristics of the baseline scheme for aeroacoustic applications.

Determining the geometry of an acoustic enclosure using microphone arrays has become an active area of research. Knowledge gained about the acoustic environment, such as the location of reflectors, can be advantageous for applications such as sound source localization, dereverberation and adaptive echo cancellation by assisting in tracking environment changes and helping the initialization of such algorithms. A methodology to blindly infer the geometry of an acoustic enclosure by estimating the location of reflective surfaces based on acoustic measurements using an arbitrary array geometry is developed and analyzed. The starting point of this work considers a geometric constraint, valid both in two and three-dimensions, that converts time-of-arrival and time-difference-pf-arrival information into elliptical constraints about the location of reflectors. Multiple constraints are combined to yield the line or plane parameters of the reflectors by minimizing a specific cost function in the least-squares sense. An iterative constrained least-squares estimator, along with a closed-form estimator, that performs optimally in a noise-free scenario, solve the associated common tangent estimation problem that arises from the geometric constraint. Additionally, a Hough transform based data fusion and estimation technique, that considers acquisitions from multiple source positions, refines the reflector localization even in adverse conditions. An extension to the geometric inference framework, that includes the estimation of the actual speed of sound to improve the accuracy under temperature variations, is presented that also reduces the required prior information needed such that only relative microphone positions in the array are required for the localization of acoustic reflectors. Simulated and real-world experiments demonstrate the feasibility of the proposed method.

This thesis concerns numerical acoustic wave scattering analysis. Such problems have been solved with computational procedures for decades, with the boundary element method being established as a popular choice of approach. However, such problems become more computationally expensive as the wavelength of an incident wave decreases; this is because capturing the oscillatory nature of the incident wave and its scattered field requires increasing numbers of nodal variables. Authors from mathematical and engineering backgrounds have attempted to overcome this problem using a wide variety of procedures. One such approach, and the approach which is further developed in this thesis, is to include the fundamental character of wave propagation in the element formulation. This concept, known as the Partition of Unity Boundary Element Method (PU-BEM), has been shown to significantly reduce the computational burden of wave scattering problems. This thesis furthers this work by considering the different interpolation functions that are used in boundary elements. Initially, shape functions based on trigonomet- ric functions are developed to increase continuity between elements. Following that, non-uniform rational B-splines, ubiquitous in Computer Aided Design (CAD) soft- ware, are used in developing an isogeometric approach to wave scattering analysis of medium-wave problems. The enriched isogeometric approach is named the eXtended Isogeometric Boundary Element Method (XIBEM). In addition to the work above, a novel algorithm for finding a uniform placement of points on a unit sphere is presented. The algorithm allows an arbitrary number of points to be chosen; it also allows a fixed point or a bias towards a fixed point to be used. This algorithm is used for the three-dimensional acoustic analyses in this thesis. The new techniques developed within this thesis significantly reduce the number of degrees of freedom required to solve a problem to a certain accuracy—this reduc- tion is more than 70% in some cases. This reduces the number of equations that have to be solved and reduces the amount of integration required to evaluate these equations.

The estimation of energy dissipated during multiple particle impact is a key aspect in evaluating the abrasive potential of particle-laden streams. A systematic investigation of particle impact energy using acoustic emission (AE) measurements is presented in this thesis with experiments carried out over a range of particle sizes, particle densities and configurations. A model of the AE impact time series is developed and validated on sparse streams where there are few particle overlaps and good control over particle kinetic energies. The approach is shown to be robust and extensible to cases where the individual particle energies cannot be distinguished. For airborne particles, a series of impact tests was carried out over a wide range of particle sizes (from 125 microns to 1500 microns) and incident velocities (from 0.9 ms-1 to 16 ms-1). Two parameters, particle diameter and particle impact speed, both of which affect the energy dissipated into the material, were investigated and correlated with AE energy. The results show that AE increases with the third power of particle diameter, i.e. the mass, and with the second power of the velocity, as would be expected. The diameter exponent was only valid up to particle sizes of around 1.5mm, an observation which was attributed to different energy dissipation mechanisms with the higher associated momentum. The velocity exponent, and the general level of the energy were lower for multiple impacts than for single impacts, and this was attributed to particle interactions in the guide tube and/or near the surface leading to an underestimate of the actual impact velocity in magnitude and direction. In order to develop a model of the stream as the cumulation of individual particle arrival events, the probability distribution of particle impact energy was obtained for a range of particle sizes and impact velocities. Two methods of time series processing were investigated to isolate the individual particles arrivals from the background noise and from particle noise associated with contact of the particles with the target after their first arrival. For the conditions where it was possible to resolve individual impacts, the probability distribution of particle arrival AE energy was determined by the best-fit lognormal probability distribution function. The mean and variance of this function was then calibrated against the known nominal mass and impact speed. A pulse shape function was devised for the target plate by inspection of the records, backed up by pencil lead tests and this, coupled with the energy distribution functions allowed the iv records to be simulated knowing the arrival rate and the nominal mass and velocity of the particles. A comparison of the AE energy between the recorded and simulated records showed that the principle of accumulating individual particle impact signatures could be applied to records even when the individual impacts could not be resolved. For particle-laden liquid, a second series of experiments was carried out to investigate the influence of particle size, free stream velocity, particle impact angle, and nominal particle concentration on the amount of energy dissipated in the target using both a slurry impingement erosion test rig and a flow loop test rig. As with airborne particles, the measured AE energy was found overall to be proportional to the incident kinetic energy of the particles. The high arrival rate involved in a slurry jet or real industrial flows poses challenges in resolving individual particle impact signatures in the AE record, hence, and so the model has been further developed and modified (extended) to account for different particle carrier-fluids and to situations where arrivals cannot necessarily be resolved. In combining the fluid mechanics of particles suspended in liquid and the model, this model of AE energy can be used as a semi-quantitative diagnostic indicator for particle impingement in industrial equipments such as pipe bends.

No description available.

Given a mixture of audio sources, a blind audio source separation (BASS) tool is required to extract audio relating to one specific source whilst attenuating that related to all others. This thesis answers the question “How can the perceptual quality of BASS be improved for broadcasting applications?” The most common source separation scenario, particularly in the field of broadcasting, is single channel, and this is particularly challenging as a limited set of cues are available. Broadcasting also requires that a source separator is automated, capable of handling non-stationary, reverberant mixtures and able to separate an unknown number of sources. In the single-channel case, the time- frequency mask is common as a method of separation. However, this process produces artefacts in the separated audio. The perceptual evaluation for audio source separation (PEASS) toolkit represents an efficient way to generate a multi-dimensional measure of perceptual quality. Initial experimental work, using ideal target and interferer estimates, uses PEASS to test variations on the ideal binary mask and shows continuous masks are perceptually better than binary while identifying a trade-off between artefacts and interferer suppression. To explore the optimisation of this trade-off, a series of sigmoidal functions are used to map target-to-mixture ratios to mask coefficients. This leads to a mask, with less target-to-mixture based discrimination than those typically found in literature, being identified as the optimum. Further experiments applying offsets, hysteresis, smoothing and frequency-dependency to the mask do not show any benefit in audio quality. The optimal sigmoidal mask is demonstrated to also be superior under non-ideal conditions using a non-negative matrix factorisation algorithm to produce the estimates. A final listening test compares the outputs of binary, ratio and optimal sigmoidal masks concluding that listeners prefer the ratio mask to the sigmoidal mask and both continuous masks to the binary mask.

Acoustic levitation techniques have been widely studied within the biological and medical disciplines, primarily to manipulate cells and molecules whose size range lies outside the capabilities of optical tweezers. Since, ultrasonic assembly has been applied more widely, with the trapping of micron- to millimeter-size objects of different shapes and sizes, and the formation of ordered arrays of particles having become possible. The research presented in this thesis investigates the feasibility of applying ultrasonic particle manipulation methodologies to manufacture short fibre reinforced polymer composites. A series of ultrasonic devices is developed allowing the manufacture of thin layers of anisotropic composite material. Strands of unidirectional reinforcement are, in response to the acoustic radiation force, shown to form inside various matrix media. The technique proves suitable for both photo-initiator and temperature controlled polymerisation mechanisms. To further explore key parameters in the design of ultrasonic devices, a number of linear acoustic models are developed. One- and two-dimensional finite element analysis are employed to study the resonance characteristics, compute the acoustic pressure, and calculate the acoustophoretic force on small spherical particles. A range of fibre architectures that can be generated with devices of up to eight transducer elements is explored by plane and spherical wave propagation methods. A separate study analyses the dynamic response of both an elastic sphere and cylinder placed in a standing wave field by solving the equations of non-linear fluid dynamics for arbitrary angles of radiation incidence. A comparison with analytical results shows good agreement in the limit of small particles. For large particles, the acoustic radiation force is further evaluated across a range of pressure amplitudes, and for a number of initial particle positions. Finally, a series of glass fibre reinforced composite samples constructed via the ultrasonic assembly process are subjected to tensile loading and the stress-strain response is characterised. Structural anisotropy is clearly demonstrated, together with a 43 % difference in failure stress between principal directions. The average stiffnesses of samples strained along the direction of fibre reinforcement and transversely across it were 17.66 ± 0.63 MPa and 16.36 ± 0.48 MPa, respectively.

This work deals with including non-uniform mean flow effects into boundary integral solutions to acoustic wave propagation. A time harmonic boundary integral solution is proposed for low Mach number potential flows with small non-uniform mean flow velocities and a free-field Green’s function is recovered to solve the corresponding kernel. The boundary integral formulation can be used as a means of solving both wave extrapolation and boundary element problems. For boundary element solutions to external sound propagation, the non-uniqueness issue is worked around by extending the conventional combined Helmholtz integral equation formulation and the Burton–Miller approach to non-uniform mean flows. Nonetheless, the proposed integral formulation is shown to be consistent with a combination of the physical models associated with the Taylor and Lorentz transforms. The combined Taylor–Lorentz transformation allows mean flow effects on acoustic wave propagation to be resolved by using a standard boundary integral formulation for the Helmholtz problem with quiescent media in a transformed space. Numerical experiments are performed to benchmark the proposed integral formulations against finite element solutions based on the linearised potential equation. Numerical examples are also used to validate a weakly-coupled approach exploiting the proposed integral formulations in order to predict forward fan noise installation effects. Nonetheless, the integral formulations in a transformed space are used to simulate mean flow effects based on standard boundary element solvers for quiescent media. The results suggest that, for low Mach numbers, boundary element solutions to wave propagation with non-uniform mean flows represent a good approximation of finite element solutions based on the linearised potential equation. It is shown that the boundary element solutions including non-uniform mean flow effects improve on the corresponding approximations assuming a uniform flow in the whole computational domain. This is observed when sound propagation is predicted in the near field and in a region where the non-uniformity in the mean flow velocity is significant.

The ability to predict sound fields in coupled volumes is important for noise control and acoustic quality with buildings, cars, aircraft and trains. This thesis investigates methods to assess the diffusivity of sound fields in rooms and the prediction of sound transmission between coupled volumes using statistical approaches. Sound fields in a box-shaped room were assessed using ray tracing with the spatial correlation coefficient for instantaneous sound pressure. The results were compared with the theory for a three-dimensional diffuse field and propagating plane waves. Three different options were considered for the measurement lines: (1) pairs of points formed by one fixed point when the other point varies along the same line, (2) pairs of points with fixed spacing and (3) all permutations of points with variable spacing. The general conclusion is that option (1) can lead to conclusions that seem inappropriate. Options (2) and (3) were found to have potential as assessment procedures, but definitively characterising a sound field as diffuse was not possible. Sound transmission between coupled volumes was investigated using an empty cuboid, a cuboid with staggered barriers and a car cabin model based on Statistical Energy Analysis (SEA) and Experimental SEA (ESEA). Experimental work on corridors was used to validate the ray tracing models. For sound transmission along an empty cuboid, the direct field was significant with highly absorptive surfaces such that a propagating two-dimensional model overestimated transmission for low absorption, and underestimated it for high absorption. SEA incorporating coupling loss factors from the general form of ESEA gave improved agreement with ray tracing and showed the importance of indirect coupling between subsystems. For a corridor with staggered barriers, source locations for the Power Injection Method used in ESEA were assessed to ensure accurate predictions of sound transmission along the corridor. For the corridor and car cabin, the general form of ESEA tends to always result in a working SEA model and be more accurate when a source position (point or surface) used for the power injection process is similar to the actual source position. This tends to be more apparent when using a single source rather than multiple sources.

This thesis addresses two main subjects: the development of efficient electromagnetic-acoustic transducers (EMATs) and the synthesis of coded sequences for pulse-echo mode. EMATs are desirable because no mechanical contact with the sample is required; however, EMATs are inherently inefficient. To improve their performance, a ferromagnetic core surrounded by permanent magnets whose like poles face the core is proposed as the bias magnetic field source; this configuration can outperform a single magnet by an order of magnitude. Coil configurations that result in linear and radial polarisations of the ultrasonic waves are also compared; the linear polarisation was found to yield higher mode purity and penetration depths. Furthermore, the optimal impedance of the EMAT coil is discussed. Although improvements in EMAT sensitivity of up to 20 dB were achieved, this is not enough to obtain adequate signal-to-noise ratios (SNR) ( > 30 dB) without averaging over a long period of time. Therefore, efficient encoding techniques to achieve a greater SNR over shorter periods of time were investigated. The maximum length of conventional coded sequences and hence the maximum SNR increase is limited by the location of the closest reflectors. In this thesis, coded sequences that have receive intervals are introduced so that their overall length and therefore the SNR increase is not affected by the location of the reflectors; the proposed sequences can produce a given SNR increase an order of magnitude faster than averaging. These sequences are then used to drive EMATs using less than 0.5 W and a repetition rate of 10 Hz; this rate can be perceived by a human inspector as quasi-real-time. Moreover, a set of pseudo-orthogonal sequences that have common receive intervals can simultaneously be transmitted through several transducers. This makes it possible to increase the number of active elements in an array, which combined with synthetic focusing, can increase the resolution and contrast of the resulting image without affecting the frame rate. Binary quantisation is also investigated in this thesis. It can be used with low-SNR systems resulting in minimal loss of information while reducing the data throughput and the complexity of the electronics, especially in arrays with many elements. The theory behind binary quantisation is reviewed and the maximum input SNR range that does not cause unacceptable distortions is investigated. Finally, the first low-power, pulse-echo EMAT phased array system is proposed. This array can perform similarly to conventional piezoelectric arrays mounted on a wedge to excite angled shear wave in the sample. Racetrack coils are used as the array elements laid out in an overlapping pattern that minimises inter-element crosstalk and results in an array element width and pitch equivalent to 1 and 2/3 of the wavelength respectively. Coded sequences that have receive intervals are used to reduce the drive power to just a few watts (24 Vpp). The performance of an 8-channel, 1-MHz prototype in the detection of surface cracks which have a length of less than 1 mm is reported.