This thesis focuses on the growth of cracks which are small in relation to the material microstructure especially the situation of clusters of small cracks grown from smooth surfaces, termed micro-multi-site cracking, as is frequently the case for components in service. A proper understanding of this regime of crack growth will allow for less conservative maintenance schedules as well as the application of more sensitive health monitoring systems which are currently under development. To address the problem a significant experimental investigation of micro-multi-site cracking was conducted on 7075-T7351 aluminum alloy. Using the resulting data a micro-structurally based transition crack length is defined to determine the point which separates small and long crack growth. This definition is based upon the observed evolution of scatter in the growth rates of growing small cracks. It is shown that this scatter falls with growth until the transition point is reached where it assumes a constant value for the growth of long cracks. It is then shown that the total population of cracks within the clusters can be considered as bi-modal. One distribution consists of primary cracks which can grow and ultimately cause specimen failure. The second distribution consists of secondary cracks, the growth of which ultimately arrests. Several methods for experimentally separating the two distributions have been developed. The first method relies upon the defined transition point between small and long crack behavior. A second method based upon the second derivative of the crack length versus cycle count data has also been developed. Since the secondary cracks cannot lead to failure their data must be discarded prior to any analysis. It is then shown that failure to do so will lead to erroneous non-conservative predictions of crack growth.

Made available in DSpace on 2016-08-29T15:32:42Z (GMT). No. of bitstreams: 1 tese_2324_TeseSalomao.pdf: 1995620 bytes, checksum: a698aa7ccbb9671d5b4ff1eae00a6508 (MD5) Previous issue date: 2007-08-24 / Esta pesquisa propõe uma forma de fusão de informações no nível das características para um sistema bi-modal de reconhecimento pessoal. As características biométricas e comportamentais consideradas na validação deste trabalho são obtidas das seqüências de vídeo da forma humana de caminhar e de imagens faciais. Ela sugere arquiteturas e algoritmos que permitem a implementação de reconhecimento e autenticação de indivíduos através da fusão de características faciais e da forma humana de caminhar, objeto de poucos trabalhos disponíveis na literatura. A escolha da abordagem através da fusão das duas características justifica-se pela suposição de que ela deve oferecer um melhor desempenho de classificação, robustez e segurança ao permitir a identificação noturna e à distância. Além disso, é considerada menos intrusiva do que todos os outros sistemas biométricos, por apresentar pouca ou nenhuma necessidade de colaboração da pessoa a ser identificada. A pesquisa, pela atualidade do tema e pelas poucas referências bibliográficas, inicia-se avaliando o desempenho do sistema de reconhecimento pessoal pela forma de caminhar ao se aplicarem, sobre as seqüências de vídeo, técnicas de extração e seleção de características de silhuetas baseadas em análise de componentes principais (PCA - Principal Component Analysis), análise de componentes independentes (ICA - Independent Component Analysis), transformadas Wavelets e proporção de variâncias (PoV Proportion of Variances), em combinação com os classificadores baseados em distâncias Euclidianas, máquinas de vetores suporte (SVM Support Vector Machines) e redes neurais com funções de bases radiais (RBF Radial Basis Function). Utilizando-se bases de dados de domínio público, foi feito um estudo especial sobre a forma de caminhar objetivando uma avaliação inicial do seu desempenho em sistemas de reconhecimento pessoais, bem como a escolha de métodos e técnicas mais apropriadas para efetuar a fusão desta com as imagens faciais do indivíduo. Em seguida, obtiveram-se bases de dados próprias baseando-se em um cenário proposto que permitiu extrair as silhuetas de seqüências de vídeo da forma de caminhar e das imagens faciais correspondentes a cada seqüência. Avaliou-se o desempenho do sistema de reconhecimento quando se estabelece a fusão de ambas as características biométricas. Em uma etapa posterior, após a definição da arquitetura mais adequada, é proposta e implementada a fusão dos dois sistemas biométricos face-forma de caminhar sobre bases de dados de domínio público e proprietárias. Nesta fase final do trabalho, as implementações sugeridas utilizam-se das técnicas de extração e seleção de características baseadas na energia das silhuetas do caminhar, na proporção de variâncias (PoV) e em algoritmos de classificação baseados em redes neurais com funções de bases radiais (RBF). Os resultados apresentados permitiram avaliar as taxas de acertos, de falsa aceitação e falsa rejeição e o desempenho do sistema, quando considerados os vetores de características individuais das imagens faciais e de seqüências de vídeo do caminhar e da fusão das características de ambas, confirmando a validade e viabilidade de aplicação deste trabalho de fusão no nível das características no sistema bi-modal de reconhecimento pessoal.

The effect of annealing conditions on the microstructure evolution of CVD nickel was investigated systematically in the present study by differential scanning calorimetry, optical microscopy and transmission electron microscopy (TEM), upon both ex-situ and in-situ annealing. TEM observation revealed the as-deposited CVD nickel possessed a bi-modal grain structure, with large columnar grains embedded in nanocrystalline matrix. Ultrafine and nano growth twins were present as well as multiply twinned grains with five-fold symmetry. Microstructure observation upon annealing showed that grain growth did not occur until annealing at 400ºC. Detwinning was observed at 400ºC and higher temperatures. The ultrafine and nano twins tended to transform into dislocation cell structures and this phenomenon was driven by the excess free energy associated with the high density of grown-in twin boundaries. The five-fold twinned grains were found to be thermally stable up to 600ºC. The hardness was observed to decrease with increasing annealing temperature.

The effect of annealing conditions on the microstructure evolution of CVD nickel was investigated systematically in the present study by differential scanning calorimetry, optical microscopy and transmission electron microscopy (TEM), upon both ex-situ and in-situ annealing. TEM observation revealed the as-deposited CVD nickel possessed a bi-modal grain structure, with large columnar grains embedded in nanocrystalline matrix. Ultrafine and nano growth twins were present as well as multiply twinned grains with five-fold symmetry. Microstructure observation upon annealing showed that grain growth did not occur until annealing at 400ºC. Detwinning was observed at 400ºC and higher temperatures. The ultrafine and nano twins tended to transform into dislocation cell structures and this phenomenon was driven by the excess free energy associated with the high density of grown-in twin boundaries. The five-fold twinned grains were found to be thermally stable up to 600ºC. The hardness was observed to decrease with increasing annealing temperature.

The impact of a spherical object onto a surface of a liquid, solid or granular material, is a configuration which occurs in numerous industrial and natural phenomena. The resulting dynamics can produce complex outcomes and often occur on very short time-scales. Their study thereby requires high-speed video imaging, as is done herein. This three-part dissertation investigates widely disparate but kindred impact configurations, where the impacting object is a solid steel sphere, or a molten metal droplet. The substrate, on the other hand, is either granular material, a liquid, or solid ice. Therefore both fluid mechanics and thermodynamics play a key role in some of these dynamics. Part I, investigates the penetration depth of a steel sphere which impacts onto a granular bed containing a mixture of grains of two different sizes. The addition of smaller grains within a bed of larger grains can promote a “lubrication” effect and deeper penetration of the sphere. However, there needs to be enough mass fraction of the smaller grains so that they get lodged between the larger grains and are not simply like isolated rattlers inside the voids between the larger grains. This lubrication occurs even though the addition of the small grains increases the overall packing fraction of the bed. We compare the enhanced penetration for the mixtures to a simple interpolative model based on the results for monodispersed media of the constitutive sizes. The strongest lubrication is observed for large irregular shaped Ottawa sand grains, which are seeded with small spherical glass beads. Part II, tackles the topic of a molten metal drop impacting onto a pool of water. When the drop temperature is far above the boiling temperature of water, a continuous vapor layer can form at the interface between the metal and water, in what is called the Leidenfrost phenomenon. This vapor layer can become unstable forming what is called a vapor explosion, which can break up the molten metal drop. We study the details of these explosions and characterize the metal debris. We contrast the results for two different metals, i.e. tin and a special metal alloy called Field’s metal. For tin the drop solidifies and forms a porous foam-like solid, whereas the Field’s metal breaks up into a multitude of spherical beads, with a range of sizes as small as a few microns. We attribute this difference to the much lower melting point of the Field’s metal, which is only 60oC, compared to 230oC for the tin. This allows more fragmentation of the Field’s metal drop before it solidifies. When the temperature of the impacting metal is increased, high-speed imaging reveals a sequence of up to three vapor explosions, each of increasing intensity. We measure the acceleration of the vapor interface and compare the size-distribution of the microbeads to the fastest growing instability mode of the corresponding Rayleigh-Taylor instability. Part III, investigates the coefficient of restitution when a steel sphere impacts on an ice surface. As observed in earlier studies the restitution coefficient is largest for the smallest impact velocities, where the surface is not greatly fragmented. Our focus is on greatly heating the sphere up to 400oC to investigate how the thermal load affects the short term interaction of the sphere with the ice. We see a clear trend where hotter spheres rebound less than cold spheres. We also track the speed of ice-fragments ejected during the earliest stages of the impact.

The purpose of this study was to investigate to what extent bi-modal input improves the word segmentation ability of L2 learners of Japanese. Accurately identifying words in continuous speech is a fundamental process for comprehending the overall message, but studies show that second language (L2) learners often find this task difficult, even when all individual words are familiar to them (e.g. Field, 2003; Goh, 2000). This is where the combination of written and audio input (bi-modal input), like when providing captions in the target language, could be helpful because it can provide orthographical image of the sound they hear, which in turn makes the input more intelligible (Charles & Trenkic, 2015). This study was implemented through a single-case design (SCD), where 12 third-year Japanese learners at a public university in the Midwestern United States underwent a semester-long pre-post design experiment. Participants watched a series of Japanese documentary with sound and captions (bi-modal input) throughout the semester. Before and after viewing each video, participants took Elicited Imitation Tasks (EIT) as the pre-post-tests, as well as at the beginning and at the end of the semester. The result showed that most participants improved their EIT scores throughout the semester, even to utterances from videos and speakers to which they had not been exposed. This study provided evidence that bi-modal input has the potential to help learners’ internal phonological representations of lexical items to become more stable and sophisticated, which would in turn contribute to L2 Japanese learners’ speech processing efficiency.

The research work presented in this thesis introduces novel approaches for both visual region of interest extraction and visual feature extraction for use in audio-visual automatic speech recognition. In particular, the speaker‘s movement that occurs during speech is used to isolate the mouth region in video sequences and motionbased features obtained from this region are used to provide new visual features for audio-visual automatic speech recognition. The mouth region extraction approach proposed in this work is shown to give superior performance compared with existing colour-based lip segmentation methods. The new features are obtained from three separate representations of motion in the region of interest, namely the difference in luminance between successive images, block matching based motion vectors and optical flow. The new visual features are found to improve visual-only and audiovisual speech recognition performance when compared with the commonly-used appearance feature-based methods. In addition, a novel approach is proposed for visual feature extraction from either the discrete cosine transform or discrete wavelet transform representations of the mouth region of the speaker. In this work, the image transform is explored from a new viewpoint of data discrimination; in contrast to the more conventional data preservation viewpoint. The main findings of this work are that audio-visual automatic speech recognition systems using the new features extracted from the frequency bands selected according to their discriminatory abilities generally outperform those using features designed for data preservation. To establish the noise robustness of the new features proposed in this work, their performance has been studied in presence of a range of different types of noise and at various signal-to-noise ratios. In these experiments, the audio-visual automatic speech recognition systems based on the new approaches were found to give superior performance both to audio-visual systems using appearance based features and to audio-only speech recognition systems.