Thesis (Ph. D.)--University of Washington, 2005. / Vita. Includes bibliographical references (leaves 117-119).

Bachelor-Arbeit Univ. St. Gallen, 2006.

Universiẗat, Diss., 2005--Münster (Westfalen).

abstract: The focus of this study is statistical characterization of the significant duration of strong ground motion time histories. The significant duration is defined as the time needed to build up between five and seventy five (SD575) and ninety five percent (SD595) of the energy of a strong motion record. Energy is measured as the integral of the square of the acceleration time history and can be used to capture the potential destructiveness of an earthquake. Correlations of the geometric means of the two significant duration measures (SD575 and SD595) with source, path, and near surface site parameters have been investigated using the geometric mean of 2,690 pairs of recorded horizontal strong ground motion data from 129 earthquakes in active plate margins. These time histories correspond to moment magnitudes between 4.8 and 7.9, site to source distances up to 200 km, and near surface shear wave velocity ranging from 120 to 2250 m/s. Empirical relationships have been developed based upon the simple functional forms, and observed correlations. The coefficients of the independent variables in these empirical relationships have been determined through nonlinear regression analysis using a random effects model. It is found that significant duration measures correlate well with magnitude, site to source distance, and near surface shear wave velocity. The influence of the depth to top of rupture, depth to the shear wave velocity of 1000 m/s and the style of faulting were not found to be statistically significant. Comparison of the empirical relationship developed in this study with existing empirical relationships for the significant duration shows good agreement at intermediate magnitudes (M 6.5). However, at larger and smaller magnitude, the differences between the correlations developed in this study and those from previous studies are significant. / Dissertation/Thesis / Ph.D. Civil and Environmental Engineering 2011

Extra-state war, a conflict between a state and non-state actor outside of the state’s borders, is an understudied phenomenon. In order to begin a discussion on this topic, this paper seeks to understand the factors that affect extra-state war duration. Using literature on interstate and intrastate wars, I hypothesize that military intervention in support of the non-state actor, an equitable distribution of third-party military interventions, and economic intervention in the form of support will increase war duration. I also hypothesize that military intervention on behalf of the state and diplomatic intervention by a third party will decrease duration. I test my hypotheses using a multi-method resource design. First, using quantitative data drawn from extra-systemic wars between 1816 and 2007, I find support for the hypothesis that military intervention on behalf of the non-state actor increases duration. In addition, I find that military intervention on behalf of the state also increases war duration. I supplement my regression analysis with a series of case studies on the Western Saharan War, Cisplatine War, and Mozambican War of Independence. I find that an equitable distribution of military interventions as well as economic support increase duration. Diplomatic intervention, on the other hand, decreases war duration. Taken together, my findings suggest that the various forms of intervention play a crucial role in explaining extra-state war duration.

The current study tests the use of vowel duration as a cue for the voicing of the following stop by Saudi ESL learners. It is mainly constructed on the Language Transfer Theory (LTT) established by Gass and Selinker (1994), the Ontogeny Phylogeny Model (OPM) formulated by Major (2001), the Perceptual Assimilation Model (PAM) developed by Best (1994, 1995), and the Speech Learning Model (SLM) developed by Flege (1995). The instrument contained 30 English monosyllabic minimal pairs of the type CVC. The participants, who were living in the US, consisted of two groups: 5 advanced Saudi ESL learners with linguistic knowledge and 10 advanced Saudi ESL learners without linguistic knowledge. Results showed that both groups of participants (those with linguistics background and those without linguistics background) were fairly accurate at predicting final voiceless Coronal and Dorsal stops after a shorter vowel. On the other hand, they both had equal difficulties predicting voiced stops in this environment. This would explain why participants in both groups overall seemed to be listening for the voicing status of the final stop and disregarding differences in vowel length as a predictor of that voicing. Individual participants, overall, were quite uniform in their responses regardless of any background in linguistics. This finding suggests that participants in both groups relied much more on the actual voicing of the final stops than they did on the length of the preceding vowel.

M.Ing. (Electrical & Electronic Engineering) / Three single phase current fed converters operating in pwm mode, are combined as an active power filter of 15 kVA. The undesired limit cycle shown by the system is eliminated by feedback loop compensation. Analog, computer and analytical models which predict remarkably similar system responses are used to investigate the technique. The analytical model involves describing function methods. The losses in the converters are modelled, and used as a feedback reference to allow losses in the system to be accurately covered from the supply. The system behaves as an accurate, fast active filter for distortion in power networks. Results illustrating this include the time and frequency response of the system for different reference currents. This is shown for the computer simulation as well as the practical system. Some more results include the response of the system to both a balanced and an unbalanced set of references in the three phases. The last results shown, illustrate the ability of the system to compensate a non-linear load.

[pt] Os desvanecimentos por multipercurso atmosférico são a principal causa de degradação do desempenho de enlaces rádio digitais de alta capacidade, acentuando os efeitos do ruído térmico e da interferência intersimbólica na taxa de bits errados. A previsão da distribuição cumulativa de probabilidades destes desvanecimentos em função de parâmetros do enlace e características rádio climáticas da região é de fundamental importância no projeto de sistemas rádio digitais terrestres. Existem na literatura técnica métodos para previsão do desvanecimento em enlaces individuais, mas não para um tronco de microondas formado por vários enlaces. Medidas realizadas no Brasil indicam que a usual suposição de decorrelação total entre desvanecimentos profundos em enlaces consecutivos em um mesmo tronco é pessimista, podendo levar a uma indisponibilidade combinada maior do que a real. Neste trabalho foram analisados dados experimentais de 15 enlaces na região centro-oeste do Brasil e 5 enlaces no Japão e desenvolvido um modelo empírico de previsão da distribuição conjunta do desvanecimento por multipercurso atmosférico conjunta em enlaces adjacentes. O modelo apresenta boa concordância com os resultados experimentais e fornece uma nova expressão para o cálculo da indisponibilidade em troncos de microondas formados por vários enlaces. Os dados experimentais foram utilizados ainda para modelar a duração de eventos de atenuação por multipercurso em enlaces em visibilidade. Mostrou-se que a distribuição de duração de eventos é bem caracterizada por distribuições de Weibull. Esta caracterização permite a distinção entre eventos de perda de qualidade (duração inferior a 10 segundos) e de indisponibilidade (duração superior a 10 segundos). / [en] Multipath fading is the main cause of performance degradation in high capacity line-of-sight microwave links, increasing enhancing the effects of the thermal noise and intersymbol interference in the bit error rate. Tested methods are available in the literature for the prediction of cumulative probability distribution multipath fading in individual link but not in a microwave network formed by several tandem links. Measurements carried out in Brazil indicate that the usual assumption that deep fade events in consecutive links are totally uncorrelated may be pessimistic for tropical regions and may lead to the prediction excessive multihop links unavailability. Experimental data obtained in measurements of multipath fading in 15 pairs of tandem links located in the center- western region of Brazil were analyzed to provide cumulative distributions of attenuation in each pair. Similar results are available in the literature for 5 pairs of links in Japan. Based on these results, an empirical model was developed for the prediction of the joint probability distribution the multipath fading in adjacent links was developed. The attenuation simultaneously exceeded in both links for given percentage of time can be predicted as a function of the equiprobable values of attenuation exceeded in the individual links, path lengths and operation frequency. The predicted distributions show good agreement with the experimental results and provide an expression for the calculation of unavailability of line-of- sight tandem links. The experimental data had been used still to model the duration of events of attenuation for multipath of links in visibility. One revealed that the distribution of duration of events well is characterized by distributions of Weibull. This characterization allows the distinction between events of loss of quality (lowest duration at 10 seconds) and of non-availability (higher duration the 10 seconds).

This study represents an attempt to model the disaggregation of daily rainfall at a point into individual storms for the summer rainy season in southeast Arizona. The purpose of the model is to simulate the number of storms per day and the amount, duration and time of occurrence of each event while maintaining daily statistics. Various storm ratios are introduced to describe the disaggregation of a daily amount into individual storm amounts. Marginal distributions are fit to the storm ratios and event starting times. On multiple-event days, starting times are obtained through the use of order statistics. Conditional distributions are fit to the number of storms per day, given a daily amount, and the duration of each event, given a storm amount. The simulated data compare favorably with the observed data. Statistical tests indicate that the model adequately transforms daily rainfall into a sequence of individual storms.

M.Ing. (Electrical & Electronic Engineering) / Please refer to full text to view abstract