In this study, activity allocation is analyzed in two stages, by modeling the activity data of household individuals. In the first stage, activity participation patterns of individuals are examined for both out-of-home maintenance and discretionary activities. The influence of socio demographic, trip related, person, household role, role of constraints, and household related information are investigated in this stage using empirical data from the San Francisco Bay Area Travel Survey. In the second stage, interactions between Internet and communications technology use, and activity and travel patterns are analyzed by developing a series of econometric models, using the Bay Area Travel Survey. These interactions are examined at three levels: (a) whether the Internet is used or not, if yes, used for what purposes, (b) what is the relation between physical and virtual activities in terms of internet use, and (c) to what extent does the Internet influence the daily trip patterns such as frequency and duration. Internet use, ICT device, predicted probabilities, socio demographic, work, and household related variables are used as explanatory variables in these models. The results from this study are described and have important implications for congestion management, air-quality mitigation and travel demand forecasting.

This study focuses on computationally efficient techniques for reliability-based design optimization (RBDO), and develops extensions that are useful in robust design. Various RBDO techniques can be grouped into nested, decoupled or single loop approaches. This thesis extends the decoupled and single loop methods to include standard deviations as design parameters and a technique to replace the traditional first order analytical method with simulation for more accurate reliability assessment. The methods are extended to robust design and their applicability is investigated. The accuracy and computational efficiency of the various RBDO methods are compared. The proposed methods are applied to the problem of maximizing the star rating (NCAP safety ratings) of vehicle design.

Highway incidents are a major source of traffic congestion, the paramount operational problem on urban freeways. Considerable national attention has been directed towards the effective management of traffic incidents to alleviate congestion-related problems on freeways. Evaluating such incident management programs requires developing models that can characterize incident properties. This study provides two statistical models for predicting incident durations and secondary incident causation probabilities based on eleven months of incident data from the Nashville Traffic Management Center. The non-linear relationship between the predictor variables (descriptive incident characteristics) and the response variables (incident duration and secondary incident causation probability) was accounted for by using logistic regression methods. In the secondary incident causation model the clearance time of an incident was used as the sole explanatory variable to predict the secondary incident causation probabilities.

Single-plate connections, also called shear tabs, are an economical and efficient way to transfer load from one member to another, and are often found in steel-frame buildings. One type of shear tab is the extended shear tab, and it is unique in that the eccentricity of the bolt centroid to the weld centroid is greater than the 3½ allowed for standard shear tabs. This larger eccentricity develops a larger moment in the plate, sometimes requiring a thicker plate. There are instances, however, when this extra material cost is offset by other savings, and the use of extended shear tabs is economical. One such instance is when a beam or girder frames into the web (weak axis) of a column that is supporting a moment connection on its flanges (strong axis), as in a rigid frame. According to the AISC Manual of Steel Construction 3rd ed., an official design procedure does not exist for extended shear tabs. It is the goal of this Masters Thesis to investigate different extended shear tabs in the aforementioned framing situation to determine their strength capabilities. A design procedure will be proposed and various shear tab connections will be designed using the proposed procedure. Full-scale testing of the connections will be undertaken in the laboratory to accrue experimental data and the connections will be simulated for analytical analysis using finite element models. The proposed procedure assumes a reduced eccentricity due to the stiffening of the continuity plates. The connection strength is contingent upon the capacity of the continuity plates, which must meet minimum requirements. The test results supported the hypothesis, but the results of the finite element models were insufficient for making any definitive recommendations.

Structural control through bracing modification is presented in this thesis. First existing structural control devices discussed. Then the idea of how to adjust the stiffness properties of a structural frame is introduced. After studying the effect of induced stress on the stiffness of structural elements is considered, analytical studies using modal and transient analysis in ANSYS are undertaken. This analysis involves using various bracing configurations with different magnitudes of induced stress to find the natural frequency of each and how these models response to the El Centro earthquake excitation. Following the analytical studies, laboratory experiments are carried out using sinusoidal input and El Centro excitation. The analytical and experimental results are compared and the effect of bracing and pre-induced forces is evaluated for single-story and double-story frames.

This study proposes a general methodology for mechanical/structural fatigue reliability analysis under multiaxial loading. A new characteristic plane approach is proposed to predict the fatigue crack initiation and propagation life under general multiaxial loading. The proposed fatigue analysis methodology is combined with advanced finite element analysis to predict the fatigue life of railroad wheels. Parametric studies are performed to identify the most important factors affecting the service life of railroad wheels. Uncertainties in material properties, external applied loadings, structural geometries and observed failure profiles are incorporated in the fatigue damage model to evaluate the reliability. A general methodology for stochastic fatigue life prediction is proposed, which combines random process theory, response surface method, design of experiments and the Monte Carlo simulation technique. The proposed methodology is applicable to damage tolerant design and maintenance scheduling of various mechanical and structural components.

This study investigates day-to-day dynamics in an urban traffic network induced by internal and external factors under real-time information. A robust cost network optimization algorithm to account for the randomness of trip time is proposed and its application is demonstrated for the static traffic assignment problem. In addition, a simulation-based day-to-day network analysis framework is developed by using an agent-based approach to modeling user behavior under information. Unique features of this framework includes its day-to-day simulation capability, quantification of system performance in non-equilibrium states, capability to model and study the influence of system shocks, richer representation of user behavior, and a wide array of system performance measures that enable assessment of reliability and trip time jointly. Computational experiments are used to investigate the effect of the following experimental factors: user behavior rules, users responsiveness, demand reduction control measures, and unplanned supply shocks (incidents). Insights and models along these lines will have important implications for congestion mitigation, improvement of travel time reliability, and assessment of different travel demand management strategies.

Full-scale testing of large engineering systems for assessing performance could be infeasible and expensive. With the growth of advanced computing capabilities, model-based simulation plays an increasingly important role in the design of such systems. When computational models are developed, the assumptions and approximations introduce various types of errors in the code predictions. In order to accept the model prediction with confidence, the computational models need to be rigorously verified and validated. When the input parameters of the model are uncertain, model prediction has uncertainty. On the other hand, the validation experiments also have measurement errors. Thus model validation involves comparing prediction with test data when both are uncertain. Appropriate validation metrics that address various uncertainties and errors are developed in this study, for both component-level and system-level models. Both classical and Bayesian statistics are used for this purpose. Another goal of model validation is to extend what we can learn about the models predictive capability within the tested region to an inference about the predictive capability in the untested region of actual application and quantify the confidence in the extrapolation being performed. Sometimes the response quantity of interest in the target application may be different from the validated response quantity. Validation inferences may need to be extrapolated from nominal to off-nominal (tail) conditions or component level data may have to be used to make partial inference on the validity of system-level prediction. In all of the above cases, the methodology of Bayesian networks is developed to extrapolate inferences from the validation domain to the application domain. This study also proposes a methodology to estimate the errors in computational models and to include them in reliability-based design optimization (RBDO). Various sources of uncertainties, errors and approximations in model form selection and numerical solution are included in a first order-based RBDO methodology.

A reinforced concrete structure may degrade through a combination of coupled physical, chemical and mechanical deterioration processes. This study develops an integrated finite element-based computational framework to quantitatively evaluate the resultant mechanical responses of reinforced concrete structures exposed to multiple coupled environmental loadings. Heat transfer and associated thermal expansion/contraction process, moisture transport and associated wetting expansion/drying shrinkage process, carbon dioxide transport and associated carbonation process, chloride penetration process, reinforcement corrosion and rust expansion process due to chloride contamination, as well as the subsequent crack initiation and propagation are taken into account. Interaction effects among multiple coupled processes are considered by adjusting the transport properties of concrete through multifactor equations. The crack propagation analysis is implemented using a smeared cracking approach, and the crack-induced accelerated deterioration process is modeled using relative crack density concepts. The developed computational framework is general and flexible, which makes it possible to include additional deterioration processes. Due to a large amount of random variability and uncertainty existing in the material properties of concrete, various environmental loadings and the developed computational framework, an appropriate approach for assessing the structural durability is to combine the deterministic computational framework with advanced stochastic modeling. The most influential random variables in the computational framework are identified, and a nonlinear response surface is established through combining experimental design with multilevel regression. Monte Carlo simulation is then performed with the response surface to predict the statistics of service life and assess the associated time-dependent failure probability of reinforced concrete structures. Sensitivity analysis is implemented to quantify the influence of each random variable on the estimation of service life.

Modern aerospace systems are increasingly complex and multidisciplinary, and are required to achieve dramatic improvements in performance, cost effectiveness, and reliability. For this reason, program requirements must include reliability standards and design processes must incorporate methods for assessing and engineering to these requirements. Probabilistic analysis methods have been well developed as a means to assess reliability by propagating uncertainties in the form of stochastic variables through performance analysis models. In addition, reliability-based design optimization (RBDO) has been given considerable attention in recent years as a means to make design decisions to ensure reliability goals are met. However, for multidisciplinary systems, these tools exacerbate the already intensive computational effort required for design optimization. This dissertation develops efficient methods for the probabilistic (or reliability-based) design of multidisciplinary systems, considering system integration from two fronts. The first front is the integration of disciplinary analyses at a single level, for which methods are presented to improve the efficiency of reliability analysis and optimization. These methods are applied in the design of the power supply system for an unmanned aerial vehicle. On the second front, two alternative strategies are developed for the synthesis of reliability-based design across two design levels (as distinguished by the scope, detail, and fidelity of the performance analysis). In the second of these strategies, model form error uncertainty is presented as a valuable metric representing the fidelity of disciplinary analysis models, allowing probabilistic methods to determine the sensitivity of the system design to model fidelity as a basis for model selection. These concepts are applied to the design of conceptual reusable launch vehicle geometry and that of a component liquid hydrogen tank.