Many important engineering phenomena such as turbulent flow, fluid-structure interactions, and climate diagnostics are chaotic and sensitivity analysis of such systems is a challenging problem. Computational methods have been proposed to accurately and efficiently estimate the sensitivity analysis of these systems which is of great scientific and engineering interest. In this thesis, a new approach is applied to compute the direct and adjoint sensitivities of time-averaged quantities defined from the chaotic response of the Lorenz system and the double pendulum system. A stabilized time-integrator with adaptive time-step control is used to maintain stability of the sensitivity calculations. A study of convergence of a quantity of interest and its square is presented. Results show that the approach computes accurate sensitivity values with a computational cost that is multiple orders-of-magnitude lower than competing approaches based on least-squares-shadowing approach.

Novel aircraft configurations tend to be sized by physical phenomena that are largely neglected during conventional fixed wing aircraft design. High-fidelity fluid-structure interaction that accurately models geometric nonlinerity during a transient aeroelastic gust response is critical for sizing the aircraft configuration early in the design process. The primary motivation of this research is to develop a continuum shape sensitivity method that can support gradient-based design optimization of practical and multidisciplinary high-fidelity analyses. A local continuum sensitivity analysis (CSA) that utilizes spatial gradient reconstruction (SGR) and avoids mesh sensitivities is presented for shape design derivative calculations. Current design sensitivity analysis (DSA) methods have shortcomings regarding accuracy, efficiency, and ease of implementation. The local CSA method with SGR is a nonintrusive and element agnostic method that can be used with black box analysis tools, making it relatively easy to implement. Furthermore, it overcomes many of the accuracy issues documented in the current literature. The method is developed to compute design derivatives for a variety of applications, including linear and nonlinear static beam bending, linear and nonlinear transient gust analysis of a 2-D beam structure, linear and nonlinear static bending of rectangular plates, linear and nonlinear static bending of a beam-stiffened plate, and two-dimensional potential flow. The analyses are conducted using general purpose codes. For each example the design derivatives are validated with either analytic or finite difference solutions and practical numerical and modeling considerations are discussed. The local continuum shape sensitivity method with spatial gradient reconstruction is an accurate analytic design sensitivity method that is amenable to general purpose codes and black box tools. / Ph. D.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

A collection of salt-sensitive vacuolar (ssv) Saccharomyces cerevisiae mutants were selected for analysis in an attempt to reveal components of the osmosensing signal transduction pathway. A previous screen of these mutants had been used to select those with an impaired glycerol response to salt stress. In this study the glycerol-3-phosphate dehydrogenase activity was measured in all the strains which showed reduced glycerol accumulation to demonstrate a corresponding low enzyme activity which could be caused by lack of signalling through the high osmolarity glycerol (HOG) response pathway. However, enzyme activity was found to be impaired in only one of the strains tested. This demonstrated that the measurement of glycerol accumulation is not a particularly useful screening method for defects in the HOG pathway. The activity of the promoter of CTTl, another stress-responsive gene, was measured in selected ssv strains using a lacZ reporter gene attached to the CTTl promoter stress response element. This gave further information about the stress-responsiveness in the strains tested. CTTl promoter activity did not correlate with GPDH activity in all of the strains tested. As CTTl is subject to control by more than one type of stress the results imply that in at least one of the strains another stress response may be impaired. The VACl homologue (VACIH) on chromosome XIV was characterised as a candidate for one of the 55V genes, 55W7. Although it was demonstrated not to be 55W 7, a role for VACIH in vacuolar protein sorting was discovered. The Δvaclh strain also displayed class E vacuolar morphology. Sequence analysis and complementation experiments demonstrated that VACIH is identical to VPS27.

The study focuses on the sensitivity and uncertainty analysis of occupancy-related parameters using Energyplus modeling method. The model is based on a real building Zero Energy Lab in Discovery Park, at University of North Texas. Four categories of parameters are analyzed: heating/cooling setpoint, lighting, equipment and occupancy. Influence coefficient (IC) is applied in the sensitivity study, in order to compare the impact of individual parameter on the overall building energy consumption. The study is conducted under Texas weather file as well as North Dakota weather file in order to find weather’s influence of sensitivity. Probabilistic collocation method (PCM) is utilized for uncertainty analysis, with an aim of predicting future energy consumption based on history or reference data set. From the study, it is found that cooling setpoint has the largest influence on overall energy consumption in both Texas and North Dakota, and occupancy number has the least influence. The analysis also indicates schedule’s influence on energy consumption. PCM is able to accurately predict future energy consumption with limited calculation, and has great advantage over Monte Carlo Method. The polynomial equations are generated in both 3-order and 6-order, and the 6-order equation is proved to have a better result, which is around 0.1% compared with real value.