Browsing by Subject "experimental"
Now showing 1 - 7 of 7
Results Per Page
Sort Options
Item A Recommendation System for Preconditioned Iterative Solvers(2011-02-22) George, ThomasSolving linear systems of equations is an integral part of most scientific simulations. In recent years, there has been a considerable interest in large scale scientific simulation of complex physical processes. Iterative solvers are usually preferred for solving linear systems of such magnitude due to their lower computational requirements. Currently, computational scientists have access to a multitude of iterative solver options available as "plug-and- play" components in various problem solving environments. Choosing the right solver configuration from the available choices is critical for ensuring convergence and achieving good performance, especially for large complex matrices. However, identifying the "best" preconditioned iterative solver and parameters is challenging even for an expert due to issues such as the lack of a unified theoretical model, complexity of the solver configuration space, and multiple selection criteria. Therefore, it is desirable to have principled practitioner-centric strategies for identifying solver configuration(s) for solving large linear systems. The current dissertation presents a general practitioner-centric framework for (a) problem independent retrospective analysis, and (b) problem-specific predictive modeling of performance data. Our retrospective performance analysis methodology introduces new metrics such as area under performance-profile curve and conditional variance-based finetuning score that facilitate a robust comparative performance evaluation as well as parameter sensitivity analysis. We present results using this analysis approach on a number of popular preconditioned iterative solvers available in packages such as PETSc, Trilinos, Hypre, ILUPACK, and WSMP. The predictive modeling of performance data is an integral part of our multi-stage approach for solver recommendation. The key novelty of our approach lies in our modular learning based formulation that comprises of three sub problems: (a) solvability modeling, (b) performance modeling, and (c) performance optimization, which provides the flexibility to effectively target challenges such as software failure and multiobjective optimization. Our choice of a "solver trial" instance space represented in terms of the characteristics of the corresponding "linear system", "solver configuration" and their interactions, leads to a scalable and elegant formulation. Empirical evaluation of our approach on performance datasets associated with fairly large groups of solver configurations demonstrates that one can obtain high quality recommendations that are close to the ideal choices.Item An Empirical Survey of the Analytic / Continental Divide(Texas Digital Library, 2023-05-16) Barta, WalterWhat is the difference, if any, between analytic philosophy and continental philosophy? Contemporary philosophers tend to identify themselves, and so contemporary philosophy is divided, roughly along these lines. Thus, two fields of discourse, the analytic and the continental, have emerged. But is this division substantive or rhetorical? Is it helpful or harmful? What are its effects inside and outside the discipline? Can these categories be concretely defined, or are they nebulous and provisional? Some metaphilosophers, like C. G. Prado, in A House Divided, have suggested that analytic philosophy and continental philosophy are distinguishable based on respective focuses, and that the divide falls along lines, such as these: problem solving versus political engagement, philosophical analysis versus synthesis (e.g. a genealogical orientation), bottom-up versus top-down approaches to disciplinary hierarchy. Others have suggested that the analytic/continental divide is erroneous, illusory, merely rhetorical, or simply nonexistent. Each of these constitutes a claim that is empirically testable against the fields of discourse of the two (sub)disciplines through a comparison of textual data: the presence of logically valent as opposed to politically valent words, manner of reference to historical figures, and distribution of textual citations. Inasmuch as these correspond to quantifiable metrics, each of the aforementioned questions may have a discrete and demonstrable empirical answer. The use of empirical and statistical surveys in philosophy is unusual, but its recent implementation by David Chalmers (as in his recent “What do Philosophers Believe?”) and other influential philosophers has proven its ability to develop important insights.Item Bursting and spalling in pretensioned U-beams(2009-12) Dunkman, David Andrew; Bayrak, Oguzhan, 1969-; Jirsa, James O.An experimental program was conducted at the Ferguson Structural Engineering Laboratory of The University of Texas at Austin, under the auspices of Texas Department of Transportation (TxDOT) Research Project 5831, to investigate the tensile stresses that develop in the end regions of pretensioned concrete U-beams at transfer of prestress. Understanding the effect of these “bursting” and “spalling” stresses is essential in order to design standard details that might lead to reliably-serviceable end regions. Two full-scale beam specimens, designed to be worst-case scenarios for bursting and spalling, were fabricated. Each beam had one square and one highly skewed end. Extensive instrumentation, including strain gages on transverse and lateral reinforcing bars, was employed in the end regions of these U-beams. Experimentally determined bursting and spalling stresses in these bars were compared to results of past projects (from the literature) investigating I-beams and inverted T-beams. Preliminary recommendations are made for changes in the standard reinforcing details for U-beam end regions. Such recommended details will be tested in the upcoming phase of Research Project 5831.Item Descriptive epidemiology of 2007 homebuilt aircraft accidents and accident and fatality rates(2009-06-30) Gregory Giancarlo Shaskan; Richard Jennings; Melchor Antunano; Daniel FreemanAbstract: Homebuilt aircraft are those that the builder has completed at least 51% of the construction themselves. From 1993 National Transportation and Safety Board data, homebuilt aircraft made up only 3% of the flying hours but were involved in 10% of the general aviation accidents in the United States. This research investigates if and how these rates have changed over the past 15 years. The National Transportation and Safety Board Aviation Accident Database and Synopses website was queried for all types of United States homebuilt accidents during 2007. Select variables were used to compare homebuilt accidents and fatality rates with non-homebuilt aircraft rates. Human and mechanical errors were implicated in homebuilt accidents similar to non-homebuilt general aviation accidents. Homebuilt accident rates per 100,000 flight hours are significantly higher than non-homebuilt accident rates (26.44 +3.36 vs. 4.89 + 0.29, p<0.001) Homebuilt fatality rates per 100,000 flight hours are also significantly higher (8.67 + 1.92 versus 1.56 + 0.16, p<0.001). Safety efforts that focus on human and mechanical factors may reduce overall accident rates of both homebuilt and non-homebuilt aircraft.Item Examining A Hypersonic Turbulent Boundary Layer at Low Reynolds Number(2013-05-15) Semper, Michael ThomasThe purpose of the current study was to answer several questions related to hypersonic, low Reynolds number, turbulent boundary layers, of which available data related to turbulence quantities is scarce. To that end, a unique research facility was created, instrumentation was developed to acquire data in the challenging low Reynolds number (low density) domain, and meaningful data was collected and analyzed. The low Reynolds number nature of the boundary layer (Re_theta = 3700) allows for tangible DNS computations/validations using the current geometry and conditions. The boundary layer examined in this experiment resembled other, higher Reynolds number boundary layers, but also exhibited its own unique characteristics. The Van Driest equivalent velocity scaling method was found to perform well, and the log layer of the law of the wall plot matched expected theory. Noticeably absent from the data was an overlap region between the two layers, which suggests a different profile for the velocity profiles at these low Reynolds number, hypersonic conditions. The low density effects near the wall may be having an effect on the turbulence that modifies this region in a manner not currently anticipated. The Crocco-Busemann relation was found to provide satisfactory results under its general assumptions. When compared to available data, the Morkovin scaled velocity fluctuations fell almost an order of magnitude short. Currently, it is not known if this deficit is due to inadequacies with the Strong Reynolds Analogy, or the Morkovin scaling parameters. The trips seem to promote uniformity across the span of the model, and the data seems to generally be in agreement across the spanwise stations. However, additional information is needed to determine if two-dimensional simulations are sufficient for these boundary layers. When the turbulent boundary layer power spectra is analyzed, the result is found to follow the traditional power law. This result verifies that even at low Reynolds numbers, the length scales still follow the behavior described by Kolmogorov. Moving downstream of the trips, the peak RMS disturbance value grows in amplitude until it reaches a critical value. After this point, the peak begins to decrease in amplitude, but the affected region spreads throughout the boundary layer. Once the influenced region covers a significant portion of the boundary layer, transition occurs.Item Measured and Predicted Rotor-Pad Transfer Functions for a Rocker-Pivot Tilting-Pad Journal Bearing(2012-02-14) Wilkes, Jason ChristopherMany researchers have compared predicted stiffness and damping coefficients for tilting-pad journal bearings (TPJBs) to measurements. Most have found that direct damping is consistently overpredicted. The thrust of this research is to explain the difference between measured and predicted stiffness and damping coefficients for TPJBs, and to provide some confidence to designers that TPJB dynamic coefficients can be accurately predicted. Most analytical models for TPJBs are based on the assumption that explicit dependence on pad motion can be eliminated by assuming harmonic rotor motion such that the amplitude and phase of pad motions resulting from radial and transverse rotor motions are predicted by rotor-pad transfer functions. In short, these transfer functions specify the amplitude and phase of pad motion (angular, radial, translational, etc.) in response to an input rotor motion. A new pad perturbation model is developed including the effects of angular, radial, and circumferential pad motion and changes in pad clearance due to pad bending compliance. Though all of these pad variables have previously been included in different analyses, there are no publications containing perturbations of all four variables. In addition, previous researchers have only perturbed the journal, while both the bearing and journal motions are perturbed in the present analysis, and the applicability of comparing rotor-perturbed bearing impedance predictions to impedances measured on a bearing-perturbed test rig is discussed. This perturbation model was implemented in a Reynolds-based TPJB code to predict the frequency-dependent bearing impedances and rotor-pad transfer functions. Direct measurements of pad motion during test excitation were recorded to produce measured transfer functions between rotor and pad motion, and a comparison between these measurements and predictions is given. Motion probes were added to the loaded pad (having the static load vector directed through its pivot) of a 5-pad TPJB to obtain accurate measurement of pad radial and tangential motion, as well as tilt, yaw, and pitch. Strain gages were attached to the side of the loaded pad to measure static and dynamic bending strains, which were then used to determine static and dynamic changes in pad curvature (pad clearance). Good agreement was found between the amplitude of the measured and predicted transfer functions concerning radial and transverse pad motions throughout the range of speeds and loads tested, while pad tilt was moderately underpredicted. For the bearing investigated, radial pad motions resulting from pivot compliance were as large as 60% of the radial component of shaft motion when operating at 4400 rpm under heavily loaded conditions. Hence, if a dynamic load applied to the shaft resulted in a shaft displacement of 25 microns (1 mil), the pad would displace radially 15 microns (0.6 mils), and the fluid film height would only decrease by 10 microns (0.4 mils). The consequence of this pad motion is that fluid film stiffness and damping forces produced by relative rotor-pad motions are significantly reduced, resulting in a bearing having significantly less direct stiffness and damping than predicted. A similar effect occurs when shaft motions produce significant changes in pad clearance due to pad compliance. For the pad tested here, the measurements show that predicting TPJB stiffness and damping coefficients without accounting for pad and pivot compliance will produce large errors, and is not advised. Transverse pad motion was predicted and observed. Based on phase measurements, this motion is lightly damped, and appears to be caused by pivot deflection instead of slipping. Despite observing a lightly damped phase change, an increase in magnitude at this natural frequency was not observed. Predicted direct stiffness and damping for unit loads from 0-3200 kPa (0-450 psi) fit through 1.5? running speed are within 18% of measurements at 4400 rpm, while predictions at 10200 rpm are within 10% of measurements. This is a significant improvement on the accuracy of predictions cited in literature. Comparisons between predictions from the developed bearing model neglecting pad, pivot, and pad and pivot flexibility show that predicted direct stiffness and damping coefficients for a model having a rigid pad and pivot are overestimated, respectively, by 202% and 811% at low speeds and large loads, by 176% and 513% at high speeds and high loads, and by 51% and 182% at high speeds and light loads. While the reader is likely questioning the degree to which these predictions are overestimated in regard to previous comparisons, these predictions are based on measured operating bearing clearances, which are 20-30% smaller than the cold bearing clearances that previous comparisons were based on. The effect of employing a full bearing model (retaining all of the pad degrees of freedom) versus a reduced bearing model (where only journal degrees of freedom are retained) in a stability calculation for a realistic rotor-bearing system is assessed. For the bearing tested, the bearing coefficients reduced at the frequency of the unstable eigenvalue (subsynchronously reduced) predicted a destabilizing cross-coupled stiffness coefficient at the onset of instability within 1% of the full model, while synchronously reduced coefficients for the lightly loaded bearing required 25% more destabilizing cross-coupled stiffness than the full model to cause system instability. This overestimation of stability is due to an increase in predicted direct damping at the synchronous frequency over the subsynchronously reduced value. This increase in direct damping with excitation frequency was also seen in highly loaded test data at frequencies below approximately 2?running speed, after which direct damping decreased with increasing excitation frequency. This effect was more pronounced in predictions, occurring at all load and speed combinations. The same stability calculation was performed using measured stiffness and damping coefficients at synchronous and subsynchronous frequencies at 10200 rpm. It was found that both the synchronously measured stiffness and damping and predictions using the full bearing model were more conservative than the model using subsynchronously measured stiffness and damping. This outcome contrasts with the comparison between models using synchronously and subsynchronously reduced impedance predictions, which showed the subsynchronously reduced model to be the most conservative. This contrast results from a predicted increase in damping with increasing excitation frequency at all speeds and loads, while this increase in damping with increasing excitation frequency was only measured at the most heavily loaded conditions.Item Multi-Location Evaluation of Agronomic Traits in Maize Hybrids(2012-12-12) McKee, Michael 1982-Maize (Zea mays L.) is one of the main crops grown in the United States. Genetic improvement over the last century has seen a shift from using open-pollinated varieties to single cross hybrids. This has resulted in major grain yield gains and improved management methodologies. However, there is still concern about reduced genetic diversity in elite corn germplasm and the potential effects this could have on future maize productivity in the presence of numerous abiotic and biotic pressures. One solution to this issue is the incorporation of exotic germplasm into existing maize improvement programs. This exotic material must be evaluated and characterized because too much or poorly matched exotic material can lead to reduced productivity. The use of multiple environments representative to the target improvement area is the best way to determine the true potential of certain material. The objectives of this research were to: i) estimate the responses of hybrids to aflatoxin and their agronomic performance across a range of environments under inoculation with Aspergillus flavus; ii) identify the hybrids within each group that exhibit the lowest levels of contamination; iii) analyze the relationship between agronomic performance and aflatoxin accumulation; and iv) determine how Genotype x Environment interactions affect these traits. Agronomic data was collected in ten Texas environments in 2005 for hybrids created from yellow, white, and Quality Protein Maize material that was crossed with one of two elite temperate inbred testers, LH195 or LH210. Response to aflatoxin was measured in eight of these environments. U.S. commercial hybrids were used as checks. Significant differences between hybrids were observed at different environments for different traits. Overall the experimental hybrids had lower aflatoxin accumulation than the commercial checks. They also yielded lower and had lower test weights and 1000 kernel weights. However, there were some hybrids that were competitive with the commercial checks for these agronomic traits. The incorporation of this material into established U.S. lines could be beneficial with regards to aflatoxin accumulation and kernel quality, which could ultimately translate to higher yields and crop quality.