136879 research outputs found
Sort by
Wind Tunnel Data Quality Assessment and Improvement Through Integration of Uncertainty Analysis in Test Design
No full textThe practical application of uncertainty quantification in wind tunnel testing is not consistently or proactively applied. Although there is a solid methodology to quantify uncertainty, the resources required to implement this methodology at the pace of testing while adapting to the unique designs for each test are rarely available. This research combines the use of Monte Carlo simulations for uncertainty quantification with a decision-based integration of uncertainty estimates into the test design process and test execution. This implementation reduces the resources required to routinely quantify uncertainty to a practicable level and aims to proactively affect data quality by incorporating uncertainty estimates into early test design decisions. This methodology is used in the design, execution, and data analysis of a wind tunnel test at the Oran W. Nicks Low-Speed Wind Tunnel (LSWT) at Texas A&M University. The test analyzes the uncertainty in the aerodynamic coefficients and performance parameters of an aircraft test model. After quantifying the uncertainty of the aerodynamic coefficients, this research investigates a potential elemental error source in the measurement of static aerodynamic coefficients due to oscillating nonlinear aerodynamic loads. Notable results from this research include the demonstration of integrating uncertainty analysis with test design in a practical way, reduction of uncertainty intervals in aircraft performance parameters measured in the LSWT by an average of more than 90% through this integration, and experimental evidence of an elemental error source from oscillating nonlinear aerodynamic loads in the measurement of static aerodynamic coefficients
Three-Dimensional Simulations of Ductile Fracture Under Arbitrary Loadings
No full textFracture leads to billions of dollars of losses worldwide every year, leading to material waste in manufacturing or disrupting the safe operation of load bearing components. For structures capable of plastic deformation, the ever-increasing demands on performance under extreme environments, combined with the challenges of an-all-experiments based approach, require the development of reliable and predictive failure models under real-world situations. In addition, new design paradigms and the development of strong, lightweight materials for use in thin-walled structures test the limits of classical methods based on linear elastic fracture mechanics. After half a century of porous material yield function development, there is still no sound basis for predicting pore-mediated ductile failure under general loadings. Tremendous progress has been achieved for modeling failure assuming mesoscopically homogeneous deformation at appropriate length scales. The inherent limitations of available theories due to the neglect of what has recently been termed unhomogeneous yielding at such scales are numerous. For example, no existing theory can predict failure in a simple torsion specimen on a sound physical basis, let alone under more general shear-dominant loadings.
In this work, a data-driven approach is followed to develop a porous material plasticity yield function that accounts for porosity, void shape and orientation. High-throughput computational limit load analysis is used to this end. The same dataset is employed to calibrate evolution equations developed on the basis of Eshelby concentration tensors. A comprehensive constitutive theory, named HUNNY, is then formulated which is applicable under general loading conditions. The theory is akin to crystal plasticity but with dependence on the resolved normal stress. In the isotropic limit, dependence upon all stress invariants is rationalized. The predictive capabilities of the theory are assessed against a large set of micromechanical unit cell calculations under combined tension and shear loading. Several realizations of the theory are implemented as user-defined subroutines to enable three-dimensional structural simulations of crack initiation and growth. Illustrations are given to simulate ductile failure in a round notched bar and a top-hat shear specimen developed at the Sandia National Laboratories. Finally, the formulation is extended to deal with more complex hexagonal materials exhibiting plastic anisotropy, such as magnesium alloys
Optimizing Produced Water Treatment in the Permian Basin: The Role of Indirect Evaporative Cooling in Hydraulic Fracturing
No full textThis study investigates the novel application of Indirect Evaporative Cooling (IEC) for treating produced water in hydraulic fracturing operations within the Permian Basin. Utilizing a lab-scale IEC system, we conducted fifteen experiments to assess Total Dissolved Solids (TDS) and volume reduction in produced water, considering variations in initial water volume, composition, and TDS concentration. The experiments included a range of synthetic samples with salinities from 0 to 70,000 ppm and real-produced water from different Permian Basin regions (Delaware, North Midland, and South Midland).
Our findings reveal that IEC's efficiency in TDS removal, achieving near 100% effectiveness for both synthetic and real produced water samples, is primarily influenced by the presence of the most abundant salts rather than the overall TDS concentration. This highlights the system's capability to handle high salinity and diverse impurities typical in oil and gas production waters. Additionally, the IEC system proved to be significantly less energy-intensive compared to traditional thermal evaporation methods.
The economic assessment of IEC versus standard evaporation methods for treating one barrel of water further demonstrates its viability. The study concludes that IEC offers a highly effective, environmentally sustainable, and economically feasible solution for high TDS oil field water treatment. It stands out as a promising alternative to conventional technologies, with potential applications extending across various industrial facilities. This research paves the way for future exploration to maximize the potential of IEC in addressing the wastewater challenges in hydraulic fracturing
Assessing Orthodontic Compliance with Tailored vs Generic Reminders
No full textThe purpose of this study is to evaluate the effect of daily automated tailored reminders compared to generic ones on the improvement of oral hygiene compliance.
A blinded, prospective, randomized controlled trial was designed to evaluate the effects of the content of reminders. Subjects were recruited from patients undergoing orthodontic treatment at the Texas A&M University College of Dentistry, Department of Orthodontics and were treated with fixed full appliances in both arches. Subjects were randomly assigned to either a tailored text message group or a generic text message group. There were 68 subjects recruited who were 12 to 17 years of age. Oral hygiene was measured at the beginning of the study and again 8 weeks later.
The generic reminder group had significant improvements in oral hygiene compliance from timepoint 1 to timepoint 2. Decreases from T1 to T2 were 1.20 to 0.59, 1.88 to 1.10, and 3.56 to 2.90 for bleeding index (BI), gingival index (GI), and plaque index (PI), respectively (p<0.001). The tailored reminder group had significant improvements in oral hygiene compliance from timepoint 1 to timepoint 2 as well. Decreases from T1 to T2 were 1.33 to 0.58, 2.02 to 1.12, and 3.75 to 2.81 for bleeding index (BI), gingival index (GI), and plaque index (PI), respectively (p<0.001). All initial values in the tailored group and total decreases for all three periodontal tests were higher than in the generic group but were not statistically significant.
Tailored daily reminders are not more effective at improving oral hygiene compliance than generic daily reminders
Deciphering Cell Systems: Machine Learning Perspectives and Approaches for the Analysis of Single-Cell Data
No full textThis doctoral dissertation delves into the application of machine learning techniques in molecular biology, exploring gene expression regulation at the single-cell level and navigating the intricacies of cellular biology. The study specifically focuses on the utilization of modern neural networks to address cell-cell communications, gene function inference, and decipher protein expression. These applications aim to elucidate the complex interactions governing cellular behavior, as evidenced by the analysis of single-cell RNA sequencing (scRNA-seq) data.
In pursuit of these goals, I have developed and implemented advanced computational methodologies that combine systems biology and modern neural networks techniques. These methods are specifically crafted to manage the high-dimensionality and complexity of single-cell data, facilitating a more nuanced comprehension of genotype-phenotype relationships.
This research makes a significant contribution to the field of computational biology by proposing the use of neural networks to tackle the longstanding optimization problem in manifold learning. Furthermore, the study investigates generative models for learning gene regulatory networks and simulates gene knockout at the single-cell resolution. Lastly, the research delves into enhancing the interpretability of black box neural network models, applying them to multimodality data.
This research also contributes to the cell biology field by first providing an in-depth analysis of cell-cell interactions, highlighting how these interactions shape cellular behavior and influence disease progression. In addition, this research investigates gene function prediction, focusing on how gene knockouts can affect cellular phenotypes and their potential therapeutic implications. Lastly, this research looks into how gene expression patterns translate into protein expression and how accurately and interpretably this translation process can be predicted. This aspect of this research yields important insights into the functional implications of gene expression, which may be applied to the understanding of disease mechanisms and drug responses.
This research serves as a valuable resource because, in addition to the three introduced tools, it provides a comprehensive overview of state-of-the-art methodologies and their respective applications in the analysis of single-cell data within the recent years.
In conclusion, this doctoral dissertation represents a significant contribution to the field of computational biology and cellular biology by providing novel methods and insights into the genotype-phenotype relationships at the single cell level. These methods and discoveries not only improve our understanding of cellular behavior, but also pave the way for the creation of novel therapeutic strategies, thereby potentially enhancing our ability to combat a wide range of diseases
Geometric Deep Learning for Science: Prediction, Generation, and Symmetries
No full textDeep learning has significant potentials in accelerating the progress of science research. However, the data in most science problems are geometric data, or graph data, which brings many unique challenges. First, designing label-invariant data augmentations for geometric data is challenging. Second, regular deep generative models need to be dramatically modified to suit for 2D molecular graphs, 3D molecular geometries, and periodic materials. In this dissertation, we study these challenges and propose several novel methods to tackle them.
We first propose GraphAug, a novel automated data augmentation method aiming at computing label-invariant augmentations for graph classification. GraphAug uses an automated augmentation model to avoid compromising critical label-related information of the graph, thereby producing label-invariant augmentations at most times. To ensure label-invariance, we develop a training method based on reinforcement learning to maximize an estimated label-invariance probability. Second, we propose GraphDF, a novel discrete latent variable model for 2D molecular graph generation based on normalizing flow methods. GraphDF uses invertible modulo shift transforms to map discrete latent variables to graph nodes and edges. We show that the use of discrete latent variables reduces computational costs and eliminates the negative effect of dequantization.
Third, we propose G-SphereNet, a novel autoregressive flow model for generating 3D molecular geometries. G-SphereNet employs a flexible sequential generation scheme by placing atoms in 3D space step-by-step. We propose to determine 3D positions of atoms by generating distances, angles and torsion angles, thereby ensuring both invariance and equivariance. In addition, we propose to use spherical message passing and attention mechanism for conditional information extraction.
Finally, we propose SyMat, a novel symmetry-aware periodic material generation method. SyMat generates atom types and lattices with a variational auto-encoder model. In addition, SyMat employs a score-based diffusion model to generate atom coordinates based on a novel coordinate diffusion process. We show that SyMat is theoretically invariant to all symmetry transformations of materials.
We demonstrate the effectiveness of our proposed methods with comprehensive benchmark experiments. In the future, we will explore developing novel predictive models for the prediction of Hamiltonian matrices and accelerating the generation of SyMat by stochastic differential equation based diffusion models
Amino Acid Type and Concentration Impact on Endothelial Nitric Oxide Synthase in Restructured Hams
No full textThis study investigated amino acid types, either singly or in combination, at varying concentration levels to determine which was most effective as a substrate for the endothelial nitric oxide synthase system (eNOS) enzyme to generate nitric oxide for its evaluation as an alternative curing system. Restructured hams were manufactured with pork semimembranosus muscle with a 20% brine consisting of salt, sugar, phosphate and sodium erythorbate and addition of either L-arginine (Arg), L-citrulline (Cit) or in combination (Arg/Cit) at concentrations of 1000, 3000 or 5000ppm. A nitrite (NaNO2) control (200 ppm) was also included. Hams were cooked to (71��C), chilled, vacuum packaged and analyzed on day 1, 7, 28 and 56 of refrigerated (4��C) storage for residual nitrate (RNO3) nitrite (RNO2) and nitroslyhemochromagen (NO-Heme). Sensory panel and textural attributes were analyzed on day 28. For RNO3 values an interaction (p=0.0001) was observed for amino acid type and concentration. Trends suggested 1000ppm concentration for amino acid treatment combinations resulted in higher NO3 values. The main effects of amin acid type and concentration did not affect RNO2 values in the restructured hams however all amino acid treatments produced ~2/3 of the amount of RNO2 than nitrite control. An amino acid type x concentration interaction (p=0.01) NO-Heme values. An amino acid x storage day interaction (p=0.001) NO-Heme values was observed, however, no amino acid treatment or concentration was more effective at generating NO-Heme values. Amino acid type influenced Cured Ham ID (p=0.004) and Ham Flavor Aftertaste (p=0.006) with Arg exhibiting scores closest in value to the nitrite control. Amino acid type also affected Soured Aromatic (p=0.03) and Chemical/Medicinal/Metallic (p=0.001) with Cit treated ham having the highest values for both attributes. An amino acid x concentration (p=0.0001) interaction was observed for objective Hardness values with Arg treated hams values increasing as concentration increased and exhibiting closest values to nitrite control. The data from this study suggests that Arg most effectively cures restructured hams by activation of the eNOS system to generate NO and that a concentration of 1000ppm may be sufficient
Investigating Deformation and Sediment Dispersal During Andean Mountain Building in the Western Cordillera of Southern Peru
No full textThe central Andes are the archetypal modern cordilleran margin. An assessment of the timing, style, and position of deformation and associated sediment dispersal remains incomplete, especially in the Western Cordillera and forearc. Here, much of the region is overlain by Neogene extrusive igneous products from the modern Andean arc that obscure the exposures and evidence of crustal deformation. This study exploits exposures in deeply incised canyons that provide key insights into the deformational and depositional records of the Western Cordillera and forearc in southern Peru. New U-Pb zircon geochronology, structural field mapping, aerial drone 3D modeling, sediment provenance modeling, and fault kinematic forward modeling results are integrated to constrain deformation timing and style, and sediment provenance of the forearc basin. Geochronologic results from two outcrops of thrust faults that verge toward the subduction trench, paired with outcrop interpretations and forward modeling, provide detailed accounts of fault kinematics in the forearc. One fault, termed here the Aplao thrust fault, shows evidence of at least three distinct slip events throughout a long-lived history: initial compressional deformation is constrained here to between early Cretaceous and ending prior to 45.24 Ma; syndepositional deformation between 45.24 Ma and ending between 30 and 26.67 Ma; and a final episode ending before 26.67 Ma. The other structure, termed here the Toran fault, also displays multiple phases of deformation: the structure initiated as a Jurassic normal fault related to pre-Andean extension, followed by normal fault inversion and three identified compressional slip events constrained to (1) between early-middle Jurassic and ending prior to 26.67 Ma, (2) 15.98���14.07 Ma, and (3) post14.07 Ma. Sediment provenance modeling reveals upsection unroofing of the Western Cordillera and recycling of forearc basin fill, with minimal contribution from distal sources in the Altiplano or Eastern Cordillera. Results from this investigation are integrated with published depositional and deformational age constraints to place forearc deformation into context with broader orogenic controls on Andean deformation. Protracted compression in the forearc was coincident with an Eocene���early Miocene episode of flat or shallow slab subduction. This ancient slab shallowing event also drove inboard deformation of the Eastern Cordillera. This new record of shortening and unroofing in the forearc while deformation was also ongoing in the Eastern Cordillera is evidence of widespread and distributed out-of-sequence deformation and consistent with the Andean orogen being in a protracted phase of subcritical taper. The observed thrust fault geometries are consistent with a broadly bivergent Andean cordilleran system and emphasize the role of selective reactivation and inversion of inherited structures on deformation localization. These results contribute towards a complete characterization of Andean deformation, emphasizing the need for additional investigation into the cause of this long-lived compressional deformation in the forearc, even when deformation was focused far inboard from the trench
Essays on Microeconometrics
No full textThis dissertation consists of three essays on Microeconometrics and Applied Econometrics.
Chapter 2 provides nonparametric identification results in first-price auctions with unknown collusion schemes. This chapter shows that regardless of the unknown collusion scheme, collusive bidders in a partial cartel can be identified from winning bids, identities of winners, and auction-specific covariates satisfying an exclusion restriction. It can be shown that the value distributions of collusive bidders are identified under two types of cartels characterized by McAfee and McMillan (1992). This chapter proposes a test procedure to recover the identities of collusive bidders. The test method is applied to California highway procurement auctions.
Chapter 3, a joint work with Yonghong An, Matthew Gentry, and Daiqiang Zhang, examines the economic impacts of anti-bid-shopping legislation. We exploit the policy changes in Washington state to evaluate the impacts of anti-bid-shopping legislation on both ex-ante and ex-post procurement auction outcomes. With the engineer���s estimate of the cost and work types of projects, we can classify the contracts affected or not affected by anti-bid-shopping legislation. We estimate the effects of the policy using Regression Discontinuity Design (RDD) or Difference-in-Difference (DID). Findings indicate that the requirement for subcontractor listing can increase the winning bid while reducing the final total payment to the primary contractor. Moreover, disallowing substitutions may expedite project completion time and reduce the final total payment.
Chapter 4 proposes nonparametric estimation and uniform inference on counterfactual distributions in the presence of proxy variables. A two-step series estimator is developed to consistently estimate counterfactual distributions and their functionals under a general framework with latent variables. The uniform asymptotic theory is built on strong approximations of Gaussian processes introduced in Chernozhukov et al. (2013). The two-sample series counterfactual distribution estimator is applied to both simulations and an empirical study of the wage gap, in which the Armed Forces Qualifying Test (AFQT) score is used as a proxy variable for premarket human capital. The difference in the distribution of premarket human capital explains the racial wage ga
Efficacy and Utility of MPP51AA2-Traited Cotton as a Management Tool for Cotton Fleahoppers [Pseudatomoscelis seraitus (Reuter)]
No full textIn Texas the cotton fleahopper (Pseudatomoscelis seriatus (Reuter)) is considered a highly economically damaging pest of cotton (Gossypium hirsutum L.). Current control methods rely heavily on the use of foliar applied chemical insecticides during the growing season. The Mpp51Aa2.834_16 gene in cotton (ThryvOn) has proven effective against thrips and Lygus spp. with piercing and sucking feeding behaviors, suggesting the trait may also provide similar efficacy on cotton fleahoppers.
Field trials were conducted in 2019, 2020, and 2021 comparing a ThryvOn cultivar to a non-traited isoline under insecticide-treated and untreated conditions. While cotton fleahopper population differences between the traited and non-traited plants were not consistently noted during the pre-bloom squaring period, there was a consistent increase in square retention in cotton expressing Mpp51Aa2 relative to non-traited cotton. Additionally, cotton expressing Mpp51Aa2 offered similar square protection relative to non-traited cotton treated with insecticides for cotton fleahopper. These findings indicate that the Mpp51Aa2 protein should provide benefits of delayed nymphal growth, population suppression, and increased square retention.
In the choice assay, feeding by cotton fleahoppers significantly reduced square retention in the non-traited cotton to 46%, while ThryvOn cotton retained 60% of squares. In the no-choice assay, cotton fleahopper nymph feeding significantly reduced square retention in the non-traited cotton to 61%, whereas ThryvOn cotton was unaffected. Our findings indicate that the Mpp51Aa2 protein influences cotton fleahopper feeding preference and the susceptibility of cotton plants to damage caused by cotton fleahoppers.
To evaluate the feeding behavior of the cotton fleahoppers on ThryvOn cotton an electropenetrography (EPG) coupled with a Giga-8 DC EPG amplifier was used to monitor the probing activity cotton fleahopper nymphs on ThryvOn and non-traited cotton squares. Nymphs were placed on a square for 8 hours and waveforms were characterized as non-probing, cell rupturing, and ingestion. There were significantly more cell rupturing events on ThryvOn (14.8) than on non-traited squares (10.3) but were no differences in ingestion events. However, the duration of ingestion events were significantly shorter at 509s on ThryvOn compared to 914s on non-traited squares. The results of this study provide evidence that ThryvOn affects the feeding behavior of cotton fleahoppers