Doctoral Dissertations

Permanent URI for this collection

https://hdl.handle.net/10919/11041

Browse

Now showing 1 - 20 of 17452

CREATING FACULTY GUIDELINES FOR COLLABORATIVE ONLINE INTERNATIONAL LEARNING: DESIGN AND DEVELOPMENT
Mukuni, Candido (Virginia Tech, 2024-12-20)
Parametric Analysis and Life Cycle Cost Assessment for Optimizing PCM Application in Exterior Walls in the Kingdom of Saudi Arabia
Alamri, Uthman Abdullah (Virginia Tech, 2024-12-19)
The Kingdom of Saudi Arabia (KSA) aims to reduce CO2 emissions and mitigate its environmental impact as part of Vision 2030. The building sector has high energy consumption, particularly due to elevated cooling demands, which make up 70% of residential energy use. This is largely caused by uninsulated thermal mass and subsidized electricity rates. In addition, Vision 2030's housing projects and labor shortage necessitate alternatives to current housing standards. Modular housing offers a solution to the labor shortage, but its success depends on lighter materials. This study proposes replacing thermal mass with PCM in modular housing and, investigates this using EnergyPlus simulations. The research investigated the optimal placement and thickness of PCM to maximize its thermal performance in SIPs. PCM application reduced the model total site energy by 12.9% to 13.7% with a thickness of 0.5–2.0 cm and significantly reduced the HVAC energy consumption by 37% to 39%. In this study, we developed LCCA models to assess the cost-effectiveness of PCM by establishing a price range per square foot that aligns with the energy savings that are usable in the KSA. This study also identified the maximum PCM production cost based on the LCCA analysis to ensure its investment use in KSA's construction industry.
Unraveling the Role of EphA4 in Immune-Mediated Arteriogenesis After Ischemic Stroke
Ju, Jing (Virginia Tech, 2024-12-19)
Stroke, a life-threatening condition, primarily resulting from ischemic events often caused by occlusion of the middle cerebral artery (MCA). Pre-existing leptomeningeal collateral (LMC) vessels connect MCA branches to anterior or posterior arteries, situated along the brain's cortical surface or meninges, under healthy conditions these vessels remain dormant due to their small diameters and relatively low flow velocity. LMCs serve as vascular redundancies that retrogradely re-supply blood to help salvage the penumbra following cerebral vascular occlusion. Their outward growth or remodeling (arteriogenesis) is essential for promoting cerebral reperfusion and preventing tissue damage after ischemic stroke. Increased fluid shear stress on collateral vessel wall activates arteriogenesis result in the activation of the endothelium and subsequent recruitment of peripheral-derived immune cells (PDICs), which have been shown to aid this unique adaptive process in other organ systems, however their role and mechanism(s) involved in LMC remodeling in stroke has not previously been evaluated. Initial findings suggest the EphA4, a well-established axonal growth and guidance receptors, plays a novel role in LMC arteriogenesis. This dissertation examined PDIC-specific functions of EphA4 using GFP labeled bone marrow chimeric mice subjected to permanent middle cerebral artery occlusion (pMCAO). We assessed immune cell population changes, infarct volume, functional recovery, characterized subtypes of infiltrated immune cell, and measured collateral vessel diameters. Additionally, we explored the Tie2-mediated PI3K signaling pathway in peripheral-derived monocyte/macrophages (PDM) treated with soluble Tie2-Fc and a PI3K p110α inhibitor. The results from this dissertation show that loss of PDIC-specific EphA4 led to increased collateral remodeling, associated with decreased infarct volume, improved cerebral blood flow, and functional recovery within 24 hours post-pMCAO. The crosstalk between EphA4-Tie2 signaling in PDMs, regulated through PI3K/Akt axis, inhibited pial collateral remodeling. In conclusion, our findings highlight the negative regulatory role of PDM-specific EphA4 in collateral growth and remodeling by inhibiting Tie2 function via the PI3K regulated pathway. Peripheral myeloid-derived EphA4 emerges as a new regulator of cerebral vascular injury and neuroinflammation following acute ischemic stroke.
Artificial Intelligence Mystification and Data Financialization: An Intensive Case Analysis of User-data and Value Realization in the Platform Firm
Alexander, Andrew William (Virginia Tech, 2024-12-19)
The social relation between the platform firm and its users is defined by engagement with platform infrastructure and the rendition of this engagement into data. This type of data is often compared to gold, oil, and other fungible goods. User-data, however, are not generally accounted for as intangible assets and their value, economic and otherwise, is not transparent. I problematize a priori assumptions about a direct line from data to capital by asking: How do platform firms realize value from user-data? I engage the question through structural analysis by abstraction in an intensive case study of two transnational platform firms. I use qualitative content analysis to analyze annual and earnings reports, terms of service agreements, and internal documents from 2017 through 2023 with Atlas.ti data analysis software. The findings reveal a perceptual disconnection between user-data inputs and artificial intelligence (AI) related service and product outputs. I argue that the platform functions as digital real-estate to extract monetary and data rents and securitizes its users through a process of mystifying the relationship between user-data and AI related products, services, and infrastructures. I posit the processes of AI mystification and user securitization as mechanisms of value realization that suggest the dilation of an entrenched social relation rather than a divergence from capitalism. The study places financialization as a critical factor in data seeking and calls for the inclusion of the finance, insurance, and real estate (FIRE) sectors in future research of data and AI in political economy. I suggest a focus on data as property and its ownership and control as regulatory articulation points for future policy formation.
Hepatic Lipid metabolism in Neonatal Pigs
Gerrard, Samuel David (Virginia Tech, 2024-12-19)
Medium chain-fatty acids (MCFA) are a group of fatty acids containing hydrocarbon chains between 6-12 carbons. They are rapidly absorbed and taken up by cells which has led to their incorporation into neonatal formulas as an alternative source of energy. While abundant literature is available on proposed beneficial effects of MCFA at low levels of incorporation, little is known about utilization of MCFAs in formulas and the effects on growth and liver health. Therefore, we set out to test the hypothesis that MCFAs are metabolized differently than long-chain fatty acids (LCFAs) when they are the main energy substrate of the formula. We sought to investigate the mechanism that differentiates MCFA from LCFA from a metabolic and physiologic standpoint. Feeding high-fat diets of MCFA and LCFA resulted in steatosis from both classes of fatty acids. However, MCFA fed group accumulated 4  more fat in their livers than the LCFA group. Steatosis was accompanied by decreased - oxidation and increased expression of fatty acid synthetic enzymes in MCFA pigs. Peripherally, skeletal muscle displayed an upregulation of cholesterol-related genes. Lowering the amount of MCFA in the formula relieved hepatic steatosis however, only removing the MCFA source entirely from the diet lowered the steatosis below 20% of liver weight. Isolated mitochondria from pigs fed high MCFA formula were unable to oxidize pyruvate and malate as effectively as pigs without MCFA in their formula. Mitochondria also oxidized laurate more effectively when attached to carnitine. Regardless, pigs fed higher amounts of MCFA, or MCFA at any level, had higher levels of fat in their livers than the LCFA counterparts. Taken together, these data suggest that MCFA and LCFA are handled differently from a cellular perspective and MCFA changed the hepatic phenotype of neonatal pigs however, several unanswered questions arose from the completed studies.
Acoustic Frequency Domain Reflectometry
Theis, Logan Bartley (Virginia Tech, 2024-12-19)
Acoustic Frequency Domain Reflectometry (AFDR) is a novel technique employing frequency modulated continuous wave (FMCW) methods in solid acoustic waveguide reflectometry. It is particularly suited to dispersion compensation and phase compensation due to the measurement domain being the frequency domain. This work rigorously analyzes, develops, and experimentally demonstrates AFDR, alongside various compensation methods and demodulation techniques. Distributed measurement of temperature is tested using several novel signal processing algorithms for strain determination and is estimated to have a resolution of 0.58 °C over a 20 cm gauge length. An error correction algorithm to improve SNR in the measurement of strain is proposed and validated. The sensing system has a theoretical spatial resolution of 2 mm and an estimated sensing resolution limit of about 1 cm. AFDR and the associated signal processing developments are positioned to be transformative across many areas of acoustics, with significant potential for distributed sensing along an acoustic waveguide.
Adopting Plant-rich Dietary Patterns and Reducing Red and Processed Meat Intake: Examining How Diverse U.S. Food and Health Systems Actors May Support Sustainable Diet Transitions for American Adults
Stanley, Katherine Ellen (Virginia Tech, 2024-12-19)
Expert bodies recommend that populations adopt plant-rich dietary patterns and consume less red and processed meats (RPM) as a high-impact climate action. This PhD dissertation describes three studies that examined how diverse food and health systems actors may encourage sustainable diet transitions for Americans to support human and planetary health. Study one examined U.S. adults' perceptions, beliefs, and behaviors toward plant-rich dietary patterns. The International Food Information Council's Food and Health Surveys (2012–2022) were analyzed using crosstabulation and chi-square analyses. Consumers' recent RPM intake trends were mixed. Despite interest in sustainable products and principles, few U.S. adults followed plant-rich dietary patterns or purchasing practices. Leadership and coordinated action are needed to incentivize Americans to adopt plant-rich dietary behaviors. Study two conducted a systematic scoping review of media campaigns that promoted plant-rich dietary patterns and traditional and novel plant-based proteins, and that encouraged or discouraged RPM products to Americans (1917-2023). Of 84 media campaigns identified, corporate marketing (58.6%) campaigns were most prevalent compared to public information (13.8%), corporate sustainability (12.6%), countermarketing (5.7%), social marketing (4.6%), and public policy (4.6%). Civil society campaigns promoted plant-rich dietary patterns, but only one campaign was evaluated. U.S. government, academia, businesses, and civil society should commit adequate resources and evaluate media campaigns to support a sustainable diet transition for Americans that prioritizes traditional and novel plant-based proteins. Study three explored U.S. Food is Medicine (FIM) experts' views on how plant-rich dietary patterns and other sustainable diet practices could be incorporated into FIM interventions. Twenty semi-structured interviews were conducted among U.S. food and health systems actors and analyzed using inductive and deductive thematic analysis. Results indicated that many FIM actions support human and planetary health, but the co-benefits are not often discussed. The FIM movement is a unique opportunity to promote food and health systems changes that support human and planetary health, but key challenges require coordinated action across sectors. The three studies in this PhD dissertation collectively addressed knowledge gaps, used novel conceptual frameworks, and offered recommendations to inform U.S. food and nutrition policies, programs, and research to encourage sustainable diet transitions for American adults.
Astrocyte and vascular changes contribute to Alzheimer's disease
Li, Jiangtao (Virginia Tech, 2024-12-19)
Alzheimer's disease (AD), the most prevalent age-related neurodegenerative disorder, is defined by the pathological accumulation of amyloid-β (Aβ) peptides, neurofibrillary tangles (NFTs), neuronal loss, and the activation of astrocytes and microglia. One of the early indicators of AD is a global reduction in cerebral blood flow (CBF), which precedes significant plaque formation and cognitive decline. This persistent decrease in CBF, along with diminished oxygen and glucose delivery to the brain, is thought to contribute to neurodegeneration, although the underlying mechanisms remain unclear. Astrocytes, critical regulators of both Aβ clearance and CBF, have garnered increasing attention in AD research. Astrocytes, one of the most abundant cell types in the central nervous system, play a vital role in maintaining overall brain health and function. In AD, astrocytes express key AD-related genes, including APP, PSEN1, PSEN2, and APOE. While astrocyte gene expression alterations have been observed, the relationship between these transcriptomic changes, protein expression, and cellular function requires further investigation. This dissertation examines astrocyte and vascular changes in AD using a well-described preclinical AD mouse model: hAPPJ20 mice. First, a multi-omics analysis of cortical astrocyte gene and protein expression was conducted at 3, 6, 12, and 18 months in female J20 and wild-type (WT) mice, revealing significant gene and protein expression differences linked to normal aging and AD progression. Several overlapping gene-protein pairs were identified as potential biomarkers for AD treatment and diagnosis. Gene Ontology analysis highlighted enriched pathways related to inflammation, disrupted metabolism, and vascular dysfunction starting at 6 months. Additionally, pathway analysis revealed apoptotic pathways were enriched in astrocytes isolated from diseased tissue. Further analysis revealed for the first time that astrocytes significantly decline by 12 months in the cortex and hippocampus of J20 AD-disease mice. Nest, we explored vascular network remodeling and amyloid-β (Aβ) accumulation in this same model. In male J20 mice, 40% of the total pial arterial Aβ accumulation was found in the meningeal vascular network by 12 months, while females showed around 20%. Aβ deposition was associated with increased vessel diameter and tortuosity of pial collateral vessels. AD mice also exhibited reduced blood flow in the cortical meningeal arteries and significant enlargement of pial collateral vessels compared to wild-type mice.
The Precarious Man: Measuring masculine discrepancy and its relationships with aggression and misogyny
Aadahl, Sarah (Virginia Tech, 2024-12-17)
In the studies of men and masculinity, most of the focus has been on masculine dysfunction strain, or the strain males feel as it relates to the various expectations of masculinity. In contrast, the research on discrepancy strain (or the strain males feel when they fail to meet these expectations) is limited. Unlike dysfunction strain, there are not any widely accepted and utilized scales measuring discrepancy. By combining identity theory and general strain theory with gender and feminist theory, my goal is to examine how masculine discrepancy may be related to the endorsement of aggression and misogyny. The aim of this dissertation is to develop and validate a scale to operationalize masculine discrepancy as it is theorized; this scale will then be used to examine the following research questions: does masculine discrepancy impact males' individual endorsement of aggression and misogyny? And if so, are these impacts moderated by failure to meet particular aspects, or "pillars," of masculinity? I created a masculine discrepancy scale that more accurately operationalizes the theoretical concept of masculine discrepancy. First, I synthesized various masculinity scales, namely the Male Role Norms Inventory, the Conformity to Masculine Norms Inventory, and the Man Box scale to develop assessments of males' masculine ideals ("ideal") and perceptions of their lived experiences ("actual" or "experiences"). By comparing ideal to actual, we can calculate a discrepancy score, where a score of 0 indicates consistency, and scores further from 0 indicate discrepancy. These scores are calculated both as an overall assessment of discrepancy and by particular pillars of masculinity. Following a pilot study of undergraduate sociology students, 1,000 males above the age of 18 were surveyed. These surveys were conducted via Cint panel distribution in December 2023. I then use factor and cluster analysis as well as regression analyses to test the following hypotheses: (1) Masculine Discrepancy Stress will have a positive relationship with aggression and misogyny. (2) Higher endorsement of masculine ideals and lived experiences will be associated with higher levels of endorsement of aggression and misogyny when compared to lower endorsement of ideals and experiences. (3) Masculine Discrepancy, or the difference between idealized and lived masculine experiences, will be negatively correlated with aggression and misogyny. Individuals with positive discrepancies (lived experiences surpassing their masculine ideals) will exhibit lower levels of aggression and misogyny compared to those with negative discrepancies (masculine ideals surpassing lived experiences). (4) Certain masculine ideals and experiences will have stronger associations with aggression and misogyny than others. (5) Cluster analysis of ideals and experiences will reveal four groups of males: Norm-Favoring Consistents: High ideals, high experiences; Norm-Favoring Discrepants: High ideals, low experiences; Norm-Rejecting Consistents: Low ideals, low experiences; Norm-Rejecting Discrepants: Low ideals, high experiences. (6) These clusters will differ in their endorsement of aggression and misogyny. The Norm-Favoring Discrepants will exhibit the highest levels of aggression and misogyny, followed by the Norm-Favoring Consistents. The Norm-Rejecting Consistents will have lower levels than both of the Norm-Favoring groups, and the Norm-Rejecting Discrepants will have the lowest endorsement of aggression and misogyny. Hypotheses 1 through 4 were supported, while hypotheses 5 and 6 had limited support, as the two "Consistent" groups did not clearly differ as "norm-favoring" vs. "norm-rejecting."
Unbiased Filtered Rayleigh Scattering Measurement Model for Aerodynamic Flows
Warner, Evan Patrick (Virginia Tech, 2024-12-17)
The filtered Rayleigh scattering (FRS) optical diagnostic has become an attractive technique for advanced aerodynamic measurements. The appeal of FRS is that it can simultaneously quantify density, temperature, and vector velocity. Additionally, it is entirely non-intrusive to the flow since the technique leverages how laser light scatters off of molecules naturally present in the gas. Acquired FRS data considered herein is in the form of a frequency spectrum. To process this data, a measurement model for the FRS spectrum is used, where inputs to this model are the flow field quantities of interest and the output is a representative FRS spectrum. An iterative procedure on these quantities is performed until the model spectrum matches the measured spectrum. However, as observed in certain applications of this technique, there is a range of measurement configurations where the standard methods to model this spectrum do not agree with measured spectra, even at known flow conditions. This disagreement causes large bias uncertainties in determined flow field quantities. This work leverages a data-driven approach to diagnose this disagreement by utilizing an extensive FRS database. Data analysis indicates that the widely used Tenti S6 model for the Rayleigh scattering lineshape is invalid in certain operating regions. A new Rayleigh lineshape modeling methodology, the Cabannes model, is introduced that vastly improves the agreement between measured and modeled FRS signals. Analysis of the Cabannes model indicates that one only needs to use this modeling methodology for FRS and not laser Rayleigh scattering (LRS). This improved measurement model can be used to mitigate bias uncertainties, and, in turn, improve the reliability of the FRS optical instrument.
Positive Support Systems: A Qualitative Investigation into the Perceptions of Elementary School Leaders Regarding Family Engagement with School-wide Positive Behavior Intervention and Supports
Gill, Jason Martin (Virginia Tech, 2024-12-17)
Families play a vital role in their child's education both educationally and behaviorally. Schools hold family engagement events and want parent support, but schools are not including families in the planning and implementation of their School-wide Positive Behavior and Intervention Support (SWPBIS) process. Policy holds schools responsible for family engagement involvement as well as reducing discipline referrals, but there is little research on schools including family engagement with their SWPBIS. The purpose of this study was to identify elementary school administrators' overall perceptions of family involvement in SWPBIS implementation. Specifically, this study sought to identify family engagement in decision-making, barriers limiting family engagement, and family engagement activities focused on student behavior. Fourteen school administrators were interviewed and shared they have not been including families with the SWPBIS process, they need to get their school's process out to families so the families understand it and can have a voice in the school, they need training on how to involve families with the implementation process, and they need to plan events that focus specifically on their SWPBIS system. The study has created future opportunities for elementary school administrators to share ideas for involving families, utilize a common database or handbook for guidance with involving families, and ways to train school administrators on how to involve families in the SWPBIS implementation process. A suggestion for future research would be to expand the sample to include more regions of Virginia.
Developments for the Characterization of Spacecraft Proximity Operations for Improved Space Situational Awareness
Bala, Amit Gopal (Virginia Tech, 2024-08-24)
As space becomes increasingly populated by numerous individuals and organizations, the capabilities of satellites on orbit have also improved. A variety of satellites are able to perform operations in close proximity to each other in order to complete a number of different missions. Maintaining awareness of these operations helps to ensure the continued safe operations in space. This work introduces a method for identifying and characterizing rendezvous and proximity operations (RPO) in space. A Bayesian Belief Network, a probabilistic evaluation tool, is introduced in order to fuse information sources together. Various combinations of relative orbital dynamics, vehicle characteristics, and environmental conditions can be used to determine the potential intent of these close proximity operations. First, a baseline framework is developed to classify the different formations of a satellite trajectory when performing a RPO mission. Sensitivity analyses are introduced in order to understand where the assessment capabilities lose validity as uncertainty is injected into the system. Next, additions to the baseline framework are made to consider specific satellite subsystem characteristics and environmental conditions. The developed framework looks to stand as a proof-of-concept system for information fusion and the characterization of events in the spacecraft domain.
Surfactants at fluid interfaces: molecular modeling and deep learning
Ham, Seok Gyun (Virginia Tech, 2024-12-16)
Surfactants at fluid-fluid interfaces play a critical role in numerous engineering applications, including enhanced oil recovery and fire suppression by foams. This dissertation explores surfactant-laden fluid-fluid interfaces in two applications using molecular dynamics (MD) simulations and develops deep learning models to predict the interfacial properties of sur factants. The first study investigates slippage modulation at brine–oil interfaces by surfactants, which is relevant to enhanced oil recovery operations. We identified a slip length of 1.2 nm at clean decane-brine interfaces. Introducing surfactants to the interface leads to an initial linear decrease in slippage, with nonylphenol being more effective than phenol. As surfactant concentration increases, the reduction in slip length slows, ultimately plateauing at 1.4 nm and 0.5 nm for nonylphenol and phenol, respectively. The mechanisms underlying these slip modulation behaviors and the effects of surfactant tail length on interfacial slippage are examined by analyzing the molecular structure and transport properties of the interfacial fluids and surfactants. The second study focuses on oil transport across surfactant-laden fluid-fluid interfaces, which is relevant to firefighting foam applications. Despite its importance, the molecular details of this transport are not fully understood. Through MD simulations, the potential of mean force (PMF) and local diffusivity profiles of heptane molecules across surfactant monolayers was computed to evaluate their transport resistance across the interface. It was discovered that a heptane molecule experiences significant resistance when crossing surfactant-covered water−vapor interfaces. This resistance, influenced by high PMF and low diffusion in the surfactant head group region, increases linearly with surfactant density and dramatically spikes as the monolayer reaches saturation, becoming equivalent to the resistance of a 5 nm thick layer of bulk water. These observations provide insights into the design of surfactants aimed at reducing oil transport through water−vapor interfaces. The final part of the dissertation explores the development of a quantitative structure-property relationship (QSPR) model for surfactants using a graph neural network (GNN) based approach. The model was trained on 92 surfactant data points and demonstrated high accuracy (R² = 0.86 on average) in predicting critical micelle concentration, limiting surface tension, and maximum surface excess for various surfactants. The performance of the model in capturing the relationship between molecular design parameters and surfactant properties was critically evaluated. The dataset, model development, and assessments contribute to advancing surfactant QSPR models and their rational design for diverse industrial applications
Adaptive Longitudinal and Lateral Control for Autonomous Vehicles: High-Speed Platooning of Articulated Trucks
Shaju, Aashish (Virginia Tech, 2024-12-13)
Autonomous vehicle technology has seen remarkable advancements in recent years, yet significant challenges remain in ensuring robust, adaptive, and efficient control algorithms for diverse operational scenarios. This dissertation aims to address these challenges by developing and validating a generic control framework that is applicable to both independent autonomous vehicles and connected vehicle systems such as automated platoons. The versatility of the proposed framework ensures its applicability to a wide range of vehicles, including automobiles, light trucks, and rigid and articulated commercial trucks, under high-speed and complex driving conditions. The first major contribution is the development of a longitudinal control algorithm based on a nested PID structure. Designed for computational efficiency and stability, the algorithm simultaneously regulates vehicle speed and inter-vehicle distance. Its adaptability is extended to curved trajectories using an arc length-based error calculation, making it suitable for real-world scenarios. A rigorous simulation study is undertaken to demonstrate the algorithm's stability and robustness to parametric uncertainties. The second major contribution is the development of a high-speed lateral control algorithm based on a modified clothoid controller. This lateral control framework is designed to minimize lateral acceleration (improving passenger comfort and safety) and reduce cross-track errors (CTEs) across various vehicle configurations, including articulated trucks. Simulation results confirmed the superiority of the clothoid-based controller in minimizing CTEs and maintaining smooth steering profiles, even for complex vehicle configurations. Notably, tracking the steer axle center was found to significantly improve performance across all trajectory segments. The final contribution integrates the longitudinal and lateral control frameworks, enabling seamless operation in automated platooning scenarios. This integration requires adapting the longitudinal controller to curved trajectories using arc length-based calculations. Comprehensive simulations, including challenging trajectories such as dual lane changes, and actual roadways like sections of the Blue Ridge Parkway in Virginia and South Grade Road in California, validated the integrated framework. Despite minor anomalies in high-stress conditions, the results demonstrate acceptable performance in terms of spacing errors, relative velocities, lateral accelerations, and CTEs, highlighting the robustness and resilience of the proposed system. The study presents a unified control framework that bridges the gap between independent autonomous vehicles and connected vehicle systems. The generic nature of the algorithms ensures their applicability to a wide variety of vehicles and scenarios, making them a strong candidate for future deployment in autonomous systems. The findings represent significant advances toward safer, more efficient, and versatile autonomous vehicle technologies, addressing critical challenges in the path to commercial implementation
Bayesian Inference Based on Nonparametric Regression for Highly Correlated and High Dimensional Data
Yun, Young Ho (Virginia Tech, 2024-12-13)
Establishing relationships among observed variables is important in many research studies. However, the task becomes increasingly difficult in the presence of unidentified complexities stemming from interdependencies among multi-dimensional variables and variability across subjects. This dissertation presents three novel methodological approaches to address these complex associations between highly correlated and high dimensional data. Firstly, group multi-kernel machine regression (GMM) is proposed to identify the association between two sets of multidimensional functions, offering flexibility to effectively capture the complex association among high-dimensional variables. Secondly, semiparametric kernel machine regression under a Bayesian hierarchical structure is introduced for matched case-crossover studies, enabling flexible modeling of multiple covariate effects within strata and their complex interactions, denoted as fused kernel machine regression (Fused-KMR). Lastly, it presents a Bayesian hierarchical framework designed to identify multiple change points in the relationship between ambient temperature and mortality rate. This framework, unlike traditional methods, treats change points as random variables, enabling the modeling of nonparametric functions that vary by region and is denoted as a multiple random change point (MRCP). Simulation studies and real-world applications illustrate the effectiveness and advantages of these approaches in capturing intricate associations and enhancing predictive accuracy.
Thermoelectric Energy Harvesting in Harsh Environments and Laser Additive Manufacturing for Thermoelectric and Electromagnetic Materials
Sun, Kan (Virginia Tech, 2024-12-12)
This dissertation presents innovative research at the intersection of thermoelectric solutions, additive manufacturing, and nuclear safety technology, addressing critical challenges in sensor powering for extreme environments, energy harvesting, and materials fabrication. The research is divided into three key areas, each contributing to advancements in its respective domain. First, a self-powered wireless through-wall data communication system was developed for monitoring nuclear facilities, specifically spent fuel storage dry casks. These facilities require continuous monitoring of internal conditions, including temperature, pressure, radiation, and humidity, under harsh environments characterized by high temperatures and intense radiation without any penetration through their walls. The constructed system integrated four modules: an energy harvester with power management circuits, an ultrasound wireless communication system using high-temperature piezoelectric transducers, electronic circuits for sensing and data transmission, and radiation shielding for electronics. Experimental validation demonstrated that the system harvests over 40 mW of power from thermal flow, withstands gamma radiation exceeding 100 Mrad, and survives temperatures up to 195°C. The system, designed to operate stably for fifty years, enables data transmission every ten minutes, ensuring reliable long-term monitoring for nuclear safety and security. Second, the efficiency of thermoelectric generators (TEGs), unique solid-state devices for thermal-to-electrical energy conversion, was explored through a novel manufacturing approach using selective laser melting (SLM) and direct energy deposition (DED). Conventional TEG fabrication methods have limitations in achieving optimal efficiency due to design and material constraints. SLM-based additive manufacturing offers a scalable solution for creating geometry-flexible and functionally graded thermoelectric materials. This research developed a physical model to simulate the SLM and DED process for fabricating Mg2Si thermoelectric materials with Si doping. The model incorporates conservation equations and accounts for fluid flow driven by buoyancy forces and surface tension, enabling detailed analysis of process parameters such as laser scanning speed and power input. The results provided insights into temperature distribution, powder bed shrinkage, and molten pool dynamics, advancing the understanding and optimization of thermoelectric device fabrication using additive manufacturing. One step further, SLM and DED experiments were carried out to validate the simulation results and testify to the feasibility of applying laser powder bed fusion on semiconductor materials. Third, the research investigates the application of laser additive manufacturing to improve performance and reduce the production costs of magnetic materials. Soft magnetic materials, critical for various industrial applications, are fabricated using DED. The research optimizes DED printing parameters and processes through quality control experiments inspired by the Taguchi method and analysis of variance models. The resulting silicon-iron samples exhibit minimal defects and cracks, demonstrating the feasibility of the approach. Detailed optical and scanning electron microscopy, coupled with magnetic characterization, reveal that the rapid cooling process inherent to laser-based AM enables unique microstructures that enhance magnetic properties. Collectively, this work addresses pressing technological challenges in energy harvesting, materials fabrication, and extreme environment monitoring. The developed systems and methodologies have broad implications for nuclear safety, additive manufacturing, and the efficient utilization of advanced materials. By integrating interdisciplinary approaches and leveraging cutting-edge manufacturing technologies, this dissertation contributes to the advancement of sustainable and resilient solutions for modern engineering challenges.
High-dimensional Multimodal Bayesian Learning
Salem, Mohamed Mahmoud (Virginia Tech, 2024-12-12)
High-dimensional datasets are fast becoming a cornerstone across diverse domains, fueled by advancements in data-capturing technology like DNA sequencing, medical imaging techniques, and social media. This dissertation delves into the inherent opportunities and challenges posed by these types of datasets. We develop three Bayesian methods: (1) Multilevel Network Recovery for Genomics, (2) Network Recovery for Functional data, and (3) Bayesian Inference in Transformer-based Models. Chapter 2 in our work examines a two-tiered data structure; to simultaneously explore the variable selection and identify dependency structures among both higher and lower-level variables, we propose a multi-level nonparametric kernel machine approach, utilizing variational inference to jointly identify multi-level variables as well as build the network. Chapter 3 addresses the development of a simultaneous selection of functional domain subsets, selection of functional graphical nodes, and continuous response modeling given both scalar and functional covariates under semiparametric, nonadditive models, which allow us to capture unknown, possibly nonlinear, interaction terms among high dimensional functional variables. In Chapter 4, we extend our investigation of leveraging structure in high dimensional datasets to the relatively new transformer architecture; we introduce a new penalty structure to the Bayesian classification transformer, leveraging the multi-tiered structure of the transformer-based model. This allows for increased, likelihood-based regularization, which is needed given the high dimensional nature of our motivating dataset. This new regularization approach allows us to integrate Bayesian inference via variational approximations into our transformer-based model and improves the calibration of probability estimates.
Investigation of Fuel Geometry and Solid Fuel Combustion for Solid Fuel Ramjets
Gallegos, Dominic Francisco (Virginia Tech, 2024-12-10)
Solid fuel ramjets (SFRJs) are a simple means of sustaining supersonic flight. The utilization of solid fuels eliminates the need for moving parts or liquid delivery systems, and the solid fuels are typically inert, resulting in minimal handling requirements compared to solid propellants. Characteristic of SFRJ systems are the relatively high combustor velocities and the required gasification of the solid fuel prior to releasing heat through gas-phase reactions. The primary objectives of the current work were to investigate the decomposition behavior of model solid fuels typically used in SFRJ systems and to employ a novel fuel geometry to increase the flame-holding limits of an SFRJ. Two bench-scale solid fuel experiments were conducted to capture relevant performance metrics of five solid fuels. Performance parameters such as regression rates, surface temperature variations, molten layer thickness, and condensed-phase kinetic behavior were analyzed using a non-combusting laser pyrolysis experiment. Further investigations were performed for each fuel using a modified counterflow burner, which served as an analog for the boundary layer combustion in an SFRJ by introducing the effects of flame heat feedback to the fuel surface. General trends among the fuels were identified, and several mechanistic differences in the decomposition process were discussed with consideration of condensed-phase behavior. The results from the laser pyrolysis and counterflow burner studies were subsequently used as validation data for the development of a solid fuel decomposition model incorporating single-step decomposition, transient heat transfer, and surface heat losses. The developed model showed reasonable agreement with experimental pyrolysis results, particularly for regression rates and surface temperatures of polymethylmethacrylate (PMMA) and hydroxyl-terminated polybutadiene (HTPB). Investigations using two lab-scale SFRJs were conducted to determine the feasibility and performance impacts of implementing a cavity-style flame holder as a means of improving the flammability limits of a SFRJ. The results presented demonstrate the effectiveness of such a method showing that introducing a cavity flame holder enables significantly higher fuel loading in the present system. The effects of the alternate geometries on local regression rates are reported and a high local heat flux at the cavity corner is identified as a strong factor in the increased flame holding capability. The increased regression rates contribute to higher observed chamber pressures while the effects on combustion efficiency are observed to be minimal. Further investigation of the cavity geometries using an optically accessible SFRJ allowed the analysis of the reacting flow field. High-speed chemiluminescence, high-speed videography, and high-speed three-color camera pyrometry provided further insight into the reacting flow and identified key reaction regions relevant to flame holding. Observations of the spatial regression rate show similar trends to the initial experiments, revealing a large increase in regression rate associated with the cavity corner. The regression rates and observations regarding the size of the recirculation region were incorporated into a semi-empirical model describing the behavior of the recirculation region and point to the increased fuel flow rate resulting from the cavity corner as a contributing factor in the increased flammability of the cavity fuel grains.
A Multidimensional Study of Transit Ridership and Station Mode Shares in the United States: Nonlinear Effects, Data Aggregation, and Post-Pandemic Changes
Abdollahpour Razkenari, Seyed Sajjad (Virginia Tech, 2024-12-10)
Understanding the differences among public transit types allows for the development of more targeted policies at both local and regional levels. Examining how the built environment (BE) influences travel behavior (Delclòs-Alió et al.) and assessing data aggregation effects around different transit station types at the local level, along with identifying key predictors of ridership across transit modes at the regional level, offers valuable insights for policy efforts. Specifically, the dissertation comprises three studies that analyze BE-travel behavior associations and data aggregation effects locally, as well as variations in key predictors of rail and bus ridership at a regional scale within the United States. The findings emphasize the unique land-use and travel behavior associations for various public transit systems within transit catchment areas, the effects of data aggregation on BE-travel behavior models, and the critical predictors of rail and bus ridership at regional levels. The first study highlights nonlinear associations between BE attributes and commuting mode share within rail and Bus Rapid Transit (BRT) catchment areas, using data from approximately 2,790 transit stations across 34 U.S. metropolitan statistical areas. Applying a random forest model, this study reveals substantial differences between rail and BRT areas, with rail catchment areas showing greater sensitivity to BE factors in reducing car dependency. BRT, however, emerges as a viable alternative for sprawling areas lacking the compact development needed to support rail systems. The second study investigates how data aggregation influences the BE-mode share relationship around 2,794 rail and BRT stations, utilizing both inferential and machine learning approaches. Findings indicate that data aggregation affects BE-mode share models regardless of the analytical method. Optimal buffer sizes for capturing BE effects and minimizing sensitivity to data aggregation were identified as 800 meters for BRT stations and 1,000 meters for rail stations. Key BE features such as employment density, jobs per household, intersection density, residential density, distance from the central business district, job accessibility (active modes) demonstrated robustness against data aggregation for both rail and BRT stations. The third study examines changes in transit ridership predictors before and after the COVID-19 pandemic across 35 U.S. metropolitan areas. Using extreme gradient boosting on data spanning January 2019 to June 2023, the study identifies a shift from internal to external factors as key drivers of ridership post-pandemic. Socioeconomic factors, gasoline prices, telecommuting, population density, employment density and polycentric development emerged as influential for bus ridership post-pandemic, while traditional factors like vehicle revenue miles, fare, transit coverage, and service areas are more important for rail ridership. Additionally, the study uncovers unique threshold and interaction effects in the post-pandemic period, including positive interactions between African American population proportions and poverty rates for bus ridership, carless households and gasoline prices for bus ridership, and between VRM and polycentricity for rail ridership. This dissertation provides insights into the complex dynamics between BE, transit types, and travel behavior, offering valuable implications for urban transportation planning and policy development at multiple levels.
Mission-driven Sensor Network Design for Space Domain Awareness
Harris, Cameron Douglas (Virginia Tech, 2024-12-09)
This research presents a novel framework for optimizing sensor networks to enhance Space Domain Awareness in the face of a burgeoning resident space object population. By employing advanced metaheuristic optimization techniques and high-fidelity modeling and simulation, this research investigates the intricate interplay between sensor characteristics, network topology, and state estimation performance. The research aims to develop actionable recommendations for optimizing sensor network design, considering factors such as viewing geometry, sensor phenomenology, and background noise. Through rigorous simulations and analysis, this work seeks to contribute significantly to the advancement of Space Domain Awareness. A key product of this research is the development of a novel lattice-based genetic algorithm tailored for constrained metaheuristic optimization that converges in 15% fewer generations than traditional methods. This algorithm demonstrates its effectiveness in producing practical sensor network designs that can enhance space object tracking and surveillance capabilities. Results will show network designs that fill current coverage gaps over the Atlantic and Pacific oceans, remain consistent with geographical and geopolitical boundaries, and exploit regions with favorable environmental conditions. The outcome is a set of actionable solutions that triple observation capacity and reduce catalog observation gap times by up to 50%.

Browse

Recent Submissions