Doctoral Dissertations
Permanent URI for this collection
Browse
Recent Submissions
- Impaired Ulk1 Ser555 Phosphorylation Promotes Amino Acid Reliance and Links Mitochondrial Inefficiency to Systemic Metabolic InflexibilityWilloughby, Orion (Virginia Tech, 2025-11-19)Metabolic flexibility—the capacity and ability to switch between energetic substrates in response to nutrient availability—is essential for systemic energy balance and protection against metabolic disease. Loss of this adaptive capacity contributes to metabolic diseases marked by obesity, insulin resistance, and dyslipidemia, elevating the risk of type 2 diabetes and cardiovascular disease. Under nutrient stress, skeletal muscle typically adjusts substrate utilization from glucose to fatty acid oxidation, while the liver supports peripheral energy needs through glucose production and auxiliary amino acid metabolism. In insulin-resistant states, this inter-organ coordination deteriorates, resulting in reduced glucose utilization and increased catabolism of glucogenic amino acids such as alanine—leading to inefficient energy use and metabolic inflexibility. Unc-51-like autophagy activating kinase 1 (Ulk1) is a serine/threonine kinase best known for initiating autophagy, but also serves as a nutrient-sensing hub regulated by AMPK and mTORC1. AMPK activates Ulk1 via phosphorylation at serine 555 (S555) during energetic stress, while mTORC1 inhibits Ulk1 under nutrient-rich conditions. To investigate the specific role of Ulk1(S555), we used global Ulk1(S555A) knock-in mice, which prevent phosphorylation at this site, and examined responses to short-term caloric restriction (CR) and fasting. Ulk1(S555A) mice exhibited impaired glucose oxidative capacity in skeletal muscle and liver during nutrient stress, despite improved systemic glucose clearance. Metabolic cage analyses of WT and Ulk1(S555A) showed no significant differences in heat and energy expenditure, despite Ulk1(S555A) delayed substrate switching and reduced physical activity. U-14C-glucose tracing confirmed reduced glucose oxidation, while multi-omics analyses revealed a tissue-specific dependence on amino acids in response to nutrient stress. Ulk1(S555A) skeletal muscle mitochondria displayed increased basal oxidation of alanine and serine with blunted ADP-stimulated respiration, while Ulk1(S555A) hepatic mitochondria demonstrated an impairment of pyruvate entry into the TCA cycle. These findings identify Ulk1(S555) as a critical integrator of AMPK signaling, regulating substrate prioritization and metabolic communication between skeletal muscle and liver. Targeting this axis may offer new therapeutic avenues to restore metabolic flexibility in obesity, type 2 diabetes, and related disorders.
- A Sociotechnical Network Analysis of Bluegrass Practitioners and Instruments: Competing Definitions, Motivating Authenticities, and Tradition-Respecting InnovationsCarroll, William Anthony (Virginia Tech, 2025-11-19)This study applies a sociotechnical network analysis based on Actor Network Theory (ANT) to the bluegrass musical community. Three definitional emphases are identified within bluegrass; these emphases structure and bound the network based on instrumental practices, historical progression, or the community. The study shows how bluegrass practices, such as festivals and jams, temporarily restructure the network, creating and maintaining connections among actors. Further, the study shows how authenticity, conceptualized either as personal expression or as adherence to tradition, motivates actors toward these definitional emphases and connects to larger philosophical and cultural themes beyond bluegrass. The project also applies this network analysis to the instruments of bluegrass. It examines the banjo's history before bluegrass, showing how actors attempted to disentangle the banjo from its previous networks and place it into new musical and cultural networks, making changes both to the material instrument and to its meanings, thereby making possible the position it assumed in bluegrass. The project concludes by examining the place of vintage instruments in the bluegrass network and the need for boundary-crossing instrument makers.
- Investigating the Actions of Interleukin-6 Family Members on Post-Hatching Blastocyst Development in CattleOliver, Mary Alington (Virginia Tech, 2025-11-19)The goal of this work was to determine if supplementing selective members of the interleukin 6 (IL6) cytokine family influences post-hatching in vitro produced (IVP) bovine blastocyst development. The first series of studies determined if adding human recombinant IL6 improved IVP blastocyst quality before and after cryopreservation. Adding IL6 at 25, 50, and 100 ng/ml increased inner cell mass (ICM) cell number but did not affect trophectoderm (TE) cell numbers. IL6 also increased the ICM:TE ratio before and after cryopreservation and ICM cell number after freezing. All IL6 doses reduced the number of cells undergoing apoptosis in the TE, but not in the ICM. In the second study, we found that both human recombinant IL6 and leukemia inhibitory factor (LIF) increases ICM cell number before cryopreservation. Supplementing interleukin 11 (IL11) had an intermediate effect between the controls and the IL6 and LIF treatments. Supplementing IL6 also increased the ICM:TE ratio. We also observed that supplementing IL11 before freezing increased hatching rate of frozen and thawed blastocysts, but no cryotolerance effects were observed in thawed blastocyst cell number or percentage of apoptotic cells within the blastocyst. A separate series of studies tested if supplementing IL6 and/or LIF altered the morphology of blastocysts during extended culture from d 7 to 12 post-fertilization. Specifically, cell allocations within the bovine embryonic disc (ED) were investigated. Supplementing IL6 from d 5-12 increased epiblast (EPI) cell number, reduced hypoblast (HYPO) presence, and increased the EPI:HYPO ratio in d 12 blastocysts. These IL6 effects could not be replicated when IL6 was provided prior to the blastocyst stage (d 5-7), but they could be replicated when IL6 was supplemented after blastocyst formation (d 7-12). Supplementing IL6 from d 7-12 also resulted in EPI cell numbers similar to IVP blastocysts that were exposed to the uterus from d 7-12. In all studies we observed that LIF has minimal effects on EPI or HYPO allocation. These results indicate that LIF has minimal function in the bovine blastocyst after d 12, but IL6 can benefit EPI development. In conclusion, supplementing IL6 and LIF before freezing increases ICM cell number, but only IL6 has cryoprotective effects on IVP bovine blastocysts. Supplementing IL6 during IVP and extended blastocyst culture improves EPI cell number, but this is at the effect of the HYPO. Overall, results from this work indicates that IL6 plays an important role in post-hatching blastocyst survival.
- All-atom and Coarse-grained Molecular Dynamics Modeling of Various Polymeric and Composite MaterialsHao, Xi (Virginia Tech, 2025-11-14)This dissertation leverages molecular dynamics (MD) simulations to reveal the structure-property relationship of various polymeric and composite materials, including polyacrylonitrile-poly(methyl methacrylate) (PAN-b-PMMA) copolymers, polyetherimide (PEI)-graphene composites, and epoxy network polymers, by bridging molecular-level behaviors and macroscopic thermomechanical properties. First, all-atom MD simulations are used to improve the design of porous carbon fibers derived from PAN-b-PMMA copolymers for energy storage applications. A new method is developed to characterize the interfacial area between different domains. Simulation results reveal a molecular mechanism underlying the experimental findings, demonstrating that the interfacial area -- a key predictor of the electrochemical performance of the resulting fibers after oxidation and carbonization -- reaches a maximal value when the two blocks are at a 50% volume fraction. This understanding paves the way for designing PCFs with optimal energy storage capabilities. Next, all-atom MD simulations are used to explore a strategy to enhance polymer nanocomposites by mitigating nanofiller aggregation. Experiments show that coating the surface of reduced graphene oxide (rGO) nanoparticles with PEI chains can improve their dispersion in a PEI matrix and thus lead to stronger composites. Simulations reveal that the PEI chains grafted to the edge surface a rGO particle form a protective layer of the particle, preventing particle aggregation and creating a more compatible interface with the host polymer. This enhanced compatibility makes the composites perform more strongly under mechanical loading, as seen experimentally. Polymer grafting is therefore confirmed as a powerful strategy for creating stronger, more reliable composite materials. Finally, a computationally efficient coarse-grained (CG) model is developed for epoxy resins based on EPON 862 (Diglycidyl Ether of Bisphenol F) monomers and diethyltoluenediamine (DETDA) curing agent, a critical component of high-performance composite materials. The CG model is transferable across a wide range of temperatures and is used to predict the mechanical properties of epoxy resins with reasonable accuracy. It provides a facile approach to creating large epoxy networks. Then via a backmapping procedure, the CG network is mapped to an all-atom network with the same topology. The all-atom and CG networks are used for understanding the fracture behavior of epoxy resins at experimentally relevant spatiotemporal scales. Collectively, this dissertation provides a suite of validated computational tools and fundamental molecular insights to advance the bottom-up design and optimization of next-generation polymeric and composite materials.
- Cross-Laminated Timber made of Unmodified and Thermally Modified Yellow-Poplar LumberMasoumi, Abasali (Virginia Tech, 2025-11-14)Cross-laminated timber (CLT), a sustainable construction material, is transforming the construction industry. To meet diverse demands, especially for exterior applications, its durability must be improved to withstand conditions ranging from dry indoor environments to moisture-prone settings. Thermally modified wood (TMW) is a potential solution to enhance CLT's moisture durability and reduce moisture-related issues. Yellow-poplar (YP), abundant in the Appalachian region, is a promising raw material due to its availability and favorable mechanical properties and incorporating its TMW into CLT outer layers offers a sustainable strategy to improve moisture resistance, reduce moisture-induced strain, and enhance durability for exterior applications. The goal of this work was to enhance the moisture durability of CLT by integrating TMW in its outer layers. This goal was addressed through four objectives: (1) determining the water vapor permeability and resistance factor of CLT made from unmodified wood and hybrid CLT with TMW outer layers, including the influence of the one-component polyurethane (PUR) adhesive layer; (2) predicting long-term moisture diffusion performance using hygrothermal simulations; (3) establishing correlations between moisture-induced strain and water vapor diffusion to provide input for physics-informed modeling; and (4) developing a predictive model for water vapor permeability and resistance factor (µ-value) that accounts for swelling strain and the adhesive layer. The first objective was achieved by investigating steady-state moisture diffusion (ASTM-E96) through CLT panels of various configurations, including 3-layer, 2-layer (to evaluate bondline effects), and 1-layer boards for both unmodified and hybrid CLT. Results showed that hybrid CLT exhibited significantly improved moisture resistance compared to unmodified CLT, with µ-values of 51.3 versus 32.7 for unmodified 3-layer CLT, and contributions from TMW (µ = 32.9) and PUR adhesive (µ = 1345), compared to 22.3 for unmodified wood. The second objective involved simulating long-term moisture diffusion using WUFI software. While simulations indicated no significant improvement in moisture resistance for hybrid CLT compared to control, likely due to WUFI's lower sensitivity at low µ-values, this confirmed the absence of moisture accumulation in the middle layer of hybrid CLT, consistent with experimental results. Simulations also evaluated various insulation options, identifying Polyisocyanurate board (R=72) as an effective single-layer interior insulation. The third objective focused on the correlation between moisture diffusion and moisture-induced strain. Dimensional changes were monitored in 1-layer boards, 3-layer and 5-layer CLT samples under controlled conditions, revealing strong relationships, such as R² = 0.99 between diffusion and volumetric strain in unmodified wood. Swelling strain, which reduces pore sizes in the microstructure, reduced diffusion by an average of 77%. Outdoor weathering over one year showed no delamination in the 5-layer hybrid CLT, while control YP CLT exhibited significant delamination on the surface and thickness; minor surface checks on TMW were attributed to lower elasticity. These results confirm that hybrid CLT with TMW outer layers offers superior long-term durability for outdoor applications. Finally, the fourth objective was accomplished by developing a physics-informed data-driven machine learning (PIDDML) model to capture the interplay between diffusion, adhesive layers, and moisture-induced strain. Traditional Fick's law models fail to account for these effects, resulting in large prediction errors. The PIDDML model, trained on 11 features including humidity, temperature, and moisture, integrated diffusion and strain data, explicitly accounting for swelling and adhesive effects. It outperformed Fick's law, achieving a maximum absolute error (MAE) of 1.4–13.1% for spruce-pine-fir (SPF) CLT and 11.2% for hybrid CLT compared to 46.7–66.0% under Fick's law, with an overall R² of 0.96. Using a logistic regression moisture content-permeability submodel, the framework was extended to other wood species, generating data for untested species and achieving cross-validation R² = 0.94. By embedding physical constraints such as Fick's law, mass conservation, and absorption–diffusion kinetics, the PIDDML model ensures physically plausible predictions, reduces reliance on extensive experimental data, and generalizes across diverse conditions and species. By integrating experimental measurements, hygrothermal simulations, and advanced PIDDML modeling, this study provides a robust framework for predicting CLT's moisture behavior. The hybrid CLT approach enhances dimensional stability and durability, advancing its use in moisture-exposed construction applications, contributing to sustainable building practices, and supporting wider adoption of CLT in North America.
- Understanding Cooperative Extension Directors' Conceptualizations of, and Perceived Roles in, InternationalizationGrove, Benjamin Bryant (Virginia Tech, 2025-11-14)The Cooperative Extension System has been engaged in internationalization for decades. Ludwig and Barrick (1996) outlined five indicators of an internationalized Extension system. In the broader higher education literature Knight (2003) has conceptualized internationalization as a process. Extension directors have been identified as key leaders of internationalization within the Extension system. This dissertation investigated directors' conceptualizations of internationalization and their roles in such pursuits. I conducted semi-structured interviews with directors from across the U.S. to ascertain their understanding of internationalization as a phenomenon. I used Knight's (1994) internationalization cycle as a primary analytic framework and also drew on additional arguments from the higher education and Cooperative Extension System literatures. The study participants revealed a lack of a shared definition of internationalization. They articulated a perceived imperative that domestic audiences receive the primary benefits of internationalization and financial considerations as prevailing lenses for considering engagement in such initiatives. They suggested process-based, position-based, rationale-based, and system leadership-based rationales for such efforts. They also acknowledged the mediating impact of university priorities and resources on whether their systems would become involved in internationalization.
- Structure-property relationships of earth and engineered materialsEhlers, Alix Marie (Virginia Tech, 2025-11-03)Structure-property relationships, which describe the connection between an atomic-scale structure and arising functional properties, inform our understanding of the physical world, from unraveling deep-earth dynamics to developing and tuning profitable materials. A comprehensive characterization of the structure of minerals and materials (from their atomic- to microstructure) is necessary for their full and informed implementation. This dissertation considers three overarching areas of research in which mineral and material structures are constrained and resultant large-scale consequences are detailed. First, the properties inherent in atmospheric mineral particles and their consequences on aerospace-grade material are investigated. The mineralogy and particle-based characteristics of test dusts are comprehensively described using a detailed mineralogical characterization workflow. The morphologies of these particles combined with particle-target experiments (conducted for different permutations of particle impact speed, angle of incidence, and target material type) reveal that erosion of targets from impacts of test dust particles is driven by normal particle impact velocity and target yield strength. These results were implemented into a particle bounce model in a companion paper which models a particle's change in kinetic energy following impacts. Second, the high-pressure crystallographic properties were investigated for ternary oxides (ABO4 compositional space). High-pressure experiments on the rare-earth phosphate (REEPO4) group show that whole-structure compressibility is driven by the compressibility of REEOx polyhedra. Moreover, we demonstrate a linear relationship between the REE ionic radius and REEPO4 compressibility, which is consistent through the I41/amd to P21/n phase transition. We also combine high pressure and high temperature data for the mineral zircon, which demonstrates entrapment conditions of zircon inclusions in garnet hosts. Third, the dynamical properties of the entropy-stabilized oxide Mg0.2Co0.2Ni0.2Cu0.2Zn0.2O, which are instrumental to its valuable thermal properties, are described using inelastic neutron scattering experiments combined with complementary VASP simulations. This work shows that energy contributions at room temperatures and above are driven by Mg and O ions. Calculations of thermal properties from VASP simulations reveal that phonon-driven entropy contributes a significant amount to total system entropy. In combination, this work contributes to three different fields of scientific research and uncovers how valuable, desired, or complex properties of earth and engineering materials are driven by inherent structural characteristics.
- Warriors and Healers of the Eastern Band of Cherokee: Peoplehood, Survivance, and Military Service in World War I (1917-1924)Chambraud, Marie lys Therese Jeanne (Virginia Tech, 2025-11-03)My research examines the contributions of the Eastern Band of Cherokee warriors during World War I, emphasizing Indigenous agency, cultural survival, and community within a settler-colonial framework. It explores how these warriors navigated military service as both an assertion of sovereignty and a means of survival, using their enlistment to resist colonial erasure while maintaining their cultural identity. Through the application of the Peoplehood model and the concept of survivance, this study analyzes the interconnected roles of land, language, sacred history, and ceremonial practices in sustaining Cherokee identity. Furthermore, this study incorporates an Indigenous feminist perspective to emphasize the role of Eastern Cherokee women in maintaining the resilience of Cherokee communities and supporting the healing of warriors. The women served during the war as nurses, officers and occupied a generally white and male centered environment. After the war, some former warriors and women participated in political actions to improve the lives of indigenous peoples. They also had to learn to ret ingrate themselves in their community and society after such deep traumas; and for that they used their traditional healing practices and traditional stories. By combining archival research, oral histories, and an analysis of military service, my work contributes to ongoing discussions of Native sovereignty, military service, and the legacies of Indigenous resilience before, during, and after World War I. It underscores the enduring importance of Indigenous storytelling, archives, and collective memory in documenting the survival and revitalization of Cherokee culture in the face of historical violence and cultural suppression.
- Experimental and Numerical Explorations of Fire Performance of Intumescent Thermoplastic Composites for Electric Vehicle Battery EnclosuresFarhadi, Mehrnoush (Virginia Tech, 2025-10-28)The global shift toward renewable energy has accelerated the adoption of electric vehicles (EVs) powered by lithium-ion (Li-ion) batteries. While these batteries offer high energy density and efficiency, their inherent fire risk remains a critical safety concern. To mitigate this hazard, plastic-intensive, flame-retardant enclosures have been engineered as promising alternatives to traditional metal-intensive systems for EV battery pack protection. This dissertation presents a first-of-its-kind experimental and numerical investigation into the fire performance of three newly developed injection-molded, flame-retarded polypropylene (FR-PP) thermoplastic composites, each reinforced with 30% discontinuous glass fibers with differing fiber length and FR content. These systems are candidates for next-generation plastic EV battery enclosures. Understanding the fire performance of Li-ion battery systems under fire conditions has traditionally relied on full-scale pool-fire testing, as prescribed by standards such as GB/T 38031-2020 and ECE R100. While indispensable for certification and regulatory compliance, full-scale fire tests are costly, logistically challenging, and carry significant safety risks, particularly due to potential battery explosions. To navigate these challenges, this study adopts a more agile, modular approach. Instead of testing the entire system, a representative smaller-scale experimental approach was implemented to replicate pool-fire thermomechanical loading on horizontally oriented plate-scale composite specimens in a safer and more controlled manner. One-sided fire exposure generated using a sand burner reproduced pool-fire-like conditions, while steel cylinders were configured to represent mechanical loads of Li-ion battery weight and compartments. This scaled-down framework removes hazards while enabling precise, repeatable experiments and allows examination of the influences of material formulation (fiber length and level of flame-retardant content), geometric parameters (plate thickness), fire intensity (fire source size), and mechanical loading (flexural load configuration) on the fire performance of composite specimens. Through these experiments, the study identified configurations that achieved the greatest fire endurance, with the top-performing candidate selected as a benchmark for computational validation. The key characterizations of the intumescent, decomposing thermoplastic systems are thickness expansion and viscoelasticity. The evolution of thickness expansion and softening behavior across the full thermal spectrum and material states, from intact nondecomposed regions at room temperature, through condensed phases at decomposition onset, to porous layers infused with gases during decomposition, and finally to fully degraded charred systems, was captured through comprehensive property characterizations and material measurements. Advanced constitutive models derived from these characterizations were incorporated into the computational framework. The primary contribution is the development of a coupled thermomechanical, physicsbased, and experimentally driven finite element framework that simulates the composite plates under replicated pool-fire loading and the weight of the battery and its compartments. This framework is validated via fire experimental testing while integrating the key characterizations of the intumescent, decomposing thermoplastic composite plate, as well as all composite plate–steel cylinder and system-environment interactions. The integrated experimental-computational platform reliably predicts temperature distributions and out-of-plane deflections under combined thermal and flexural stresses, providing a scalable foundation for plate-to-module and full-scale simulations. Collectively, these results establish a robust foundation for the safe and effective design of composite battery enclosures under extreme fire scenarios, with computational predictions validated against observed experimental behavior.
- The Role of Charge Transfer Induced Spin Crossover Complexes on Charge-Separated State Lifetime in Photoelectrochemical DevicesCheng, Tzu-Ching (Virginia Tech, 2025-10-28)In sustainable fuel production, the conversion of solar energy into chemical energy is one of the great interests of storing high-density energy. The relatively low observed quantum yields (~19%) for such process compared to theoretical efficiencies (in the range of 30-40%, depending on target reaction) result in part from the prevalent short-lived photo-induced charge-separated states. In our previous studies, manganese (II/III) poly(pyrazolyl)borate (Mn(pzb)2) complexes demonstrated exceptional charge-separated-state lifetimes in the dye-sensitized photoanode constructs. Mn(pzb)2 undergoes a spin transition from a high-spin, sextet state for Mn(II), to a low-spin, triplet state for Mn(III). The spin change is induced upon a pseudo-octahedral-to-octahedral structural change that modifies the orbital overlap between the pyrazolyl lone pair and central Mn. The large reorganization energy associated with the structural and spin transitions increased the lifetime of charge-separated states. Inspired from the results, we sought to determine the degree of molecular reorganization/spin transition that results in the greatest modulation in back electron transfer rates. Two methods were explored 1) the use of zwitterionic ligands, 2) the formation of a multi-component molecularly sensitized interface with [M(pzb)2] (M= Mn, Fe, and Co). In the first approach, the Lewis base of the zwitterion ligand affects redox potential and the magnetic properties. By tuning the pKa of the Lewis base and the functional group on pyrazole, the extent of the spin transition can be modified. Moreover, the zwitterion ligand is charge-neutral and a counterion is needed to balance the Mn positive charge. Thus, the overall charge neutral zwitterionic [Mn(pzb)2] complex has improved its solubility in organic solvents, a downfall of our first public [M(pzb)2] complex derivation in quantum dot solar cell. The second approach is to decorate the [M(pzb)2] (M= Mn, Fe and Co) with phosphoric acid group (-PO3H2) to form a self-assembly layer with a cognizant chromophore. [M(pzb)2] (M= Mn, Fe and Co) exhibit charge transfer induced spin crossover (CTISC) but the change of spin multiplicity is different depending on the central metal. The spin multiplicity difference were hypothesized to manipulate the extent spin transition/reorganization. Furthermore, the self-assembled layer forms the molecular layer in the photoanode construct, which bypasses the aforementioned solubility issue.
- Fast and Accurate Graph ClusteringWanye, Frank Derry (Virginia Tech, 2025-10-28)Graph clustering, also known as community detection, is a fundamental problem in graph analytics with applications across a wide variety of domains including bioinformatics, social media analysis, and anomaly detection. Graph clustering algorithms can be grouped into two categories: inferential and descriptive. While inferential algorithms deliver high accuracy, they do so with poor performance and poor scalability, particularly on large graphs. Conversely, descriptive algorithms are fast and scalable but significantly less accurate. In this manuscript, we focus on accelerating the performance and improving the scalability of stochastic block partitioning (SBP), a bottom-up (agglomerative) inferential graph-clustering algorithm based on sequential Bayesian inference. Like other inferential algorithms, SBP is highly accurate even on graphs with a complex community structure but does not scale well to large real-world graphs that can contain upwards of a million vertices. Our first set of contributions centers around performing data reduction via sampling as well as shared- and distributed-memory parallelization for accelerating the bottom-up formulation of SBP. We integrate these approaches into a unified, accelerated bottom-up SBP implementation, which exhibits up to 322X speedup over sequential SBP on a 1,000,000-vertex graph processed on 64 compute nodes with 256GB of memory. Next, we fundamentally re-factor SBP's bottomup approach into a top-down one and modify the previous data reduction and parallelization methodology to realize an integrated and accelerated top-down SBP algorithm. This new algorithm demonstrates up to 403X speedup over sequential SBP on real-world graphs when run on 64 compute nodes with 48 cores and 32GB of memory. Furthermore, the top-down computational approach enables SBP to process up to 4.4X larger graphs — from 862,664 vertices to 3,774,768 vertices — on the same 32GB of memory. Importantly, our approach accelerates SBP without sacrificing the algorithm's accuracy on both real-world and synthetic graph datasets.
- Human Pose Estimation and Algorithms for Alignment and Registration Problems: Applications in Robotics, Computer Vision, and Stroke RehabilitationSarker, Anik (Virginia Tech, 2025-10-27)Estimating human motion and inter-frame orientation reliably, efficiently, and with minimal instrumentation is central to robotics, computer vision, and rehabilitation. This dissertation develops a set of learning- and geometry-driven methods spanning (i) sparse-IMU upper-body 3D pose estimation with deep sequence models, and (ii) fast, correspondence-free alignment on the sphere (S2) and on the rotation group (SO(3)) with applications to sensor calibration and robot world–hand–eye (RWHE) problems. Chapter 2: Sparse-IMU upper-body pose estimation with deep sequence models. We study the problem of reconstructing upper-body kinematics from three wearable IMUs (sternum and bilateral forearms) for stroke rehabilitation. Using a synchronized XSens MVN system as reference, we introduce a cross-sensor mapping that transfers standalone XSens DOT measurements into the MVN coordinate system. Two mapping regimes are investi- gated: a variable (session-specific) mapping and a fixed mapping obtained by quaternion averaging across sessions. On top of these mapped signals, we train and evaluate a family of deep sequence models—Seq2Seq, Seq2Seq with BiRNN and attention, a Transformer encoder, and a full Transformer—to infer multi-joint upper-body orientations (15 segments) from the three-IMU streams. Transformers, particularly the full encoder–decoder variant, achieve state-of-the-art accuracy and robustness across participants and tasks, outperforming re- current baselines while maintaining deployment-friendly throughput. The results quantify trade-offs between model class, IMU placement (two configurations), and mapping regime, and demonstrate that a fixed mapping retains accuracy while enabling practical re-use across recording sessions. Chapters 3–4: Fast, correspondence-free alignment on S2 and SO(3). We recast set alignment of orientations as alignment of Transformed Basis Vectors (TBVs): each rotation yields three unit vectors on S2, producing three spherical point clouds per dataset. Build- ing on this representation, we develop (and rigorously analyze) three spherical matchers— SPMC (Spherical Pattern Matching by Correlation), FRS (Fast Rotation Search), and a hybrid SPMC+FRS—that estimate relative rotation in linear time O(n) and without point- wise correspondences. These methods are robust under severe contamination (empirically up to 90% outliers) and avoid the cubic-log scaling of FFT-based spherical/SO(3) cross- correlation. We lift these spherical matchers to SO(3) by aligning TBV triplets per axis and projecting the fused estimate back to the group, yielding SO3_SPMC, SO3_FRS, and SO3_SPMC_FRS for axis-consistent settings (e.g., homogeneous IMUs from the same vendor). To handle axis-ambiguous scenarios (cross-vendor frames, world/hand–eye), we make the pipeline Permutation-and-Sign Invariant by enumerating signed permutations L = P S of axes and selecting the maximizer over 24 proper hypotheses (det L = +1). This preserves the O(n) profile: the heavy spherical matches (18 axis/sign pairs) are computed once and re-used during hypothesis scoring. We demonstrate PASI-SO(3) alignment as a drop-in rotational initializer (or stand-alone estimator) for RWHE calibration on real data (ETH robot_arm_real), using raw, unpaired trajectories without time alignment. Chapter 5: Motion-driven, axis-consistent automatic orientation calibration for wearable IMUs. We target session- and subject-specific orientation drift that arises in wearable IMUs due to don/doff variability and strap placement, even when devices share the same vendor axis convention. The key idea is to calibrate from motion: we maintain a library of ground-truth SO(3) motion signatures (walking, sit-to-stand, reach, etc.) collected previously in a canonical frame; at the start of a new session, the user performs a brief activity sequence, producing an unpaired SO(3) set. Assuming axis consistency, we apply SO3_SPMC to align the session's TBV distributions to the canonical signatures and re- cover the session's global rotational offset in a single, correspondence-free step. We show: (i) calibration is accurate and repeatable across participants and days; (ii) multi-activity signatures increase identifiability and reduce bias; (iii) runtime is linear in sequence length (tens of milliseconds for typical windows); and (iv) the calibrated orientations improve down- stream pose estimation and cross-session comparability without magnetometer reliance or time alignment. Empirical validation and impact. Across synthetic and real evaluations, the proposed methods deliver: (i) accurate upper-body pose reconstruction from minimal IMU instru- mentation with Transformer models and a reusable cross-sensor mapping; (ii) fast, robust S2/SO(3) alignment that scales linearly and tolerates extreme outliers; and (iii) a practical, motion-driven, axis-consistent calibration procedure for wearable IMUs that removes session/subject orientation bias from brief activity snippets. Together, these contributions pro- vide a cohesive toolkit—learning for motion inference and geometry for fast, correspondence-free alignment—that enables calibration-light monitoring and cross-device fusion in wearable sensing, robotics, and vision.
- Integrated approaches for monitoring sharks: Leveraging machine learning, big data, and molecular biologyJenrette, Jeremy Freeman (Virginia Tech, 2025-10-24)Sharks are ecologically important predators facing severe global declines, yet conservation and management are hindered by data deficiencies in taxonomy, distribution, and abundance. In this dissertation, I develop and integrate complementary technological approaches: machine learning, big data workflows, and molecular techniques—to expand scalable, non-invasive monitoring of sharks with programmatic and practical field methodologies. First, I constructed the largest global shark image dataset to date and developed the Shark Detector, a pipeline combining object detection and hierarchical classification. This system automatically locates, identifies, and classifies sharks in heterogeneous media, achieving >90% recall for detection and up to 92% species-level classification accuracy across 80 species, outperforming existing biodiversity classifiers. Second, we refined these methods for ecological survey applications by packaging the models into sharkDetectoR (R package) and SharkByte (desktop application), enabling accessible, semi-automatic processing of baited remote underwater videos (BRUVs). These tools reduced annotation effort by up to 95% while preserving high taxonomic resolution, and demonstrated iterative improvement through survey-specific data boosting. Third, I designed scalable pipelines to mine and filter >5 million social network (Instagram, Flickr) and open source (iNaturalist and Global Biodiversity Information Facility) posts and >600k opportunistic shark observations. By pairing automated classification with effort-standardized statistical models, we derived species-specific abundance indices that revealed regionally consistent population trends: increasing trajectories for coastal taxa in the Bahamas, and recent declines of reef-associated sharks in the Hawaiian Islands. Finally, I piloted molecular monitoring of critically endangered white sharks (Carcharodon carcharias) in the Mediterranean Sea using complimentary Environmental DNA detection and validation workflows. I collected 204 samples across the Sicilian Channel, Adriatic and Ligurian Seas, and detected white sharks at four stations. Detections were confirmed in the lab. Particle simulations identified the detected individuals as nearby for the purpose of tracking them in the field. A preliminary multi-species assay detected 12 elasmobranch species. These workflows provided novel spatiotemporal insights into white shark (and other elasmobranch) occurrence in hypothesized hotspots. Together, these chapters demonstrate how integrated computational and molecular approaches can overcome data limitations, provide reproducible ecological indices, and inform conservation of threatened shark populations in data-poor regions.
- Data Science for Multisectoral Water Use Management and STEM Education: A Synergistic ApproachNaseri, Mohammad Yunus (Virginia Tech, 2025-10-23)The unprecedented data deluge across disciplines offers transformative potential for addressing 21st century challenges, including sustainable water resources management amid climate change and population growth and transforming higher education. Data science provides a comprehensive framework for extracting actionable insights from this wealth of data, enabling evidence-based decision-making across complex systems at scale. However, the application of data science for sustainable water resources management across economic sectors in the United States (US) has been constrained by the lack of comprehensive data at granular spatiotemporal scales. Furthermore, data science applications for analyzing residential water use dynamics using large-scale, fine-resolution smart water meter data remain largely unexplored. Yet a more fundamental barrier to widespread data science adoption across disciplines is the insufficient integration of authentic data and data-driven modeling experiences for real-world problem-solving within undergraduate STEM curricula, limiting students' preparation for an increasingly data-driven workforce. This dissertation addresses these interconnected gaps by conducting synergistic data-driven and mixed-methods research at the intersection of civil engineering and engineering education. Using an inductive approach, the research first demonstrates specific cases of data science applications for water resources management. Then, it synthesizes lessons learned from these disciplinary cases along with instructor and student data from a multi-university and multi-disciplinary data science integration experiment to derive generalizable principles and assessment frameworks for integrating data science across undergraduate STEM curricula, advancing both water resources management and data science education in STEM disciplines. To address the fundamental data limitations constraining the application of data science methods to water resources management, this dissertation first developed the United States Water Withdrawals Database (USWWD), a comprehensive standardized compilation of user-level water withdrawal data across 42 states in the US. Through systematic collection and integration of heterogeneous state-level data sources, USWWD provides water withdrawal time series at unprecedented spatial and temporal resolutions, encompassing 188,597 unique water users, 353,082 points of diversion and use, and 57,559,412 withdrawal volumes across multiple economic sectors. The database standardizes diverse information on water users, withdrawal locations, volumes, source types, and primary water use categories, combining both direct measurements and various estimation techniques to reflect the diverse reporting methods utilized by different state agencies. By providing the most detailed national water use data to date at disaggregated spatiotemporal scales, USWWD enables comprehensive data science applications for understanding multisectoral water withdrawal patterns, trends, and drivers, directly supporting evidence-based water resource management, planning, and policy development across the US. The dissertation next focused on residential water use sector, applying data science methods to analyze residential water consumption patterns at both city and household scales using high-resolution smart water meter data. It used an unprecedented dataset of residential water use from 33,435 single-family households across 39 US cities over two winter months. At the city level, it used functional data analysis and mixed-effects random forest that revealed distinct consumption clusters, with 13 high and 6 low water-using cities (concentrated in coastal California) differing significantly from 20 medium water-using cities, where shower and toilet end uses emerged as primary drivers of water use. Extending this analysis to the household scale to assess the relative effects of behavioral versus fixture efficiency factors on total daily water use, the dissertation revealed that while behavioral factors explain most variation in total per capita indoor water use, fixture efficiency factors better differentiate between high and low water-using households, particularly around shower, toilet, and clothes washer end uses, with significant economies of scale observed as household size increases. These multi-scale findings provide critical insights for targeted urban water management strategies that combine fixture efficiency improvements with behavioral interventions, emphasizing the importance of scale-appropriate conservation approaches for different household categories and geographic contexts. Following the demonstrated applications in water resources management, this dissertation derives the principles for integrating data science into established undergraduate curricula through a multi-university research-practice partnership. Working with instructors across six courses at three universities and input from industry partners, the research documented how educators can effectively integrate discipline-specific data science modules into existing science and engineering courses, with instructors selecting discipline-agnostic topics such as data visualization and statistical analysis while adapting integration approaches to meet specific course needs, academic levels, and pedagogical requirements. Assessment of this integration approach through mixed-methods analysis of 877 student data across diverse demographics, academic levels, and disciplines revealed significant increases in students' self-reported motivation, skills, interest, and confidence in data science, with strong alignment between student self-assessments and instructor evaluations indicating effectiveness from both perspectives. The development of 12 publicly accessible data science modules across six disciplinary science and engineering fields, combined with empirical evidence of their educational impact, provides a transferable framework for preparing STEM graduates with essential data science competencies needed for an increasingly data-driven workforce, thereby addressing the fundamental educational barrier to widespread data science adoption across disciplines.
- Evaluation of Feed Ingredients and Feed Additives on Poultry Performance and HealthLyons, Alyssa Mae (Virginia Tech, 2025-10-23)The grinding of feed ingredients is a major contributor to the costs of poultry production and there is no optimal ingredient particle size for corn or calcium (Ca). Optimization of ingredient particle size can alleviate costs, improve growth and production, and promote sustainability. Intestinal health is also a key factor of poultry production, but with the limitations of conventional strategies, alternative methods to promote health and performance are necessary. This dissertation investigates corn particle size, alternative feed ingredients (aragonite), and feed additives (direct fed microbials) on their potential to improve turkey and laying hen health and performance. The second chapter is a literature review on ingredient particle size, Ca source, and varying feed additives. The third chapter determined that a smaller corn particle size (581 µm) was necessary for turkeys during the starter 1 phase, but a larger corn particle size (964 µm) is able to maintain performance later on. Turkey poults that consumed the 581 µm corn had an increase in BW (P ≤ 0.01) over the starter 1 phase (0-21d) compared to those consuming the 964 µm corn. By 42 d of age, there were no differences in BW, FI, or FCR indicating that larger corn particle sizes can be used later on to maintain performance and potentially reduce feed costs associated with grinding of ingredients. The fourth chapter evaluated varying Ca sources (blend of 50:50 fine and coarse limestone and fine aragonite) and concentrations (2.46, 3.28, or 4.10%) on laying hen performance, eggshell quality, and bone mineralization. There were minimal differences in performance regardless of Ca source and concentration, however the inclusion of 4.10% limestone resulted in the lowest performance. There was a Ca source main effect for all eggshell quality parameters including increased breaking force, shell thickness, relative shell weight, and specific gravity for birds fed the blend of fine and coarse limestone compared to fine aragonite (P < 0.05). There was a linear increase in bone mineralization for birds fed limestone (P = 0.02) whereas there was no difference for those fed fine aragonite (P = 0.70). These data indicate that fine aragonite was able to maintain production and bone mineralization, but birds fed lower levels of limestone started to pull Ca from the bone to support egg production. Fine aragonite may be used as a higher bioavailable Ca source and less Ca can be added into the diet to reduce costs. A combination of fine and coarse calcium is needed to support eggshell quality. Direct fed microbial (DFM) supplementation (Novela ECL® (ECL), Novela® (NOV), and Amnil® (AMN)) was investigated in the fifth chapter and its effects on laying hen performance, egg quality, and energy metabolism. Supplementation of all DFM altered the energy metabolism within the hen and allowed the birds to partition energy either towards egg production energy and storage. Egg weight was highest in the ECL fed birds followed by AMN then NOV and the control (P ≤ 0.01). An increase in egg weight resulted in a higher egg mass for ECL and AMN (P ≤ 0.01) which improved the feed conversion ratio by 7 and 9 points, respectively (P ≤ 0.01). Inclusion of NOV increased the body weight and stored energy within the hen (P = 0.05) whereas ECL and AMN diverted energy down a productive pathway and improved productive performance. Inclusion of both NOV and AMN increased egg breaking force compared to other treatments (P ≤ 0.01). The sixth chapter investigated the effects of a larger corn particle size on feed milling efficiency and the amelioration of a coccidial challenge in turkey poults using performance, intestinal permeability, nutrient digestibility, and litter moisture. The coarse corn (1,049 µm) had a higher mill load than the fine corn (597 µm; P < 0.01) but pellet quality was not different (P > 0.05). No interactions occurred for any of the measured parameters between corn particle size and coccidiosis challenge from 21 to 42 d. Inclusion of the coarse corn reduced FI and BW (P ≤ 0.01). Coccidiosis reduced all performance parameters and apparent ileal crude fat digestibility compared to the non-challenged birds (P ≤ 0.01). Litter moisture increased on D35 (d 7 post coccidiosis vaccination) for challenged birds (P = 0.02). A coccidiosis challenge reduced performance in turkey poults from 21 to 42 d and large particle corn was not able to overcome the coccidia challenge. This dissertation conveys possible ways to maximize turkey and laying hen production parameters through feed additives and optimization of ingredient particle size. These experiments provide insight on potential strategies to lower feed costs and varying feed additives to improve or maintain turkey and laying hen production and performance.
- Spatial Orientation Training in Virtual Reality: Designing for Cognitive Load and Spatial AbilityHsing, Hsiang-Wen (Virginia Tech, 2025-10-23)This dissertation advanced the understanding of spatial orientation in VR training by applying cognitive load theory in connection to visuospatial working memory across three related studies. Study 1 used eye-tracking during two spatial tasks (object-based rotation via Purdue Spatial Visualization Test: Rotations and perspective-change via Santa Barbara Solids Test) to extend gaze-based measures across transformation types. Gaze metrics (encoding, transformation, confirmation, strategy ratio) revealed a large difference between the two spatial reasoning tasks in gaze behaviors, with encoding fixations emerging as the most sensitive and associated with accuracy. Study 2 is a qualitative investigation of a VR basketball tactic training prototype, revealing user design inputs on the balance between realism and simplicity, and emphasis on role clarity. These insights also informed areas of interest for eye-tracking investigation in subsequent research. Study 3 experimentally manipulated graphical fidelity (low vs. high) and user perspective (egocentric, exocentric defense-up, exocentric courtside) in a recognition task on basketball tactics. Higher graphical fidelity increased subjective cognitive load, while user perspectives in training affected spatial processing (more fixation transitions between areas of interest for training in exocentric views) and recognition time (slower recognition for training in egocentric trials). Spatial ability moderated these effects, particularly when training in egocentric view. Together, the studies showed how graphical fidelity and user perspective differentially influenced cognitive load, gaze behavior, and performance, and how spatial ability conditioned these relationships. The dissertation concluded with actionable guidelines: personalize graphical fidelity, teach encoding explicitly, and tailor designs to learners' spatial abilities.
- Heterogeneous Decision-Making in Socio-Technical Systems Methodologies for Efficiency Measurement and Decision Analysis with Mixed DataMohsenirad, Saman (Virginia Tech, 2025-10-23)This dissertation advances the performance evaluation of complex socio-technical systems (STSs) by integrating systems theory with methodological innovations in Data Envelopment Analysis (DEA). Traditional DEA models, while robust in assessing relative efficiency across multiple inputs and outputs, implicitly assume homogeneity among decision-making units (DMUs) and precise, ratio-scaled data. These assumptions fall short when confronted with the realities of STSs, which are characterized by behavioral heterogeneity, contextual variability, mixed-scale and imprecise data, and emergent performance shaped by environmental conditions. Recognizing these challenges, this research reconceptualizes DEA within a systems-theoretic framework and proposes four interrelated methodological advancements that address the epistemological and practical limitations of conventional performance analysis. The first contribution introduces a novel statistical framework for testing heterogeneity among DMUs using slack-based diagnostics and nonparametric inference. This approach empirically verifies whether observed units operate under a shared technology, providing a rigorous foundation for clustering, meta-frontier analysis, and group-specific benchmarking. The second contribution develops a multivariate fuzzy DEA methodology that enables the inclusion of ordinal, categorical, and imprecise data through the integration of Multiple Factor Analysis (MFA), fuzzy set theory, and clustering algorithms. This model enhances interpretability and robustness in efficiency evaluation under data ambiguity and contextual complexity. Third, the dissertation proposes a hybrid SEM–DEA framework to estimate and adjust for environmental influences on performance. Structural Equation Modeling (SEM) is used to quantify both direct and indirect effects of contextual variables on inputs and outputs, allowing for what-if scenario modeling and fairer cross-unit comparisons. The fourth and final contribution incorporates a behavioral modeling dimension by applying machine learning techniques—specifically Random Forests and decision tree analysis—to predict household evacuation behavior in response to Hurricane Irma. It highlights adaptive decision-making under uncertainty and the value of linking behavioral insights with performance analysis. Empirical demonstrations focus on disaster evacuation as a prototypical socio-technical system, where decision-making interacts with infrastructure, resource constraints, and diverse perceptual frameworks. Across all contributions, this research maintains a commitment to interpretability, diagnostic clarity, and systems alignment. By situating DEA within a broader systems paradigm and extending its methodological repertoire, this dissertation offers a unified analytical lens to capture efficiency, contextual sensitivity, and behavioral realism. The resulting framework provides both theoretical insight and practical tools for evaluating performance in complex, human-centered environments—advancing the applicability of efficiency analysis in engineering, policy, and socio-technical design.
- A study of open education in the state of Virginia, 1973-76Norman, James Samuel (Virginia Polytechnic Institute and State University, 1977)
- Determination of high school student attitudes toward marketing and advanced marketing courses in VirginiaClodfelter, G. Richard (Virginia Polytechnic Institute and State University, 1984)
- The performance estimation of an axial-flow compressor stage using theoretically derived blade element characteristics with experimental comparisonJones, Ralph Raymond (Virginia Polytechnic Institute and State University, 1979)