Doctoral Dissertations

Permanent URI for this collection

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

Browse

Now showing 1 - 20 of 18261

Rhetorical Authority in Context: A Study of Technical Accounts of Flint, Michigan's Water Crisis from 2014-2017
Hockman, Cassandra Rose (Virginia Tech, 2026-03-30)
Essays on Modeling Human Behavior During Epidemics: Simulation, Statistical, and Optimization Approaches
Babaei Shalmani, Kian (Virginia Tech, 2026-03-30)
Human behavior is at the core of epidemics. Public risk perception shapes compliance with non- pharmaceutical interventions, mobility and contact patterns, and vaccine uptake; in turn, these behaviors alter transmission dynamics and future perceptions. A central challenge in integrating behavior into epidemiological analysis is that perception and response are not instantaneous. Information diffuses through societies with delays, and behavioral adjustment often occurs gradually and asymmetrically responding differently when risk is rising than when it is falling. Ignoring these delay structures can bias empirical inference about behavioral responsiveness and can misstate the effects of policies evaluated using models that treat behavior as exogenous or contemporaneous. This dissertation advances the modeling and estimation of behavioral feedback in epidemics by focusing on how delayed risk perception links epidemic indicators to behavioral change and policy outcomes. The first essay develops and validates a delay-aware empirical framework for estimating how mobility responds to epidemic risk. Using synthetic experiments, it shows that assuming immediate response (or relying on ad hoc fixed lags) can yield biased estimates of both the magnitude and timing of behavioral response. The essay introduces a structured approach to representing perception delays using distributed-lag formulations motivated by information diffusion and provides practical methods for estimating delay parameters alongside behavioral sensitivity. The second essay extends the framework by allowing delay structures to be asymmetric across phases of the epidemic, recognizing that behavioral responses to increasing risk may differ from responses to declining risk. Through additional synthetic tests and application to U.S. state-level COVID-19 mobility data, the essay demonstrates that the assumed delay structure materially affects inference about responsiveness and can change conclusions about how quickly behavior adjusts to worsening versus improving conditions. The third essay connects behavioral estimation to policy design by examining optimal vaccination strategies under endogenous, delayed behavioral feedback. It compares a conventional SEIRV framework with constant contact rates to a behavioral SEIRbV framework in which perceived risk reduces contacts with a perception delay. In both a homogeneous setting and an age-stratified allocation setting, the analysis shows that accounting for behavioral feedback can shift suppression thresholds and the relative performance of vaccination strategies, highlighting the marginal importance of operational levers such as earlier starts and faster rollout alongside prioritization rules. Taken together, the three essays show that delays in risk perception are a first-order feature of epidemic systems. By providing methods to estimate delay-aware behavioral responses and demonstrating how behavioral feedback reshapes vaccination policy evaluation, this dissertation contributes tools and evidence to improve inference, forecasting, and the design of effective interventions in epidemic settings.
Advanced Grid-Interface Three-Phase Converters: Grid-Support Capabilities and Grid Impact Analysis
Wang, Biqi (Virginia Tech, 2026-03-30)
The increasing penetration of power electronic converter–interfaced resources is fundamentally transforming modern power systems by displacing conventional synchronous generators, thereby introducing new challenges to system reliability and stability. To address existing challenges and technical gaps, this dissertation develops advanced grid-interface three-phase converters with grid-support capabilities and systematically assesses their impacts on power system operation, including dynamic behavior, small-signal stability, fault current characteristics, protective relay performance, and black-start restoration. Control frameworks are developed for both grid-forming inverters (GFIs) and newly proposed grid-supporting rectifiers (GSRs), enabling grid-regulation support and enhancing overall system stability. Impedance-based small-signal assessments incorporating the generalized Nyquist criterion (GNC) are conducted, with emphasis on (i) comparative stability analysis between conventional grid-tracking rectifiers (GTRs) and the proposed GSR under weak-grid conditions (Chapter 3), and (ii) stability of grid–GFI interconnections during black-start operations, particularly during restoration initial stages (Chapter 5). The proposed converters exhibit more benign impedance characteristics and enhanced stability performance, therefore reducing adverse dynamic interactions and enabling stable operation under weaker grid conditions. In addition, a GFI control strategy with improved fault ride-through capability is proposed. The impacts of inverter-interfaced distributed energy resources on grid fault current characteristics and protective relay performance are analytically investigated, focusing on potential issues of desensitization effect and selectivity deterioration (Chapter 4). Furthermore, a black-start-friendly GFI control is developed, and the feasibility of black-start operations supported by GFI-based renewable resources is systematically studied (Chapter 5). The effectiveness of proposed grid-support control strategies the and the accuracy of the grid-impact analyses are validated by comprehensive simulation studies and hardware experimental results. Overall, this dissertation provides a comprehensive control, modeling, and assessment framework for grid-interface converters in future converter-dominated power systems. The proposed GFI and GSR control strategies, together with impedance-based stability and protection analyses, contribute practical insights for the improving the stability, protection reliability, and restoration capability of low-carbon power grids.
Do Clinicians Have Patience? Examining Delay Discounting, Perceived Stress, and Low-Value Antibiotic Prescribing
King, Mary Jane (Virginia Tech, 2026-03-30)
Antimicrobial resistance has been deemed one of the top global threats by the World Health Organization, and the overuse of antibiotic medications contributes to antimicrobial resistance. Yet low-value antibiotic prescribing (LVAP; the prescription of antibiotics when not clinically indicated) is still widely prevalent across multiple healthcare disciplines, often for viral illnesses such as acute bronchitis. Behavioral factors such as delay discounting (DD; the extent to which someone prefers smaller, sooner rewards over larger, delayed rewards) and perceived stress may independently and interactively contribute to clinicians' high rates of LVAP. Examining how DD and perceived stress levels are related to each other and LVAP can help develop interventions to improve prescribing rates and promote high-quality patient care. In study one, we compared DD and probability discounting (PD; the extent to which someone prefers smaller, guaranteed rewards over larger, risky rewards) between cross-sectional survey samples of primary care clinicians and non-healthcare workers, finding that clinicians had lower DD and higher PD rates compared to non-healthcare workers. In study two, we analyzed data from the survey of non-healthcare workers to determine whether there was a significant relationship between DD, PD, and perceived stress levels, finding that both DD and PD were associated with perceived stress. In study three, we examined the associations between DD, perceived stress, and LVAP in a sample of clinicians from multiple departments across a large healthcare system. This included both self-reported likelihood of LVAP based on two clinical vignettes in a cross-sectional survey, as well as electronic health record-based incidence rates within 12 months prior to survey administration, and we found that DD was associated with self-reported LVAP likelihood in both clinical scenarios assessed. Ultimately, these three studies suggest that DD may be an important behavioral marker that warrants further investigation in the context of LVAP, a complex and recurring issue in healthcare. Future studies investigating the connection between LVAP and behavioral factors such as discounting and stress should seek to examine workplace-specific stress levels and further explore possibilities for interventions involving DD as a behavioral marker.
Voices of Strength: Counselor Experiences of Resilience and Wellness Among Refugee Youth
Perinchery, Saudamini Agarwal (Virginia Tech, 2026-03-30)
An estimated 117 million people worldwide are currently displaced due to war and other human rights violations, with refugee children and adolescents among the most vulnerable. Their heightened exposure to trauma, violence, and chronic instability places refugee youth at increased risk for mental health challenges such as post-traumatic stress disorder, anxiety, and depression. While resilience and wellness are increasingly recognized as protective factors that foster positive adaptation, little is known about how school and clinical mental health counselors understand and support these constructs in their work with refugee youth. This study explored how counselors experience the resilience of refugee youth in their care, as well as uncover their lived experiences in providing wellness support to them. Using Interpretative Phenomenological Analysis (IPA), six practicing counselors were interviewed to gain in-depth insight into how they experience resilience and implement wellness-based interventions. The analysis revealed four Group Experiential Themes (GETs) describing counselor experiences of resilience and four GETs illustrating their experiences providing wellness support to refugee children and adolescents. An additional GET emerged from the data analysis of the two original research questions established for this study, reported below. The following themes emerged for Research Question 1 pertaining to resilience: Counselors Experience the Resilience of Refugee Children by Observing their Responses to Significant Stressors; Counselors Experience the Resilience of Refugee Children by Observing their Individual Internal Strengths; Counselors Experience the Resilience of Refugee Children by Observing the Influence of Relational Support Systems; and Counselors Experience the Resilience of Refugee Children by Observing the Impact of Schools. The following themes emerged for Research Question 2 pertaining to wellness: Supporting Wellness by Viewing Wellness Through a Holistic Lens; Supporting Wellness by Building Trust; Supporting Wellness by Empowering Clients Through Advocacy; and Supporting Wellness by Practicing Culturally Responsive Care and Using Creative Approaches. The following theme emerged from the data analysis of Research Questions 1 and 2, pertaining to broader counselor experiences of working with refugee children: Working with Refugee Youth as a Source of Growth for the Counselor. Findings indicated that counselors conceptualize resilience as a dynamic interplay of internal (e.g., faith, hope, self-belief, and persistence) and external factors of resilience (e.g., family, community, educational access, and resources). Counselors reported utilizing holistic, strengths-based, and evidence-based practices to support wellness, and schools were observed to play a critical role in providing stability and promoting well-being. Results demonstrated a shift away from deficit-based perspectives toward approaches that recognize and leverage the existing strengths of refugee youth, illustrating how counselors translate theoretical concepts of resilience and wellness into practice. This study contributes a foundational framework to the literature by offering one of the first comprehensive examinations of resilience and wellness as enacted by counselors serving refugee youth. Implications for counselor education include the need for a linguistically diverse workforce, integration of resilience and wellness-oriented training in counselor curriculum, and increased emphasis on advocacy and collaboration with local agencies and schools. Recommendations for future research include longitudinal studies, greater attention to intersectionality, and exploration of effective training, supervision, and interdisciplinary collaboration models. Overall, this study reframes refugee youth experiences through a lens of strength and positive adaptation, advancing research, training, and practice within counselor education and systems of care.
Towards Unified and Generalizable Multimodal Foundation Models
Xu, Zhiyang (Virginia Tech, 2026-03-30)
Recent advances in multimodal models have reshaped multimodal learning by leveraging large language models (LLMs) as backbones and integrating them with vision encoders or diffusion models. These approaches have achieved strong performance in either multimodal understanding or multimodal generation. However, no existing system offers a unified framework capable of performing both understanding and generation across flexible input-output modality combinations while also generalizing to unseen, real-world tasks. Unification is essential not only for enabling a single model to perform traditional understanding and generation tasks, but also for enabling cross-modal tasks, such as visual storytelling, report generation, and script creation, that neither understanding nor generation models alone can accomplish. By processing and generating inputs and outputs that span multiple modalities, unified models more closely mirror the way humans naturally acquire and construct knowledge. Generalization, in turn, enables models to adapt to novel tasks in open-world environments under human-specified instructions or principles. Together, unification and generalizability constitute two fundamental pillars for advancing toward general-purpose multimodal intelligence. This dissertation advances unified and generalizable multimodal modeling through novel architectures, post-training paradigms, and reinforcement learning algorithms. It addresses two enduring challenges in multimodal foundation models: (1) the lack of modality unification, and (2) limited generalizability in open-world settings. The contributions are fourfold. First, we introduce Modality-Specialized Synergizers (MOSS), a framework that enables interleaved multimodal generation in pretrained models. Second, we propose an efficient unified architecture that bridges vision-language models and diffusion models, providing a novel pathway for joint understanding and generation. Third, we establish multimodal instruction tuning as a new post-training paradigm to improve zero-shot generalization and robustness. Finally, we extend image understanding to the spatiotemporal domain by developing a novel reinforcement learning algorithm that promotes temporal awareness, enabling vision–language models to reason effectively about videos. Extensive experiments across diverse multimodal benchmarks demonstrate that these approaches significantly enhance unification, generalizability, and overall capability. Collectively, this research strengthens the foundations of multimodal AI and outlines a pathway toward universal models that can understand, reason, and generate across modalities in complex open-world environments.
New Frontiers in Seismic Imaging of the Critical Zone
Eppinger, Ben Julius (Virginia Tech, 2026-03-27)
The critical zone (CZ) is the life-supporting skin of our planet, spanning the top of vegetation to the base of weathered bedrock. This thin layer is the only place in the known universe where biota, water, atmosphere, and geologic materials interact and transform. Direct observations of the critical zone's subsurface structure typically require invasive methods such as drilling boreholes or excavating soil pits. Seismic imaging has been used for decades to circumvent direct measurements of the subsurface, but traditional approaches often rely on a limited subset of the information contained within the seismic wavefield, such as the timing of first arrival waveforms. By leveraging more of the data contained in the seismic wavefield, new facets of critical zone evolution can be revealed. This thesis develops and applies advanced geophysical techniques, specifically full-waveform inversion (FWI) and multi-component surface wave analysis, to constrain high-resolution models of p-wave velocity, s-wave velocity, and radial anisotropy. In the first manuscript, a workflow is developed to implement 2D full waveform inversion (FWI) within the critical zone. The workflow involves inverting surface waves and body waves separately to ensure that high-amplitude surface waves do not dominate and overprint the information contained in body waves. When applied to a site near Laramie Wyoming, the resulting FWI models reveal that bedrock fracture density serves as an important bottom-up control on CZ architecture. These findings show that the transition from saprolite to intact bedrock is sharp in areas with low fracture density but more diffuse where the underlying rock exhibits higher fracture density. Additionally, the FWI models show better agreement with borehole data as compared to previously published first arrival travel time tomography models. The second study explores the role of inherited rock fabric in the development of critical zone porosity by measuring radial anisotropy with surface waves. This novel method utilizes multi-component surface seismic data associated with Rayleigh and Love waves to quantify radial anisotropy at the hillslope scale. Field data and in situ measurements from the South Carolina Piedmont demonstrate a strong correlation between seismic anisotropy and porosity, with both properties developing concurrently as rock undergoes in situ weathering. This empirical evidence suggests that weathering processes do not act stochastically, and instead, are guided by the geologic fabric of the parent material. Moreover, this research provides further evidence that inherited rock fabric plays a major role in dictating the form and function of landscapes. The final study investigates subsurface structure and water stores beneath giant sequoias in Yosemite National Park. By employing dense arrays for multicomponent nodal geophones, a revised time-frequency-phase FWI algorithm, and geostatistical rock physics modeling, this research estimates volumetric water content beneath giant sequoias at different landscape positions. The results indicate that giant sequoias located on ridges and hillslopes lack sufficient shallow soil moisture and must instead rely on deep rock moisture from depths exceeding 2 meters to avoid water stress during arid summers. As such, this work underscores the importance of rock moisture to valued species in arid landscapes. These three studies present several avenues for seismic imaging to catalyze research in the critical zone. The advent and integration of multicomponent, dense nodal data sets with advanced processing methods such as FWI means that previously undiscernible subsurface characteristics can now be elucidated. By contextualizing novel images of the shallow subsurface within the vibrant field of critical zone science, we can better understand how Earth supports life.
Multiphysics Modeling of Environment-assisted Cracking Properties of Advanced Materials for Aerospace and Marine Application
Mathew, Christian Chukwudi (Virginia Tech, 2026-03-23)
This dissertation develops a Multiphysics phase field-based model to predict the initiation, propagation, and recovery of corrosion damage in metallic alloys by coupling electrochemical dissolution, mechanical deformation, and repassivation within a thermodynamically consistent formulation. The framework addresses the challenge of predicting stress corrosion cracking (SCC) and corrosion fatigue by capturing the competing effects of passive film rupture, localized dissolution, and film reformation under mechanical loading. A coupled electro-chemo-mechanical phase-field model is established to simulate localized corrosion and pit evolution under both activation-controlled and diffusion-controlled regimes, with benchmark simulations—including pencil-electrode and semicircular-pit tests—used to validate the model against analytical solutions and experimental observations. The framework is extended to incorporate anisotropic elasticity and crystal plasticity, enabling analysis of corrosion-assisted crack initiation in single-crystal, bicrystalline, and polycrystalline 316L stainless steels. Orientation-dependent corrosion behavior observed in aluminum and other face-centered cubic metals is captured, producing anisotropic pit morphologies consistent with electron backscatter diffraction–based microstructural observations. Comparisons between conventional and laser powder bed–fused 316L microstructures demonstrate that grain morphology, crystallographic texture, and grain boundaries govern corrosion susceptibility and pit-to-crack transitions. An additional contribution is the formulation of a film rupture–dissolution–repassivation cycle that quantifies the cyclic interaction between electrochemical kinetics and mechanical stress through a time-dependent interface mobility, capturing passive film rupture, active dissolution, and repassivation-driven surface healing. Under cyclic loading, the model reproduces the asynchronous coupling between mechanics and corrosion, wherein tensile stresses promote rupture and dissolution, while compressive stresses enhance repassivation and crack closure.
Pavement Structural Monitoring of CCPR and FDR Using Non-Destructive Testing Methods
Benavides Ruiz, Carolina (Virginia Tech, 2026-03-23)
Cold Central Plant Recycling (CCPR) and Full Depth Reclamation (FDR) are pavement recycling techniques that provide a sustainable option for pavement construction and rehabilitation. However, these methods are not commonly implemented on high-volume routes due to the need for more detailed information on their field performance and the interactions between material properties and environmental conditions. This dissertation aims to document and evaluate the performance of semi-rigid pavement structures with CCPR and FDR layers through multi-year monitoring, including data collected using three nondestructive testing methods: the falling weight deflectometer (FWD), the traffic speed deflectometer (TSD), and embedded pavement instrumentation. The analysis includes three pavement sections from the NCAT Test Track (N3, N4, and S12) and two instrumented segments (Segment II and Segment III) on Interstate 64 in Virginia. The first study characterized CCPR and FDR materials by calculating in-situ moduli for three full-scale pavements on the NCAT Test Track, with a focus on the moduli of individual layers using linear and viscoelastic software. FWD data were collected at the beginning of three research phases (2012, 2015, and 2018). The results showed that the section including CCPR and FDR (S12) exhibited the lowest overall deflections compared to sections with only CCPR (N3 and N4). Deflections decreased over time, indicating increased pavement stiffness. Additionally, low deflection and estimated strain levels suggest that fatigue cracking is unlikely. The second study reports measured pavement responses, including strains and stresses, for two recycled sections on Interstate 64 subjected to actual environmental and traffic-loading conditions. Pavement structures were also modeled using layered-elastic software to compare measured and calculated responses. The results indicate that the pavement sections exhibited low strain, pressure, and deflection, suggesting a long structural life. Instrumentation remained generally functional after five years and may continue to provide valuable performance data. The third study assesses the functional and structural performance of two recycled pavement segments on Interstate 64 in Virginia. Structural performance was evaluated using calculated structural condition indexes and deflection velocity data from TSD. Deflection velocity profiles were compared with estimated values from 3D-Move software under various modulus scenarios. The scenarios that most closely matched observed data were further compared with measured pavement responses from in-situ instrumentation under loading conditions similar to those of the TSD rear axle. The findings demonstrate that both recycled pavement structures maintain good functional condition and favorable structural behavior after several years of service. Collectively, these studies advance understanding of pavement structural behavior under loading and environmental conditions, highlight the potential of recycled materials in flexible pavements on high volume routes, and promote innovation in pavement engineering. The results support the consideration of similar design approaches, including recycled-material foundations, when existing pavements require deep repair or reconstruction.
Machine Learning-Based Predictive Health Model of Turbofan Engine
Jung, Jin-sol (Virginia Tech, 2026-03-23)
Turbofan engine is one of the major elements providing power and thrust for aircraft. Maintaining the engine is vital for both safety and economy of aircraft operation. Besides, for an engine original equipment manufacturer, aftermarket services take account approximately 60% of company's revenue. Furthermore, predictive maintenance can reduce up to 30% of unscheduled removals which will lead to extended on-wing time for airliners. In this context, the impact of predicting in-service engine performance or deterioration has been of interest for many years. The idea is also aligned with digital twins supporting decision making for engine maintenance, repair, and overhaul. And it is not limited to safety improvement, engine life extension, efficiency in maintenance, and financial benefit. However, there are practical challenges in pursuing a predictive engine health model for in-service engines. One of these challenges is that it is difficult to consider all factors affecting the engine performance while building the model. Because each engine experiences a variety of environments as it travels worldwide. The emergence of artificial intelligence and machine learning (ML) opens opportunity to circumvent those challenges by building a surrogate model over data. For in-service turbofan engines, there is engine health monitoring (EHM) system that collects sensor signals during flights. EHM system provides limited but numerous information such as pressure, temperature, vibration, fuel flow, aircraft data, etc. It becomes useful data source as it conveys engine operational data for over its life cycle spanning over 20 years. The aim of this research is to build a framework of predictive engine health model applying ML to EHM data, by evaluating various approaches in each process of building a ML-based model. Turbine gas temperature was target variable as it is one of useful health indicator correlated with engine life. Additionally, it aims to provide insights and ideas that are deemed valuable to share within this community as a further contribution. The observation on EHM data provided a useful insight on its characteristics related to engine maintenance interval. Based on the observation, the new training approaches were proposed using data segmentation based on time between major overhaul. The proposed approaches were to use much less data (up to 65%) than a conventional train-test split method for model training. A sensitivity study to select predictor variables were conducted after Pearson covariance coefficient analysis. The models were built with long short-term memory (LSTM) network, however, linear and nonlinear regression algorithms were compared. Missing values were identified from the data observation and several missing value imputation (MVI) methods were discussed to evaluate the impact of those methods. One of the MVI methods used a physics-based engine performance model embedding the performance maps of engine components such as compressors and turbines. Furthermore, cross-engine Transfer Learning approach was proposed to explore a scalability and versatility of large generalized model. The results demonstrated that the predictive model using LSTM showed robust performance considering its prediction accuracy and number of outliers. There were promising results among linear and nonlinear regression algorithms. Furthermore, the proposed training approaches by the data segmentation based on time between major overhauls achieved prediction accuracy with a RMSE value of 4--6 C, even with 65% less amount of data than train-test split method. It was shown that a large generalized model with 45--300 times more data enhanced robustness of the predictive model by substantially reducing outliers and variability. Among various MVI methods, using the engine performance model showed the most improvement in prediction, 36% better than remove rows method. The detrending with extended Kalman filter significantly improved prediction accuracy across all algorithms and training approaches, with improvements reaching up to 87% in mean RMSE value, or 1.7C. The results of the Transfer Learning across different engines implied that the one ML model can be employed to different family engines when there is a commonality between them. Finally, to enhance the model's reliability for decision-making in safety-critical applications, uncertainty quantification was performed. An evaluation of Delta and Bayesian methods showed that the Delta method provided robust prediction intervals with high Prediction Interval Coverage Probability, demonstrating its effectiveness in quantifying prediction uncertainty.
"Where Does the Blood Go?": Constructing an Electronic Medical Records System in the United States, 1960-1990
Parrish, Roan Gabriella (Virginia Tech, 2026-03-20)
Electronic medical record systems are complex technologies that reflect the decisions of both its developers and its users. The construction of a new electronic record system in 1960-1990 Massachusetts and Vermont is an opportunity to explore these decisions and the specific ways in which technologies have values, priorities, and demands embedded within the artifacts. The two cases examined in this analysis act as contrasting foils to each other, demonstrating how the same type of technology can be constructed in distinctly different ways that shape the artifact's place within the workplace. One case is the Laboratory of Computer Science at the Massachusetts General Hospital, developing the Hospital Computer Project in the 1960s, the programming language MUMPS in the 1960s-1970s, and the publicly available system COSTAR in the 1970s-1980s. The second case is the problem-oriented medical record and its computerized version, PROMIS, from Lawrence Weed at the University of Vermont, 1964-1990. The cases are analyzed through the lens of Andrew Abbott's The System of Professions in order to explore the social, professional, and work components of these technical artifacts. The dissertation falls into the science and technology studies traditions of studying sociotechnical systems, boundary objects, controversies, and histories of technology.
Exploring The Realm of Legionella in Water Heaters: Hormetic Copper, Antagonistic Neochlamydia, Holistic Insulation and Protective Sediment
Roman Jr, Fernando Adali (Virginia Tech, 2026-03-20)
Opportunistic premise plumbing pathogens (OPPPs), including Legionella pneumophila and Mycobacterium avium, are increasingly associated with waterborne disease outbreaks in the United States. Despite their rising prominence, several factors influencing the growth and mitigation of Legionella and other OPPPs, such as pipe materials, microbial ecology, temperature gradients, water heater configuration, and the role of sediment are poorly understood. This dissertation sheds light on the role of these factors using complementary microcosm and pilot-scale experiments to examine effects of various control strategies on OPPP growth. Chapter 2 investigates how copper (0–2000 µg/L) impacted L. pneumophila and M. avium over an 11-month period using microcosms to simulate warm premise plumbing conditions. Copper was confirmed to have a hormetic effect, acting as a nutrient at low levels and an antimicrobial at high levels. At an intermediate copper dose of 250 μg/L, M. avium and total cell counts peaked in concentration, whereas the impact of copper on L. pneumophila at this dosage was stochastic amongst replicates. Comparatively, 2000 μg/L of copper suppressed M. avium, L. pneumophila, and total cell counts, though total cell counts steadily rebounded at this dosage throughout the entire experiment Chapter 3 closely examined the stochastic behavior of L. pneumophila in microcosms. A retrospective analysis on the copper-dosed microcosms described above and an earlier study of conditions leading to the Flint Water Crisis Legionnaires' Disease outbreak reveals a potential antagonistic relationship between Neochlamydia and L. pneumophila. This is consistent with prior evidence from pure culture studies that amoebae harboring Neochlamydia are resistant to L. pneumophila infection. The background water chemistry and pipe materials can influence the outcome of the competition between Neochlamydia and Legionella, suggesting probiotic approaches for OPPP control are already naturally occurring in some buildings. Two chapters of the dissertation focus on full scale experiments in residential water heaters. In chapter 4 recirculating, standard, and on-demand (tankless) water heater configurations were operated and evaluated for delivered hot-water temperatures, internal thermal profiles, energy efficiency, and volumes of water within temperature ranges at risk for Legionella growth. At a setpoint of 48°C, the recirculating configuration had the lowest energy efficiency and the highest volume of water at risk for Legionella growth, whereas at 66° C the standard tank had highest growth risk. Adding extra insulation to standard electric water heaters could eliminate Legionella growth habitat while reducing heat loss--a holistic approach to improving environmental sustainability and public health. Although thermal control is widely recommended for Legionella mitigation, chapter 5 reveals important limitations to the approach. Experiments using a typical 40-gallon residential water heater demonstrated that sediment, which can naturally accumulate at the bottom of water heater tanks, created dead space with temperatures that cooled with sediment depth. Even when the bulk water in the tank was completely mixed and the bulk water temperatures were ≥55–60 °C, portions of the sediment were cooler and permissive for Legionella growth. Two distinct mechanisms were responsible for cooling the sediment and its associated dead space: contact of the bottom of the tank with a relatively cold floor and settling of cold influent water. The sediment also increased the available surface area by several orders of magnitude compared to the internal surface area of the water heater. The cooler temperatures and high surface area could create extraordinary pathogen growth potential that causes culturable and molecularly detectable L. pneumophila to persist indefinitely. Use of external insulation between the tank and the floor can sometimes heat the sediment to a point that could eliminate Legionella growth. Chapter 6 is a Viewpoint that examines the interplay of truth, trust, and utility practices in pre-venting waterborne disease outbreaks by shaping water quality before it enters buildings to re-duce Legionnaires' Disease risk. For instance, infrastructure upgrades and improved treatment can increase disinfectant residuals and reduce sediment in supply waters, honoring a utilities to the social contract with their customers. We argue that continued denial and deception associated with this responsibility will lead to distrust and disinvestment, similar to that which arose from past failures to deal with health treats from lead contamination due to corrosive water. The overall dissertation illustrates how OPPP and L. pneumophila control in premise plumbing is influenced by ecological and engineering factors, including water chemistry, microbial inter-actions, temperature gradients, sediment and water heater design. These novel findings help explain why previously recommended mitigation strategies may not always suppress L. pneumophila in buildings and highlight the need for holistic control approaches targeting worst case niches in which Legionella can thrive.
Design and Assessment of an Embedded Die PCB-Based Traction Inverter
Spieler, Matthias (Virginia Tech, 2026-03-20)
Modular Microanalytical Systems for Trace-level Chemical Analysis
Thamatam, Nipun (Virginia Tech, 2026-03-20)
Airborne organic compounds, such as volatile organic compounds (VOCs) and organic aerosols (OAs), carry rich information about various ecological, environmental, physiological, and manufacturing processes. Growing concerns about environmental conditions and public health have driven the rapid development of analytical tools capable of characterizing trace levels of VOCs and OAs. Among other technologies, gas chromatography (GC) is considered the gold standard for analyzing organic compounds. However, conventional GC systems are expensive, bulky, and power-hungry, and require trained technicians to operate the system, limiting their adoption to a lab setting. Consequently, there is a growing demand for portable analytical systems capable of performing in situ, trace-level chemical analysis. To address this need, microelectromechanical systems (MEMS)-based microanalytical tools and micro gas chromatography (μGC) systems have emerged as powerful, miniaturized platforms for (bio)chemical analysis. They miniaturize the key components of a bench-top GC by implementing microfluidic devices, including a micropreconcentrator (μPC), a microseparation column (μSC), and photoionization detectors (PIDs). These developments drastically reduce sample-reagent volume requirements, accelerate processing times, and enable automation, making them ideal for in-situ applications. Although μGC miniaturizes the functionality of a bench-top GC, effective collection of trace samples and seamless integration of microfluidic components remain challenging. These challenges arise from design constraints, manufacturing complexities, diverse functional requirements, demanding operating conditions, and incompatible maintenance procedures, collectively limit system scalability and widespread adoption. This dissertation addresses these challenges by developing highly modular microanalytical tools for trace-level analysis of gas- and particle-phase organic compounds. Central to this work is a standardized Fluidic and Electrical Modular Interfacing (FEMI) architecture that provides removable, gas-tight, high-temperature-resistant fluidic connections for implementing complex microfluidic systems. Through the development of FEMI and a high-throughput μPC, this work demonstrates a portable μPC-based sample injection system compatible with bench-top GC, detecting VOCs down to 100 parts per trillion (ppt). This work also demonstrates FEMI-GC, a highly modular μGC that integrates μPC, μSC, and PID within a 3.75 L, 2 kg footprint, achieving a 700 ppt detection limit and a dynamic range over 50,000x. This work extends the functionality by developing a CNC-machined impactor and a μGC that successfully collected particles as small as 100 nm, enabling the detection of compounds with boiling points up to 450 ˚C, a first for μGC platforms. Finally, this work introduces the next-generation μGC called SPOCK (Size-filtered Particle Odor Chromatographic Kernel), which combines a CNC-machined cyclone, a CNC-machined impactor, μPC, μSC, and PID to simultaneously analyze VOCs and OAs. This lays the groundwork for a near real-time, highly configurable, fabrication-agnostic μGC platform for trace-level chemical analysis.
The Effect of Time and Safety on the Retention of Peace Education Concepts
Schupp, Julie Rebecca (Virginia Tech, 2026-03-19)
Peace education programs are widely implemented in schools and community settings worldwide, most often through short-term or time-limited interventions. Although these programs are widely used, research often measures success based on immediate, visible changes rather than on how children retain and express learning over time. This limitation is especially consequential for elementary-aged learners, whose learning develops gradually and is expressed in developmentally specific ways. This dissertation examines how elementary-aged children engage with and retain peace education concepts following a short-term instructional intervention, with particular attention to time and perceived safety. Using a mixed-methods design, the study integrates survey data and semi-structured interviews collected at multiple time points. Research was conducted in the Khanke Camp in the Kurdistan Region of Iraq, a displacement-affected setting where participants were Yezidi children living in an IDP camp, and Culpeper County, Virginia, a comparatively stable educational environment examined as a smaller, exploratory case. Findings indicate that short-term peace education programs generate recognition-based engagement and surface alignment with program values but rarely produce sustained conceptual retention over time. Children most often recalled peace education concepts through narrative memory, repetition, and emotionally salient experiences rather than through abstract explanation. Physical safety and institutional stability did not function as straightforward predictors of retention; instead, learning gained salience when program content intersected meaningfully with children's lived experiences and social environments. By conceptualizing learning as a developmentally mediated process of meaning-making and foregrounding recognition alongside articulation, this study challenges evaluation models that equate learning with immediate behavioral change or verbal sophistication. The dissertation contributes a developmentally grounded and contextually responsive framework for evaluating short-term peace education interventions with children, emphasizing the analytic importance of time, context, and ethical restraint.
Computational Models for Resistome Risk Assessment and Environmental Surveillance
Rumi, Monjura Afrin (Virginia Tech, 2026-03-19)
Antibiotic resistance (AR) is a major global threat to human health and economic stability. Without effective intervention, AR is projected to cause substantial loss of life and severe global economic burden. Addressing this challenge requires coordinated national and international efforts that consider all pathways through which antibiotic resistance can emerge and spread. Many public health and regulatory organizations advocate for a One Health approach, which integrates human, animal, and environmental health perspectives. Core components of this approach include the monitoring, risk assessment, and mitigation of antibiotic resistance. This dissertation addresses all three of these components through the development and application of computational methods for analyzing antibiotic resistance genes (ARGs) in environmental metagenomic data. In Chapter 2, I presented a risk assessment framework designed to evaluate the potential health risk associated with ARGs at a given location. Using metagenomic sequencing data, this framework computes a resistome risk score that quantifies the level of ARG contamination. The method was shown to be robust across different genome assembly strategies and varying sequencing depths. In Chapter 3, I applied this framework to 1,326 metagenomic samples collected from 12 distinct environmental types. This large-scale analysis disentangles biological (e.g., ARG relative abundance), ecological (e.g., taxonomic diversity), and technical (e.g., sequencing coverage) factors that influence resistome risk scores. The results demonstrate that risk scores are significantly affected by taxonomic diversity and are strongly correlated with anthropogenic markers. In Chapter 4, I introduced ARGfore, a forecasting model designed to predict future ARG abundance based on longitudinal surveillance data. Forecasting ARG trends enables earlier detection of emerging resistance risks and supports proactive mitigation strategies. Finally, in Chapter 5, I described WWTPredictor, a regression-based model that predicts the abundance of metagenomic features—specifically ARGs and microbial taxa—released into the environment from WWTPs based on incoming wastewater data. This model provides a quantitative framework for anticipating environmental discharge risks and supports data-driven decision-making in public health and environmental management.
A Graphical Approach to Identifying Structural Bias Using Directed Acyclic Graphs: Its Application to Two-Wave Nonequivalent Control Group Designs
Shin, Jaehyun (Virginia Tech, 2026-03-18)
It is well known that the analysis of covariance (ANCOVA) and the change-score analysis (CSA) can produce quite different treatment-effect estimates when applied to data from two-wave nonequivalent control-group designs, a phenomenon known as the Lord's paradox. Pearl's (2009) structural causal model (SCM) provides a useful and intuitively appealing tool to address the Lord's paradox. Using the SCM, Kim and Steiner (2021) combined the backdoor criterion with the path-tracing rules and showed that it identified the exact bias for the CSA. Though they implied that this graphical causal model approach could be applied to the ANCOVA case in a similar way, they did not explicitly show the details. Therefore, in the present study, to examine their implication, I applied the graphical approach to the ANCOVA and compared the results with the bias derived by the population ordinary least squares (OLS) method (Lüdtke and Robitzsch, 2025). The comparison exhibited a discrepancy, though the core part of the bias obtained by the graphical approach was correct. Specifically, the discrepancy occurred in the terms that were proportional to the core part of the bias implied by each backdoor path. This means that, though the detection of the sources of bias and the identification of the conditions to eliminate the bias could be completed by the graphical approach, the exact quantification of the bias was not possible. To resolve this shortcoming, I applied the so-called regression anatomy formula, also known as the Frisch–Waugh–Lovell (FWL) theorem in econometrics, and found that the proportional term could be expressed as the residualization-induced scaling factor. I then extended this graphical approach to different data-generating scenarios within two-wave nonequivalent control-group designs and confirmed that it worked well in all cases. The residualization procedure makes a graphical approach self-contained to identify the exact structural bias.
From Policy to Pathways: A One Health Assessment of Antimicrobial Resistance (AMR) in Wastewater and Surface Waters
Okeshola, Idowu Kayode (Virginia Tech, 2026-03-17)
Antimicrobial resistance (AMR) represents a complex environmental and public health challenge driven by interactions among engineered systems, natural ecosystems, and global antimicrobial use practices. While antimicrobial misuse in clinical and agricultural settings contribute to the evolution of resistance, environmental pathways such as wastewater discharge and watershed processes play a critical role in the dissemination of antibiotic resistant bacteria (ARB) and antibiotic resistance genes (ARGs). We applied a One Health (i.e., humans-animals-environment) framework to assess AMR policies, the extent to which wastewater infrastructure addresses AMR, and AMR inputs to surface waters. Comparative analysis of antibiotic use policies in Nigeria, Germany, and the United States through a comprehensive literature review highlighted how national stewardship strategies and regulatory differences can influence resistance dissemination across socioeconomic contexts. Longitudinal field studies in an urban watershed were conducted to assess the impact of wastewater treatment plant effluent and tributary inputs on shaping microbial communities and antibiotic resistance dynamics. Amplicon sequencing, shotgun metagenomic, and physicochemical measurements were used to characterize microbial community, ARGs, and metal resistance genes (MRGs) across spatial gradients. Treated effluent showed limited impact on downstream microbial communities and resistance gene profiles, highlighting treatment efficiency, whereas tributaries contributed distinct microbial signatures and elevated resistance signals. These findings highlight the need to integrate environmental monitoring, antibiotic stewardship, and infrastructure investment to address AMR while advancing understanding of environmental pathways that influence its dissemination.
Characterization of Antibiotic Resistance Genes using Network-based Approaches
Moumi, Nazifa Ahmed (Virginia Tech, 2026-03-17)
Antibiotic resistance is a natural evolutionary response to the selective pressures created by antibiotic use, but its rapid acceleration has become a major public health crisis. Resistance emerges through both vertical inheritance of chromosomal mutations and horizontal transfer of antibiotic resistance genes (ARGs) across diverse bacterial lineages, enabling the spread of drug-resistant infections and undermining effective treatment. Because ARGs circulate across clinical, agricultural, and environmental settings, mitigating resistance requires robust surveillance and mechanistic understanding of how ARGs function, move, and pose risk in real microbial communities. This dissertation develops network-based, multi-scale frameworks for studying ARGs from individual genomes to complex environmental metagenomes. In Chapter 2, we investigate what distinguishes ARGs from other genes within bacterial genomes using protein–protein interaction networks. By applying machine learning to interaction profiles, we identify patterns that differentiate ARGs from non-ARGs and reveal interaction signatures linked to resistance mechanisms and dissemination potential. Chapter 3 extends this analysis to metagenomic settings by extracting ARG genomic neighborhoods from metagenomic assembly graphs, enabling context-aware characterization of ARG mobility and horizontal gene transfer potential within microbial communities. Chapter 4 advances from genomic context to ecological risk by introducing a hazard quantification framework that scores ARGs based on their co-occurrence with mobile genetic elements, virulence factors, pathogens, and other resistance determinants, and applies this framework across diverse environments to study how hazards shift over time in response to external pressures. Finally, Chapter 5 synthesizes these insights into a predictive framework for identifying ARGs directly from metagenomic data. By integrating protein language model embeddings with graph-derived features from gene neighborhood graphs, this context-aware model captures both sequence-level signals and neighborhood structure, improving ARG recovery in complex metagenomic samples. Collectively, this work provides an integrated view of ARGs across biological scales, connecting molecular interaction patterns, genomic neighborhood organization, and environmental hazard to build more accurate and interpretable approaches for resistome profiling and hazard characterization.
The Development and Study of Protein Photocatalysts for Photoinduced Electron/Energy Transfer Reversible Addition-Fragmentation Chain Transfer Polymerizations
Anderson, Ian Carey (Virginia Tech, 2026-03-16)
This dissertation reviews literature relevant to the broader project of Biocatalyst Development for photoinduced electron/energy transfer-reversible addition-fragmentation chain transfer (PET-RAFT) Polymerizations. Chapter 1 traces the history of reversible deactivation radical polymerization (RDRP) and discusses the utility of photo electron/energy Transfer (PET) catalysis in RAFT polymerizations. We report, for the first time, an inherently photoactive protein that catalyzes PET-RAFT polymerizations. Zinc myoglobin (ZnMb) was identified as an inherently photoactive protein that was uniquely suited to photoinduced electron/energy transfer chemistry, and this protein was synthesized and used for PET-RAFT polymerizations. ZnMb proved to perform well, demonstrating all of the required features for a well-controlled polymerization, such as linear pseudo-first-order kinetics, linear growth in molecular weight with conversion, and maintained living chain ends, as evidenced by successful chain extension experiments. Inspired by the method developed for using ZnMb as a protein photocatalyst, we aimed to explore some of the consequences of using a protein in polymerization. For instance, we examined how polymer molecular weight affects chain-extension kinetics due to steric interactions with a restricted protein active site. Additionally, we investigated other reaction parameters to tune when using a protein, such as buffer composition and resulting protein stability. We demonstrate that by changing buffer composition and adjusting the salinity of the mixture, we can alter the kinetic performance of the polymerization while still maintaining a well-controlled process, as evidenced by linear pseudo-first-order kinetics, linear growth in molecular weights, and low dispersities.

Browse

Recent Submissions