Masters Theses

Permanent URI for this collection

https://hdl.handle.net/10919/9291

Browse

Now showing 1 - 20 of 23538

Designing Answer-Aware LLM Hints to Scaffold Deeper Learning in K–12 Programming Education
Bhaskar, Sahana (Virginia Tech, 2025-12-23)
Studies have shown that many K–12 students develop misconceptions about programming concepts such as variables, conditionals, and loops, particularly when learning through block-based environments like Scratch, where visual abstractions can obscure underlying computational logic. While tools powered by artificial intelligence (AI) can provide quick help, they often give direct answers that reduce students' opportunities to think and learn. This work explores how AI can support learning without encouraging overreliance. In a study with 105 students using CodeKids, 31.4% showed misconceptions about variable assignment and data types, and only 20% correctly solved conditional problems, highlighting the need for better scaffolding to address these conceptual gaps. To tackle this challenge, we designed and implemented an LLM-powered hint generation system within CodeKids, an open-source, curriculum-aligned learning platform developed by Virginia Tech in collaboration with local schools. The system generates short, step-by-step hints when students ask for help, encouraging reasoning rather than direct answer-seeking. Grounded in Vygotsky's Zone of Proximal Development, this approach balances guidance and autonomy through structured prompting that preserves productive struggle. The system was tested with real students and evaluated through automated analysis and surveys, which showed that the hints were clear, helpful, and easy to use. Students who used the hints reported higher confidence and improved problem-solving skills. These results demonstrate promising progress in using AI to support K–12 programming education and lay the foundation for future tools that personalize hints, adapt to different learners, and make AI-driven learning more effective and engaging.
Persona Reinforcement for Secure Programming AI Tutors: Adaptive Assistance in Action
Srinivasan Manikandan, Adithya Harish (Virginia Tech, 2025-12-23)
The world needs safe software, but producing it requires secure programming skills. Unfortunately, effectively teaching students secure coding skills remains a critical open problem. Pedagogy suggests that students acquire skills best through a combination of conceptual reasoning and practical experience. In computing, students gain this practical experience through hands-on programming exercises and projects. In the age of generative AI, modern tools, such as large language models (LLMs), often hinder effective learning by providing solutions to coding problems without engaging students in the learning process. Instead of internalizing the key concepts, students can simply copy answers without understanding, undermining the learning objectives. Unexpectedly, generative AI offers a promising opportunity to address the problem it has created, providing appropriate constraints and design. To that end, we present Secure Programming with Adaptive Reasoning Companion (SPARC), an AI-powered tutor designed to guide students through secure programming exercises, rather than directly provide solutions. Our design reinforces SPARC's tutor persona through a confluence of three techniques: (1) tailored prompt engineering, (2) a novel combination of AI techniques---coined as a learning safeguard proxy ---designed to prevent the tutor from directly providing solutions, and (3) a responsive algorithm that adapts responses to student proficiencies. We have integrated SPARC with SecureCoder, a drill-and-practice platform for secure coding skills, and evaluated its effectiveness via a pilot study. Across 120 study sessions (80 with SPARC and 40 with GPT-4o-mini), SPARC facilitated a 95% exercise completion rate compared to 80% for GPT-4o-mini, and pilot study participants demonstrated statistically higher satisfaction with SPARC's adaptability than GPT-4o-mini. Further, unlike GPT-4o-mini, all interactions with SPARC avoided providing participants with complete solutions. Finally, our study demonstrated that more than 85% of participants found SPARC's guidance to be clear, adaptive, and helpful, with 80% reporting improved understanding of secure programming concepts. Our evaluation suggests that SPARC's novel design achieves its goal of serving as a secure programming tutor. SPARC provides helpful guidance that most students found to enhance their learning experiences. As secure programming skills are vitally important, this work contributes to secure computing education by employing generative AI as an educator's ally, rather than its adversary.
Repurposing Antibacterial Compounds and Natural Products to Combat Vancomycin-Resistant Enterococci
Abdelmegeed, Somaia Mahmoud Abdelaziz (Virginia Tech, 2025-12-23)
Antibiotic resistance is one of the greatest threats to modern medicine, leaving clinicians with shrinking treatment options for life-threatening infections. Among the most concerning pathogens are vancomycin-resistant Enterococcus (VRE), which causes serious bloodstream, urinary tract, and wound infections, particularly in hospitalized patients. This thesis explores two complementary strategies to address this challenge: natural product discovery and drug repurposing. The first study investigated two drug candidates, CRS3123 and ridinilazole, originally designed to target Clostridioides difficile, for their activity against VRE. Both compounds showed exceptionally low minimum inhibitory concentrations and were bacteriostatic against VRE in vitro. In a Caenorhabditis elegans infection model, treatment with either compound significantly reduced the bacterial burden, demonstrating in vivo efficacy. Safety profiles were favorable, with minimal cytotoxicity and negligible hemolytic activity, highlighting their potential as safe and targeted therapies against VRE infections. In the second study, I evaluated maslinic acid, a naturally occurring plant-derived triterpene, for its potential against VRE. Maslinic acid inhibited bacterial growth and significantly reduced biofilm formation, an important mechanism that allows VRE to persist in hospital environments and resist treatment. Importantly, it showed low cytotoxicity in mammalian cells, indicating promise as a safe therapeutic scaffold. Together, these studies highlight the value of diverse approaches to antibiotic discovery. Natural compounds like maslinic acid expand chemical diversity, while repurposing candidates such as CRS3123 and ridinilazole accelerate potential clinical application. This thesis provides new insights into strategies for combating multidrug-resistant Enterococcus, a pathogen of urgent medical concern.
The Mangrove Mosaic: An Ecological Landscape Design Strategy for Everglades City's Climate Adaptation and Phased Transition Amidst Sea-level Rise.
Nandi, Prema (Virginia Tech, 2025-12-23)
Mangrove-based systems have significant potential to strengthen climate resilience in vulnerable coastal communities by protecting the coastline and adapting to rising sea levels. To explore ways to enhance the unique capabilities of mangroves, an action plan was implemented in Everglades City, Florida, chosen as a focused site. Located at the intersection of major ecological reserves and facing serious threats from significantly higher sea level rise, Everglades City presents ecological and cultural challenges as well as opportunities for land-use transformation. The project aims to restore natural hydrology in a historically disturbed landscape through ecological design strategies. The first step is to restore natural hydrology and appropriate salinity levels to support healthy mangrove growth and ecosystem function. A key part of the project involves "supercharging" mangroves by restoring the coastal mangrove belt and using a combined double-breakwater and Thin-Layer Placement (TLP) method to capture and retain sediments. The thesis also examines adaptation pathways for residents by exploring how mangrove restoration, migration corridors, community-driven decision-making, and long-term resilience planning can collectively create a sustainable future for Everglades City amid increasing climate change challenges.
Geospatial Trends of Per- and Polyfluoroalkyl Substances (PFAS) Incidence in Private Drinking Water in Virginia
Mclelland, Nicholas James (Virginia Tech, 2025-12-23)
Per- and polyfluoroalkyl substances (PFAS) are a class of synthetic organic compounds that are hydrophobic, thermally stable, and resistant to environmental degradation. Widespread industrial and household use has resulted in frequent environmental and drinking water detection, raising concerns about adverse human health effects associated with PFAS exposure. In response, the United States Environmental Protection Agency (USEPA) has established mandatory monitoring campaigns and future maximum contaminant levels (MCLs) for two PFAS compounds (PFOA and PFOS) in public water systems under the Safe Drinking Water Act. However, the 20-40 million Americans who rely on private drinking water supplies remain unregulated and comparatively understudied. This study investigates the incidence of PFAS in private drinking water under 'baseline' conditions and assesses the impacts of contributing land cover types, point sources, household characteristics, and traditional water quality parameters on PFAS incidence across Virginia. Point-of-use samples (n=382) were collected from private wells across 10 counties and analyzed for 30 PFAS compounds using USEPA Methods 533 and 537.1. Geospatial variables, household characteristics, and traditional water quality parameters (e.g., lead and bacteria) were analyzed using GIS and RStudio. At least one PFAS compound was detectable in all samples, with 90% exceeding method reporting limits, although median total sum PFAS concentrations were low (1.50 ppt). Short-chain PFAS compounds were more prevalent than long-chain legacy compounds in both total concentration and unique compound detection rates. The USEPA MCL of 4 ppt was exceeded in 2.4% and 5.2% of samples for PFOA and PFOS, respectively. While most samples had generally low total sum PFAS concentrations, 10% of samples exceeded 10.03 ppt with a maximum total sum PFAS concentration of 303 ppt. High PFAS sampled homes were associated with increased urban land cover, closer proximity to point sources, higher frequency of nearby point sources, older well age, elevated lead, and indicators of corrosive water chemistry, including low pH, and higher conductivity/total dissolved solids. These findings suggest PFAS concentrations in private drinking water are associated with more anthropogenic activity as well as potential mobilization of PFAS from in-home sources such as plumbing networks. Traditional water quality concerns remain prevalent, with exceedance of public water standards observed for lead (5.01% > 0.01 mg/L health-action-limit), E. coli (4.19% > absence), and total coliform bacteria (34.8% > absence). While 70% of homes employed some form of treatment, only 22% of homes used health based treatment types (e.g., reverse osmosis and activated carbon) which are capable of removing heavy metals, bacteria, or PFAS. These findings highlight the continued vulnerability of private drinking water users to both emerging and established contaminants and underscore the need for improved monitoring, targeted treatment adoption, and enhanced support for private drinking water supply stewardship.
Predicting Corn Response to Variable Synthetic Fertilizer Treatments Using UAV-Derived Imagery
Khulal, Aarati (Virginia Tech, 2025-12-23)
Efficient nutrient management is essential for optimizing corn (Zea mays L.) productivity while minimizing environmental and economic costs. Traditional methods for assessing crop responses to nutrients are often damaging and labor-intensive, limiting accurate assessment of spatial and temporal variations. Accurate in-season yield potential estimation plays a vital role in guiding nutrient management decisions and supporting grain marketing strategies. This study evaluated the potential of Unmanned Aerial Vehicle (UAV) derived imagery to estimate chlorophyll (Chl) status and predict corn grain yield potential in-season under variable nitrogen (N), phosphorus (P), and potassium (K) fertilizer treatments across different growth stages. Two field trials (NP and K) were conducted at two locations in Virginia, Kentland Farm in Blacksburg (Kentland), Valley and Ridge province, and the Northern Piedmont Center in Orange (Orange), Piedmont province. These sites vary in altitude, soil type, and climatic conditions, providing contrasting environments for evaluating crop responses to fertilizers. In both trials, factorial arrangement of treatments (varied N, P, and K fertilizer rates) with four replications was implemented with a randomized complete block design (RCBD). Chlorophyll readings (ChlR) were collected using the Soil Plant Analysis Development (SPAD)-502 and atLEAF Chl meters at three growth stages: early vegetative (EV), late vegetative (LV), and reproductive (Repr). These measurements were synchronized with UAV flights performed on the same day. UAV flights were conducted using DJI Mavic equipped with an RGB sensor for visible light and four monochrome sensors for multispectral imaging (red: 650 nm ± 16 nm, green: 560 nm ± 16 nm, near-infrared (NIR): 840 nm ± 26 nm, red-edge: 730 nm ± 16 nm). UAV-derived vegetation indices (VIs) responsive to Chl and indicative of crop yield potential were computed to model ChlR and yield through single and multi-index regression analyses. Multi-index model performance was evaluated through repeated k-fold cross-validation (CV) (k = 5; 30 repetitions). Indices included the Normalized Difference Vegetation Index (NDVI), Chlorophyll Index Red-Edge (CIRE), Normalized Difference Red-Edge Index (NDRE), Green Normalized Difference Vegetation Index (GNDVI), MERIS Terrestrial Chlorophyll Index (MTCI), Normalized Difference Chlorophyll Index (NDCI), Canopy Chlorophyll Content Index (CCCI), and Optimized Soil-Adjusted Vegetation Index (OSAVI). Weather variation, early-season drought, and late-season rainfall strongly influenced yield formation and grain moisture, overshadowing fertilizer treatment effects. No significant yield differences were detected among N, P, or K levels (p > 0.05). UAV-derived VIs demonstrated significant correlations with both ChlR and yield, with stronger relationships observed during the LV stage when canopy closure and Chl concentration were most stable. In the K trial at Kentland, the relationships between VIs and both ChlR and yield were generally moderate, while in the K trial at Orange, correlations were consistently strong and significant. For ChlR prediction, green and red-edge based indices (GNDVI, NDRE, CIRE, and MTCI) were the most reliable indices, explaining 40 to 55 percent of the variation across both sites and trials. For yield prediction, GNDVI, NDVI, and NDCI consistently exhibited strong relationships at Orange during the LV stage, with R² values ranging from 0.50 to 0.72 across both trials. In contrast, Kentland showed comparatively lower predictive performance, with only moderate relationships observed in the K trial during the LV stage. The use of a polynomial regression (quadratic) model further improved prediction accuracy compared to the linear model in all trials. Multi-index regression further improved predictive accuracy. The best-performing yield models were observed in the K trial at Orange during the LV stage, achieving CV R² values up to 0.71 (CV RMSE of 13.9), while the best ChlR models were found in the NP trial at Orange with CV R² of 0.46 (CV RMSE of 2.45). Model performance was lower for EV and Repr stages. Overall, these findings demonstrate that UAV-based multispectral imaging is an effective tool for monitoring corn canopy Chl status and assessing yield potential, with prediction accuracy varying across growth stages.
If These Walls Could Talk: How Architecture, Adaptive Reuse and Historic Preservation can be one.
Roberts, Sabrina Michelle (Virginia Tech, 2025-12-22)
The Built Environment represents on of the most complex and evolving challenges in contemporary society. As populations grow and environmental pressures intensify, the need to adapt and reuse existing structures has become increasingly urgent. Adaptive reuse and historic preservation, well known practices in architecture, offer sustainable solutions that bridge cultural heritage and modern function. These strategies are widely implemented across Europe, but those strategies have not quite reached the United States. This Thesis examines how adaptive reuse and historic preservation can be integrated into the American architectural context. Particularly through smaller scale, underutilized buildings that fall outside of the traditional preservation priorities. The studies focuses on a building the Washington D.C. that is similar to many building across the city that also have had the same fate. The proposed design re-imagines the structure as a mixed use development that com-bines affordable housing with ground floor retail. By employing federal and local historic preservation tax credits, the project demonstrates how economic incentives can support sustainable, community oriented redevelopment. The resulting design provides unique unites, shared amenities and spaces the enhance the livability and connection to the existent urban fabric. Through this case study, the thesis argues that adaptive reuse and preservation can serve as powerful tools for addressing the challenges of the built environment, in this case the hosing crisis, sustainability and conservation. It advocates for more of an inclusive approach to preservation, to value the everyday historic buildings as important resources for the future development of the urban built environment.
Beyond Visual Line of Sight Drone Simulator with User-defined Risk Layers
Anshebo, Surafel Tesfaye (Virginia Tech, 2025-12-22)
As Beyond Visual Line of Sight (BVLOS) operations become increasingly prevalent across a range of UAV applications, the need for reliable tools to support safe mission planning and dynamic risk assessment is growing. This work introduces a web-based simulation framework for BVLOS flight planning and validation, developed as a full-stack Flask application. The system integrates FAA sectional charts, NOAA weather overlays, and user-defined air and ground risk layers into an interactive browser interface. Users can draw waypoint paths, adjust altitudes, and dynamically reroute flights in response to hazards such as adverse weather or temporary airspace restrictions. To support rapid deployment and accessibility, the simulation stack is fully containerized using Docker and incorporates a Software-inthe-Loop (SITL) engine via ArduPilot. In addition to simulation, the platform supports scaled-down Hardware-in-the-Loop (HIL) validation using S-500 quadcopter drone, enabling a safe pre-flight evaluation of high-risk missions. The system architecture is modular and extensible, allowing integration of external tools for mission visualization and log playback. This framework offers a practical toolset for both researchers and operators seeking to improve UAV reliability in complex environments. Use cases include disaster response, remote infrastructure inspection, and agricultural monitoring. The results demonstrate the utility of simulation-driven planning in enhancing safety, mission effectiveness, and operator preparedness for real-world BVLOS deployments.
Integrating Machine Learning Techniques with Measurement-While-Drilling Data for Subsurface Characterization in Open-Pit Mines.
Addy, Jesse (Virginia Tech, 2025-12-22)
Measurement-While-Drilling (MWD) systems generate continuous drilling data that reflect subsurface conditions in real time. With the increasing availability of this data, there is a growing opportunity to use data-driven methods to support geological interpretation and geotechnical risk assessment in mining. However, the complexity and variability of drilling signals require analytical workflows that go beyond traditional threshold-based interpretation. This thesis integrates machine learning techniques with MWD data to improve subsurface characterization in two open-pit mines, each influenced by different operational and geological conditions. The first research component focuses on identifying zones of disturbed or weakened ground by detecting drilling behavior indicative of voids and compromised rock mass conditions in a mine affected by historic underground workings. The second component applies a structured data preparation and analysis workflow to develop predictive models for lithology and penetration rate in a separate open-pit operation, demonstrating how MWD data can support geological classification and drilling performance evaluation. Across both studies, the research highlights the importance of exploratory data analysis (EDA), feature engineering, and appropriate model selection. The results show that machine learning offers a scalable and effective way to extract meaningful information from MWD data, enhancing both geotechnical hazard detection and geological modeling. These findings demonstrate the value of integrating modern data science methods into mining workflows, contributing to safer and more informed operational decision-making.
Novel Synthesis and Characterization of Uranium-Zirconium Carbonitride by Direct Casting
Sanches, David Miguel (Virginia Tech, 2025-12-22)
UZrCN has exhibited thermophysical properties beneficial to high temperature reactor applications such as nuclear thermal rockets. The present work investigates a novel liquid phase synthesis method involving the admission of nitrogen gas during arc melting of uranium, zirconium, and carbon. Initial microstructural examinations using scanning electron microscopy indicated that the samples remain heterogenous with zirconium-rich cores in a uranium-rich matrix. Heterogeneity resulted from large differences in melting temperature of the major constituents and rapid solidification. Additional energy dispersive x-ray spectroscopy, combustion analysis, and inert gas fusion analysis proved that the core regions have both uranium and zirconium and establish that carbon and nitrogen are retained during the fabrication process. Powder x-ray diffraction analysis clarifies that the light elements did not form compounds with the uranium matrix but rather incorporated into the core region forming a sub-stoichiometric UZrCN. Heat treatment performed on equimolar U-Zr-C resulted in an increase in the homogenous phase present with a diffraction pattern reflecting a ternary UZrCN despite small amounts of segregation in the final button.
Evaluation of the Aerodynamic and Acoustic Performance of a Novel Passive Bat Deterrent for Wind Energy Applications
Van Horn, Charles Anders (Virginia Tech, 2025-12-22)
The growth of wind energy installations and the continued increase in turbine size over recent decades have contributed to a concerning rise in bat fatalities. Current mitigation strategies suffer from limitations, with curtailment decreasing energy production and active acoustic bat deterrents constrained by atmospheric attenuation, preventing full coverage of the rotor swept area. Passive ultrasonic deterrents integrated along the length of turbine blades offer a potential alternative, but their aerodynamic impact has not been thoroughly investigated. Implementation in the field requires an understanding of these devices to ensure they do not impose a significant penalty on turbine efficiency. This study evaluates the aerodynamic and acoustic performance of a novel passive bat deterrent design that utilizes resonant cavities exposed to airflow over the surface of an airfoil to generate noise. Two wind tunnel experiments were conducted across a range of flow speeds and angles of attack. The first investigated larger cavities designed to generate lower-frequency tones using direct lift, surface pressure, and acoustic measurements. The second tested smaller resonators targeting ultrasonic frequencies using surface pressure, wake pressure deficit, and acoustic recordings. The large cavities generated noise at approximately 10 kHz, though the acoustic performance was irregular at low freestream velocities, with some lower than expected frequencies and occasional lobes rather than tonal peaks appearing in the spectra. These resonators were also more efficient in deflecting the wind tunnel jet, leading to a drop in lift. The smaller cavities produced sound at around 23 kHz, with no risk of human audible noise pollution at lower frequencies. Lift reductions were on the order of 1% across the majority of flow conditions, and the small resonators decreased drag between 2% and 10% only at high angles of attack. Additionally, the aerodynamic impact of the resonators was less significant than that of an upstream trip, and the trip only decreased the acoustic response at high angles of attack.
Comradery, Class, and Consciousness: A Case Study in Comradery
McKinney, Ranger Egalite-Dionysus (Virginia Tech, 2025-12-19)
Recent attempts to save the US labor movement from declining relevancy and membership through social unionism have met with limited results. This work identifies this failure with social unionism's lack of interest in helping members create a robust class consciousness. To this end, I develop the idea of a comrade social bond through engagement with a diverse set of theorists and an intensive comparative case study of two unions during the period of the First Red Scare (1919-1920). I argue that comradery as a strong, disciplined, relationship oriented towards a revolutionary goal can create an environment in which members of unions understand their short-term efforts as part of the class struggle and thus build their class consciousness. My case studies test this theory by carefully demonstrating that a union with this social bond was able to retain members during a period of political repression, whereas a union without this social bond lost considerable membership, which indicates a strong class consciousness in the union with this social bond. Developing this theory holds direct implications for the ongoing struggle of labor unions to retain members, as well as vast potential for further theoretical development of the comrade bond as a method of organizing workers into sturdier unions ready to face the challenges of increasingly authoritarian capitalist systems of rule.
From Static to Adaptive: Dynamic Cost Function Weight Adaptation in Hierarchical Reinforcement Learning for Sustainable 6G Radio Access Networks
Viana Fonseca Abreu, Jefferson (Virginia Tech, 2025-12-19)
The rapid growth of mobile network traffic and the densification required for 6G networks significantly increase energy consumption, with base stations (BS) accounting for up to 70% of total network energy use. Energy-efficient BS switching has therefore become a critical research focus. Traditional solutions rely on static thresholds or fixed cost function weights, limiting adaptability in dynamic environments. This thesis investigates how cost function design and weight adaptation influence the trade-off between energy consumption and Quality of Service (QoS) degradation in Deep Reinforcement Learning (DRL)-based BS switching. Using a realistic spatio-temporal dataset, we show that static cost weights lead to suboptimal performance under varying traffic conditions. To address this, we propose a Hierarchical Reinforcement Learning (HRL) architecture in which a high-level controller dynamically selects low-level policies trained with different cost function weights. Experimental results demonstrate that the proposed HRL approach achieves up to 64% energy reduction—improving by 5% over the static DRL baseline—while maintaining acceptable QoS levels. These findings highlight the potential of hierarchical control and adaptive weighting in achieving scalable, sustainable 6G Radio Access Networks operations.
Unprecedented Candidate, Uncertain Coverage: Media Framing of Trump and January 6th
Barco, Michael William (Virginia Tech, 2025-12-18)
This thesis examines how U.S. news outlets framed the January 6, 2021 attack on the U.S. Capitol during the year surrounding Donald Trump's 2024 presidential candidacy. The study analyzes over 40,000 articles from fifteen outlets across nonpartisan-centrist, Democratic-favoring, and Republican-favoring categories, coding articles that mentioned both Trump and January 6th for use of the term "insurrection." Results show that nonpartisan-centrist outlets steadily reduced their use of "insurrection" between May 2022 and April 2023, declining by more than half across the period. Democratic outlets also showed a modest decline, while Republican outlets remained inconsistent due to low baseline use and smaller sample sizes. Alternative framings such as "protest" were rare and did not replace "insurrection." The findings suggest that language softened gradually rather than shifting at a single point, influenced by newsroom caution, public polling that showed Trump as a likely frontrunner, and professional pressures to maintain neutrality.
Restoring the Chocó: Growth and Survival of Native Chocoan Trees Using Different Propagation Methods
Aparicio, Sebastian (Virginia Tech, 2025-12-18)
The Chocó region is one of the most biodiverse places on the planet and has been identified as a priority area for conservation due to its high levels of endemism and diversity, currently threatened by the ongoing loss of forest cover. In this context, establishing baseline data on the survival and growth rates of native species can contribute to the design of effective ecological restoration strategies in the region. With this objective, we evaluated the establishment capacity of 28 native species propagated from large cuttings. The results showed that survival depended primarily on species identity, suggesting that propagation capacity by this method is associated with intrinsic adaptations. Although the use of large cuttings is a more economical method than using seedlings, there are certain limitations, such as the fact that not all species can be propagated by this method or the requirement of a high number of donors in order not to affect genetic variability. Additionally, the survival and growth rates of 23 native species (seven of them endemic to the Ecuadorian Chocó) were documented in a restoration experiment based on the applied nucleation strategy. The results showed high variability among species, while edaphic factors such as more acidic, less dense soils with gentle slopes and eastward exposure favored growth and survival. Pioneer species such as Castilla elastica, Ficus tonduzii, and Cecropia sp. promoted the development of highly diverse plantations, possibly by acting as nurse species. Finally, plantations with a higher density of islands (each composed of 25 individuals) showed increased growth rates, which could reflect competitive or facilitative interactions between individuals. This thesis presents key insights into the performance of various propagation methods and provides a scientific foundation to inform future forest restoration efforts in the Ecuadorian Chocó.
The Residual Strength of Liquefied Soil
Gallus, Erika (Virginia Tech, 2025-12-18)
Flow (or static) liquefaction is one of the most detrimental forms of ground failure. To determine flow slide potential, the residual strength of liquefied soil is needed. However, this is an extremely difficult parameter to estimate for soil deposits due to spatial variability of soil properties, potential for the formation of water films, the intermixing of soils, and the potential for partial drainage during flow liquefaction. Thus, the current state of practice for estimating the residual strength of liquefied soil (S^{r) is via back calculations using case histories. However, the complexity of flow slides makes case histories difficult to interpret, and combined with the limited number of case histories, it inherently implies large uncertainties in the derived empirical relationships. Although such empirical relationships define the current state-of-practice for estimating S^{r, laboratory studies and fundamental soil mechanics can provide insights and/or can be used to guide the form of the empirical relationships. For example, one issue that is an active area of debate is whether S^{r normalizes by initial vertical effective stress (σ'^{vo). Olson and Stark (2002) present an empirical relationship estimating residual undrained strength ratio (i.e., S^{r/σ'^{vo) as a function of normalized cone penetration test tip resistance whereas Kramer and Wang (2015) showed that S^{r does not scale linearly with σ'^{vo. Hence, the objective of this study is to develop a more mechanistic understanding of the residual shear strength of liquefied soils based on fundamental soil mechanics and laboratory studies. Specifically, an expression for S^{r/σ'^{vo is derived in terms of the effective angle of internal friction for residual strength, ϕ'^{r, and Skempton's pore water pressure coefficient for residual conditions, �^{r. Data from published laboratory studies are used to develop correlations for estimating both ϕ'^{r and A ^{r. The results show that ϕ'^{r is relatively constant for a range of sands, and the variability in its value does not significantly affect the computed value of S^{r/σ'^{vo. Additionally, the results show that �^{r correlates with the state parameter (ψ) for initial conditions and depends on whether the soil grains crush. Additionally, the value of �^{r significantly affects the computed value of S^{r/σ'^{vo. The derived laboratory-data-based ψ - �^{r relationship is used in conjunction with an empirical relationship relating ψ and normalized cone penetration test (CPT) tip resistance (Q^{tn) to develop a relationship relating S^{r/σ'^{vo to Q^{tn, which is then compared to similar relationships derived from back-analysis of case histories. The comparison shows that the proposed correlation most closely resembles Robertson's (2010) correlation once adjustments are made to the relationship between Q ^{tn and ψ.}}}}}}}}}}}}}}}}}}}}}}}}}}}
Nonlinear Polymer Nanocomposite for Field Grading in Medium-Voltage Power Converters under High-Altitude and Humid Environments
Zintak, Zachary (Virginia Tech, 2025-12-18)
This thesis presents the development and characterization of a nonlinear resistive polymer nanocomposite (PNC) coating designed to enhance insulation within medium-voltage (MV) power modules and suppress flashover on printed-circuit boards (PCBs) at high altitudes. Electric field simulations of the triple-point (TP) region revealed strong E-field intensification at conductor-ceramic-silicone and conductor-FR4-air interfaces, leading to premature partial discharges and breakdown. To mitigate these effects, a PNC coating composed of a polymer matrix with dispersed conductive nanoparticles was applied as a conformal field-grading layer. Electrostatic force microscopy (EFM) measurements exhibit an average distance of 135 nm between nanoparticles within the polymer matrix. Finite element simulations conducted in COMSOL demonstrated that the nonlinear conductivity of the PNC effectively redistributed the local electric field, reducing the peak intensity at the TP by approximately 50% compared to an uncoated interface. Experimental validation through partial discharge inception voltage (PDIV) and breakdown voltage (BV) tests confirmed that the PNC coating increased surface flashover voltage by approximately 30% under both ambient and low-pressure conditions when exposed to air. Humidity aging and condensation tests were performed to assess the long-term reliability of the coating within power modules. The PNC maintained its insulation improvement ability under prolonged high-humidity exposure, showing no measurable degradation in insulation strength. Overall, this work demonstrates a robust and environmentally stable nonlinear coating for surface field grading in MV power modules and converters. The PNC provides a promising pathway toward improving partial discharge immunity and insulation reliability in high-power, high-voltage electronic packaging applications.
Categorizing and Comparing Students' Interactions in eTextbooks
Sapkota, Jharana (Virginia Tech, 2025-12-17)
The rise of interactive eTextbooks opens new opportunities for enhancing student engagement and learning outcomes. However, analyzing student interactions within these digital platforms remains challenging. This study examines student engagement profiles in OpenDSA, an interactive eTextbook for data structures and algorithms courses. Using session-level interaction data, we categorize engagement into four distinct engagement states, defined as types of student activities: Reading, Visualization, Proficiency Exercises, and Multiple-Choice Exercises. Although OpenDSA also integrates third-party programming exercises through CodeWorkout, these activities were excluded from our analysis because the fine-grained interaction logs required for behavioral modeling were not accessible. We then apply clustering techniques to identify distinct engagement profiles, characterized by the frequency of transitions and total engagement time spent in each engagement state. Our research addresses two key questions: (1) What engagement profiles can be identified from students' interactions across these four engagement states? (2) How do these engagement profiles correlate with students' academic performance? Our findings reveal four distinct engagement profiles: Highly Engaged Learners, exhibiting frequent transitions and high engagement across all engagement states; Moderately Engaged Learners, characterized by sporadic interactions and below-average overall engagement; Balanced Learners, maintaining consistent and moderate engagement across engagement states, and Minimally Engaged Learners, demonstrating limited engagement and infrequent state transitions. Statistical analysis confirms that students in profiles with frequent and diverse engagement significantly outperform minimally engaged learners academically. These results underline the critical role of active, varied engagement in student success and underline the potential of session-level data for monitoring and optimizing student engagement. We believe our findings will be valuable to eTextbook developers, providing actionable insights to guide the design of digital content and targeted interventions that improve student engagement and performance.
"Behind Closed Doors": Identity and Influence in State-Level Legislative Decision-Making on School Discipline
Barnes, Jordan Isaiah (Virginia Tech, 2025-12-17)
This thesis critically examines state-level legislative decision-making in Virginia, New York, and Texas, with a particular focus on school discipline reform. Specifically, it explores the extent to which the race and gender of state legislators correlate with policy decisions concerning disciplinary practices in public education. Through a rigorous analysis of legislative actions and policy outcomes, this study interrogates the role of "diversity, equity, and inclusion" as a mechanism for addressing systemic inequities within a multicultural society. Through an observational study of three states, this thesis reveals how legislative priorities shape school discipline policies, thereby highlighting the importance of cultivating a diverse and representative state legislature. It argues that increased legislative diversity can serve as a critical tool in dismantling the School-to-Prison Pipeline and mitigating systemic disparities within public education.
A Comparative Study of Machine Learning and Traditional Techniques for Grade Prediction and Grade-Tonnage Evaluation in a Small VMS Deposit
Bag, Cemile Dilara (Virginia Tech, 2025-12-17)
Small-scale, high-grade volcanogenic massive sulfide (VMS) deposits present unique challenges for resource estimation due to their strong grade variability and complex geological structures. This thesis evaluates whether machine learning methods can improve grade prediction and tonnage estimation compared to traditional methods. A three-dimensional block model with 5 x 5 x 5 m resolution was constructed in Vulcan, and grade estimation was performed using Inverse Distance Weighting (IDW), Simple Kriging (SK), Ordinary Kriging (OK), and ensemble tree models. Traditional methods were assessed using cross-validation within Vulcan, while machine-learning models were evaluated using an independent train-test split. Approximately six million block centroids were exported for full model prediction to compare all methods directly. Machine learning models produced the highest accuracy in the test set but generated low-level noise predictions across sparsely informed areas. A filtering threshold of Au > 0.0001 g/t was applied to mitigate this effect and achieve geologically realistic tonnage estimates. Spatial block-model comparisons, residual analyses, and grade-tonnage curves showed distinct behaviors among methods. IDW yielded the highest tonnage at low cutoffs, Simple Kriging and Random Forest exhibited similar behavior in sparsely informed areas, and Ordinary Kriging consistently produced conservative tonnage estimates. After filtering, ensemble machine learning models provided improved grade discrimination and preserved localized high-grade zones more effectively than traditional methods. This study demonstrates that machine learning approaches can complement traditional methods and offer enhanced performance for small VMS deposits. The results highlight practical considerations for applying machine learning in early-stage resource evaluation and emphasize the need for domain-based modeling in later stages.

Browse

Recent Submissions