41656 research outputs found
Sort by
Administrator Reflections on Youth Sport Programming: Moving Forward After the COVID-19 Pandemic
Full text linkAbstractYouth sport has been a context where positive youth development (PYD) can be promoted (Fraser-Thomas et al., 2005). However, for youth sport to effectively foster PYD, all adult leaders need to understand and invest in the approach. The youth sport context is a large system that encompasses multiple stakeholders across the developmental lifespan of athletes (Dorsch et al., 2022). Much of PYD research reflects the contributions of parents and coaches (e.g., Harwood et al., 2019; Vella et al., 2011), yet often overlooked is the administrator role in PYD. Administrators are tasked with communicating and reinforcing organizational missions across stakeholders for the duration of the season (Schwab et al., 2010). Despite the essential responsibilities administrators hold within the organization, little research has examined how well missions are enacted in youth sport. Within PYD research, studies have found the importance of structuring programs to reach desired outcomes (e.g., life skills; Bean & Forneris, 2016). Thus, it would be beneficial to understand how administrators perceive the missions of, along with the implementation within, youth sport organizations
Grain Boundary Migration and Radiation Induced Segregation in Fe-Cr Alloys
Full text linkRadiation-induced segregation (RIS) is a significant phenomenon that occurs in alloys subjected to irradiation, particularly in environments such as nuclear reactors. This thesis investigates RIS in ferritic Fe- Cr alloys through the use of Atomic Kinetic Monte Carlo (AKMC) simulations, focusing on the interaction between solute atoms and migrating grain boundaries. The study explores the influence of temperature, solute concentration, and grain boundary velocity on solute drag, a critical process driving RIS. The results show that solute migration is strongly influenced by the presence of vacancies and interstitials generated under irradiation, which are absorbed by grain boundaries and other defect sinks. Simulations were conducted for different Cr concentrations (6% and 9%) at temperatures ranging from 500K to 700K, under different applied stresses. The work investigates how grain boundary migration rate and the segregation energy profile impact solute accumulation and redistribution at defect sinks. Notably, the thesis analyzes both thermal and radiation-induced segregation, highlighting the complex interactions between thermal vacancy diffusion and radiation-induced solute transport. The solute drag effect, governed by the velocity of grain boundary migration, exhibits distinct behaviors at different temperatures, with higher temperatures generally decreasing the effectiveness of solute drag. Additionally, the thesis explores the counterintuitive case of negative solute drag, where local compositional asymmetries result in apparent acceleration of grain boundary (GB) motion due to energetically favorable rearrangements of solute atoms. This phenomenon challenges traditional models and suggests that RIS can sometimes promote grain boundary migration instead of impeding it. Cluster analysis across varying conditions reveals the size, density, and spatial distribution of Cr-rich clusters, which are closely tied to segregation amplitude and GB kinetics. These clusters are formed as a result of local solute enrichment near the grain boundary, and their properties are influenced by the migration rate and the competition between different solute diffusion mechanisms. The migration of chromium precipitates near grain boundaries under radiation-induced conditions is also explored. Precipitate dissolution and reprecipitation processes were observed as the grain boundary migrates through the microstructure, with precipitation occurring under low-temperature and high-radiation conditions. The simulation results offer valuable insights into the roles of both vacancy-mediated and interstitial-mediated diffusion processes in governing the microstructural evolution of Fe-Cr alloys under irradiation, ultimately influencing their stability and performance in nuclear applications
Assessing Grafted Triploid Watermelon and Yellow Nutsedge Growth in Saline Agroecosystems
No full textIn an effort to better understand how saltwater intrusion events will impact watermelon growers a field trial was conducted to study how saline irrigation water will impact weed growth and seedless watermelon productivity at a joint Clemson University and USDA ARS facility. Weed management has always been a hurdle in vegetable crop production systems, requiring extensive labor even in a conventional setup. For organic growers, often the only economical approach is a combination of mechanical weed control techniques, including plastic mulches, tarps, and hand weeding. For watermelon growers, plastic mulch remains the most effective form of organic weed control available. In a watermelon plasticulture system, nutsedge species such as C. esculentus can remain a pest, as they have the ability to puncture plastic and must be removed by hand. Farmers in the coastal Southeastern United States also face a growing threat from extreme weather events and intruding salt water, which can impact the quality of irrigation water and deposit high levels of salt in their fields. It is well documented that high levels of salinity can negatively impact agricultural yields, however the interaction between salinity and weed competition has only been explored to a limited degree. The study was carried out under organic conditions, while sea water obtained locally was mixed with water from an irrigation pond to imitate an irrigation source that had been contaminated by a seawater intrusion event. While watermelon productivity was reduced in areas treated with saline water, the yellow nutsedge population in these areas was higher during the early season, indicating a potential increase in weed competitiveness under saline conditions. More research is needed to study long-term effects on weed populations; however, this could demonstrate an additional hurdle to growers trying to reclaim marginalized land that has been damaged by saltwater intrusion
Estimating Price Responsiveness for Service of a Luxury Transportation Company: A Zero Inflated Poisson Approach
Full text linkEstimating price sensitivity is crucial for premium service providers who face inconsistent client engagement. In this study, I investigate both observed ride demand and periods of client inactivity among ABC Transfer’s customers, meaning individuals who may not book rides in a given period but are not necessarily disengaged. ABC Transfer is a luxury ground transportation firm headquartered in upstate South Carolina. Using an unbalanced panel dataset spanning 2012 to 2024, I estimate a Zero-Inflated Poisson (ZIP) model to distinguish between individuals who might still be clients from those who are not likely to be clients any longer. The model accounts for price sensitivity across client types (personal, corporate, aviation) through interaction terms and incorporates time-fixed effects to control for changes in market conditions.
Results show that price is significantly negatively associated with ride volume, with a one-dollar increase in price predicting a 0.55% decrease in expected ride count for the baseline aviation client. While corporate and personal accounts exhibit similar sensitivity to price, their interaction terms are not statistically significant in the count model, indicating that the effect of price on ride demand does not differ meaningfully between account types after controlling for other variables. Corporate clients take significantly fewer transfer rides overall, with an estimated 72% lower ride activity relative to aviation clients.
In the inflation portion of the ZIP model, which estimates the likelihood that an individual who does not book a ride is no longer an active client, higher prices significantly reduce the probability of disengagement. These results suggest that while price increases may suppress ride frequency, they do not necessarily lead clients to abandon the service. Price sensitivity varies by client type in this regard: both personal and corporate clients are significantly less likely than aviation clients to disengage in response to price increases, indicating a stronger likelihood of remaining engaged with the platform.
Simulations using the Poisson component of the initial model predict that a $100 price reduction would yield the largest increases in ride volume among aviation clients, with more modest effects for corporate and personal clients. Strong temporal effects in both model components reflect market growth over time and cohort effects among clients in early periods. These findings suggest that while price influences ride frequency, it plays a more complex role in long-term client engagement. For luxury transportation providers, differentiated pricing strategies by client type may help maintain participation even as prices fluctuate, highlighting the importance of loyalty among high-value clients and the strategic potential of targeted retention efforts.
A second specification of the zero-inflated Poisson model was estimated to address the possibility that the average price an individual would face while potentially being a client is correlated with unobserved individual characteristics that also influence ride demand. This average price, calculated as the mean of the prices a client did pay or would have paid across observed periods, is included as a Mundlak control. In this version of the ZIP model, the previously negative effects of price on both the number of rides booked and the probability that an individual remains a client are no longer statistically significant. One possible explanation is that ABC Transfer’s pricing may not have varied enough over time or across clients for true effects, if present, to be statistically distinguishable from zero. In light of these findings, the results from the initial model should be viewed as suggestive rather than conclusive. While price is negatively correlated with ride frequency, it may also influence long-term client engagement in a counterintuitive manner
Life Cycle Assessment of Sustainable Wastewater Systems - Algae-Based Wastewater Treatment
Full text linkThis thesis studies the integration of an algae-based membrane bioreactor (MBR) and photobioreactor system into a domestic wastewater treatment facility, using the Clemson University WWTP as a reference plant. The proposed system replaces the conventional sequencing batch reactor and aerobic digester with a microalgae-bacteria symbiosis process, coupled with submerged ceramic membranes, an algal photobioreactor and anaerobic digestion for enhanced nutrient and energy recovery, which paves the way for water reuse and increased biological capacity in the future.
The study shows that overall energy consumption is reduced by approximately 18% compared to the existing conventional configuration. Further, methane production from anaerobic digestion of algae-enriched sludge enables on-site biogas recovery, offsetting nearly 19% of the plant’s total electricity demand through combined heat and power (CHP) generation. These improvements significantly enhance the energy profile of the facility.
The Life Cycle Assessment (LCA) of the current SBR system and the algae-based system shows a significant impact reduction in energy consumption, greenhouse gas emission, acidification, ecotoxicity, etc., for the algae-based system. The strategy presents a viable and sustainable upgrade path for future wastewater treatment infrastructure
Optimizing Park Locations While Considering Resident Behavior
Full text linkUrban parks and green-spaces significantly enhance community well-being by improving physical health, mental wellness, and environmental quality. Given these extensive benefits, ensuring fair and widespread access to urban parks represents a critical priority in urban planning. Despite the advantages of parks, optimizing their location poses unique and complex challenges distinct from traditional facility location problems, such as those involving emergency services or schools. The core distinction arises from the decentralized nature of residents’ park selection behaviors. Unlike centralized allocations typically managed by public administrators, park usage decisions are driven by individual preferences and behaviors. This decentralized decision-making introduces two additional challenges: ensuring fairness and addressing the uncertainty derived from varied and unpredictable individual preferences. Consequently, accurately capturing and modeling decentralized resident behaviors is central to effectively addressing these distinctive challenges. This thesis mainly contributes to existing literature by developing a mathematical model that integrates decentralized resident behavior into park location optimization decisions. To represent this behavior, the model employs an equilibrium allocation framework to capture endogenous factors, particularly the impact of park crowding level. Crowding level significantly influences park attractiveness, as increasing usage decreases comfort and convenience, thus reducing residents’ valuations. Incorporating crowding as an endogenous factor improves the realism between residents’ choices and park usage levels. The developed mathematical model seeks to maximize fairness in residents’ park access, targeting the improvement of conditions for the most underserved residents. Specifically, the objective function is structured to maximize the minimal perceived park valuation among all residents, ensuring that resource allocation addresses areas of greatest unfairness. Through this, the model actively reduces disparities in park access, enhancing overall community satisfaction.
Computational experiments were conducted using real-world data from the city of Asheville, North Carolina. Sensitivity analyses systematically evaluated how critical input parameters could impact model decisions and resident benefits outcomes. In summary, this thesis advances the field of park location optimization by addressing decentralization, fairness, and uncertainty through a equilibrium-based mathematical modeling frame- work. By incorporating user-choice behaviors and conducting analyses, the proposed approach provide decision-makers and urban planners with practical tools and critical insights to enhance fairly park access, improve community well-being, and optimize resource allocation strategies
Sensitivity Analysis of Flexible Pavement for South Carolina Conditions
Full text linkThe objective of the study was to evaluate the sensitivity of various input variables on the flexible pavement design thickness of high-speed, high-traffic routes in South Carolina using the Mechanistic-Empirical Pavement Design Guide (MEPDG) by means of the AASHTOWare Pavement ME Design software using global calibration coefficients with a focus on bottom-up fatigue cracking. The variables considered in this investigation included two-way average annual daily truck traffic (AADTT), asphalt mix type, climate station, subgrade type and resilient modulus, and aggregate base thickness. The study includes comparative analysis using older methods like the AASHTO 1993 method and the South Carolina DOT Pavement Design Guidelines, which is primarily based on the AASHTO 1972 method. Additionally, the study discusses the local calibration of the bottom-up fatigue cracking model for South Carolina for medium-level traffic and using machine learning methods like Artificial Neural Network (ANN), Gradient Boosted Method (GBM), and Random Forest (RF) to enhance model prediction. The results of the sensitivity analysis indicated that the asphalt mix type did not have a significant impact on the results. However, one of the five climate stations evaluated resulted in significantly thicker pavements than the others. Both subgrade types, as well as resilient modulus, had a significant effect on the pavement thickness. Finally, pavements were more sensitive to total truck traffic changes at lower AADTT values and then became somewhat less sensitive when exposed to the highest levels of traffic. The findings from the sensitivity study were used to develop a preliminary asphalt thickness design catalog for interstate routes in South Carolina. The results from the local calibration model showed high errors for bottom-up fatigue cracking in all the trials, but machine learning algorithm was able to increase prediction accuracy
Chemical & Electromagnetic Signatures of Binary Neutron Star Mergers
Full text linkBinary neutron star (BNS) mergers are some of the most energetic events in the universe, producing both short gamma-ray bursts (GRBs) and kilonovae. GRBs are brief, bright flashes of gamma rays, powered by ultra-relativistic jets launched during the merger. These jets produce a prompt burst of gamma rays, followed by a broadband afterglow generated by external shocks as the outflow interacts with the surrounding medium. Kilonovae are optical-infrared transients from the radioactive decay of heavy, neutron-rich nuclei formed through the rapid neutron capture process (r-process) in the merger ejecta. Together, GRBs and kilonovae provide unique laboratories for probing fundamental physics under extreme conditions, as well as studying the synthesis and transport of heavy chemical elements in the universe. This dissertation investigates the aftermath of BNS mergers through a combination of analytical modeling and numerical simulation, focusing on three interconnected aspects: the galactic-scale transport of r-process material, the long-term detectability of kilonova remnants, and the early-time dynamics of relativistic outflows in the context of GRBs.
Chapter 2 examines the chemical evolution consequences of BNS mergers occurring in the galactic halo. Developing a toy model for Rayleigh-Taylor-unstable ejecta, we demonstrate that the resulting r-process-enriched clouds rapidly cool, fragment, and are ultimately destroyed by Kelvin-Helmholtz instabilities during their descent toward the galactic disk. As a result, direct enrichment of star-forming regions is unlikely; instead, the ejecta becomes assimilated into the halo medium and can only contribute to chemical evolution via large-scale accretion flows or turbulent diffusion.
Chapter 3 explores the prospects for detecting long-lived gamma-ray line emission from kilonova remnants (KNRs), which persist for up to ~106 years following the merger. By combining binary population synthesis, galactic orbital modeling, and nucleosynthesis yields, we quantify the distribution of KNRs in the Milky Way and predict their gamma-ray fluxes. Our results show that while current gamma-ray telescopes lack the sensitivity to detect KNRs, next-generation instruments with moderate improvements could open a new observational window into late-time r-process decay signatures.
Chapter 4 focuses on the hydrodynamics of the early GRB afterglow phase. We re-derive a semi-analytic two-zone model for the shocked ejecta and circumburst medium, capturing the structure between the forward and reverse shocks, and test the accuracy of this model against detailed special relativistic hydrodynamic (SRHD) simulations. Comparisons reveal that the two-zone model systematically overestimates thermal energy in the reverse shock region when the reverse shock itself is Newtonian. We identify and clarify the relationships among key timescales - such as the deceleration time and the reverse shock crossing time - and demonstrate that standard approximations can mischaracterize the dynamics and emission of early GRB afterglows, particularly when the reverse shock is weak.
As summarized in Chapter 5, together, these studies provide a comprehensive picture of the physical and observable consequences of BNS mergers, spanning from the earliest moments of jet medium interaction to the gradual diffusion of heavy elements into the galactic halo, and culminating in the prospects for detecting the long-lived signatures of radioactive decay. Future work will include extensions to the project outlined in Chapter 4, as well as the study of tidal disruption events (TDEs) using both semi-analytic and numerical tools
Towards Securing AI Systems: Investigating Threats in Multimodal Autonomous Driving & RAG Systems
No full textArtificial Intelligence (AI) systems have become central to high-stakes applications such as autonomous driving and language-based decision support. As their deployment accelerates, ensuring the security and trustworthiness of these systems becomes paramount. Among the most stealthy and potent threats are backdoor attacks, where models behave as expected under normal conditions but exhibit malicious behavior when triggered by specific inputs, either digital or physical.
This thesis investigates novel backdoor and adversarial vulnerabilities across two emerging classes of AI architectures: (1) multimodal 3D object detection systems that fuse LiDAR and camera data, and (2) Retrieval-Augmented Generation (RAG) systems that pair large language models with document retrievers. Our work is structured across four chapters, each addressing a critical dimension of secure AI perception and reasoning.
We begin by establishing a robust baseline in multimodal perception through a weather-aware, attention-based 3D object detector. Designed for adverse driving conditions such as fog, rain, and poor visibility, the proposed model leverages dynamic multi-scale global-local attention to improve detection accuracy under diverse weather scenarios, laying the groundwork for understanding model behavior under complex operational conditions.
In the second part, we investigate digital backdoor attacks on camera-LiDAR fusion models. By inserting small, view-consistent 2D digital triggers into RGB images, we demonstrate that such artifacts can survive the fusion process and significantly distort 3D bounding box predictions. These findings reveal how cross-modal information propagation introduces new vulnerabilities, even in state-of-the-art camera-LiDAR fusion models.
Extending the threat model further, we present the first material-specific physical backdoor attacks for fusion-based detectors. By leveraging high LiDAR reflectivity materials as physical triggers, we show that attacks can be reliably activated in real-world driving environments across varying angles, lighting conditions, and viewpoints. This chapter bridges the gap between digital and physical threats by aligning simulation-based trigger design with field-level deployability.
Lastly, we examine adversarial threats in Retrieval-Augmented Generation (RAG) systems, which pair large language models with document retrieval mechanisms. We introduce a novel attack strategy, Adaptive Instruction Poisoning (AIP), that injects stealthy context-aware triggers through instructional prompts and retrieved malicious documents, without requiring model retraining or access to user queries. This work highlights how the modular design of RAG pipelines, particularly the separation between retrieval and generation, creates new and underexplored attack surfaces beyond traditional model-centric vulnerabilities.
Together, these contributions expose blind spots in modern AI pipelines where multimodal fusion, retrieval mechanisms, and deployment interfaces introduce overlooked but critical vulnerabilities. We conclude by outlining principles for designing secure-by-default AI systems and emphasizing the urgent need for holistic security evaluations that go beyond static model architectures