Dissertation Defense Announcements

Climate change presents a pressing challenge for natural disaster management, to quantify its effects and associated disasters is a persistent challenge for regional climate risk studies. As climate-induced hazards escalate in intensity and frequency, infrastructure in hazard-prone regions faces growing risks – A situation especially critical to transportation infrastructures. Recent events, such as Hurricane Helene in 2024, which caused widespread damage to life supporting infrastructures and roadways closures, underscore the urgency of addressing these combined hazards. This dissertation assesses multi-hazard risks to bridge infrastructure in North Carolina’s mountainous regions, focusing on the interplay between landslide, flooding, wildfire, and earthquake risks. We approach the multi-hazard issue using landslide as the basic quantifier and investigate the nesting effect of earthquake and rainfall triggered landslides.
Because forest fire has the potential of diminishing soil moisture and can encourage landslides, wildfire risk is also included as a predictor. Analysis identifies key wildfire-related variables, such as distance to roads, elevation, and proximity to populated areas, as significant predictors of landslide susceptibility, highlighting the role of remote sensing data in extreme weather event prediction. Soil type, included in the landslide model, had limited impact, suggesting the need for refined soil classification methods in future studies.
Utilizing logistic regression (LR) and random forest (RF) models, this study develops predictive maps for landslide and wildfire susceptibility, achieving accuracy rates of 75.7% and 83.9% for landslide prediction and 68.5% and 72.9% for wildfire prediction, respectively. The higher sensitivity of the RF model, as shown in ROC curve analysis, demonstrates its effectiveness for multi-hazard risk modeling.
The wildfire susceptibility map is then incorporated as an independent variable in predicting landslide occurrences, revealing critical interactions between wildfire and landslide risks. The result are two different landslide susceptibility maps. Finally, a novel index, the Assumed Flooding Potential (AFP), is introduced to quantify flood risk. Since it is hard to establish flooding scenarios for bridges in mountain regions. AFP is calculated as the mid-span clearance for bridges. Furthermore, bridges-in-valleys are identified for high flooding risk analysis.
The integration of multi-hazard data allows for a dynamic understanding of bridge vulnerability, resulting in a shift in risk probability for certain structures. Specifically, the number of bridges with over a 50% probability of multi-hazard risk exposure decreased from 47 to 26, while four new bridges emerged in high-risk zones due to the addition of wildfire susceptibility data. These findings provide actionable insights for decision-makers, enabling proactive mitigation strategies tailored to bridges that face increased vulnerability from wildfire-triggered landslides.
This research delivers a high-resolution multi-hazard risk map and model for infrastructure resilience planning, offering critical tools for bridge engineers and policymakers. The 2024 Hurricane Helene landslides and bridge damage data from the state have been used to validate the risk maps. The results indicated reasonable accurate predictions, thus, ascertaining the study contributed to the potential to anticipate future multi-hazard risks. However, it also highlighted the need to address the complex interactions between environmental and anthropogenic factors and the urgency for future studies to advance our understanding of climate effects and to enhance our ability to anticipate and mitigate multi-hazard impacts on critical infrastructure in the face of evolving climate challenges.

Photoconductive detectors are semiconductor optoelectronic devices that absorb optical energy and convert it to electrical signal. However, photoconductive gain or quantum efficiency (QE) theory of photodetector exhibits considerable controversy in optoelectronics literature. Gain is generally defined as the ratio of the number of photogenerated charge carriers collected by the electrodes and the number of photons absorbed in the semiconducting photoconductor. This gain is often expressed as the ratio of the carrier lifetime over the carrier transit time. The lifetime is the average time before an electron recombines with a hole, and the transit time is the time needed for photogenerated carriers to travel from one electrode to another under an applied voltage. This simple theory implies that it is possible to obtain high gain by reducing the transit time.
In this dissertation, the gain theory of photoconductive detector with an intrinsic (undoped) semiconductor is reexamined by assuming primary photoconductivity. In contrast to the widely adopted gain formula as a ratio of the carrier lifetime to transit time, allowing for a value much greater than unity, it is shown that this ratio can only be used as QE under the low-drift limit, but has been inappropriately generalized in the literature. The analytic results for photocarrier density, photocurrent, and QE in terms of normalized drift and diffusion lengths are obtained, which indicates that QE is limited to unity for arbitrary drift and diffusion parameters. A distinction between the two QE definitions used in the literature, but not explicitly distinguished, is discussed. The accumulative quantum efficiency (QEacc) includes the contributions of the flow of all photocarriers, regardless of whether they reach the electrodes, whilst the apparent quantum efficiency (QEapp) is based on the photocurrent at the electrodes. In general, QEacc > QEapp; however, they approach the same unity limit for the strong drift. Furthermore, it is shown that the photocurrent in the photoconductive channel is in general spatially nonuniform and that the presence of diffusion tends to reduce the photocurrent. As one form of secondary photoconductivity, it is confirmed that doping in a photoconductive device can yield a gain, limited by the ratio of the mobilities of majority and minority carriers. Based on the simulation results, new analytic results that show good agreement with simulated results are proposed.
This work lays the ground for understanding mechanisms of experimentally observed, above-unity photoconductivity gains. Moreover, these findings should offer new insights into photoconductivity and semiconductor device physics and may potentially lead to novel applications.

Family engagement is a crucial component of student success, impacting academic performance, attendance rates, and behavior. However, many families, particularly those from historically marginalized communities, remain disengaged from their child's school due to barriers such as a lack of trust, negative experiences, and language or cultural obstacles. A foundational reason for this disengagement is the unpreparedness of teachers to intentionally engage families. Teacher education programs often do not have an explicit focus on family engagement, resulting in teachers who may feel unprepared and who do not understand the cultural context of their students' families; thus, hindering effective communication. This dissertation explored the preparedness of beginning teachers to engage families in elementary schools, and how they perceive this preparedness, particularly in urban settings. By examining how beginning teachers perceive their readiness, it provided insights into the strengths and weaknesses of teacher education programs in this regard. The research sought to answer two central questions: 1) How are beginning teachers prepared to engage parents and families in elementary schools, and 2) How do they perceive their teacher education program's preparedness for this task? The study employed a mixed methods approach, involving curriculum analysis, online surveys , and semi-structured interviews. The findings of this study informed recommendations for teacher education programs, looking to equip future teachers with the skills and knowledge needed for effective family engagement.

The widely reported increase in the frequency of high impact, low probability extreme weather events pose significant challenges to electric power system's resilient operation. This dissertation research explores strategies to enhance operational resilience that addresses the distribution network's ability to adapt to the changing operating conditions. We introduce a novel Dual Agent-Based framework for optimizing the scheduling of distributed energy resources (DERs) within a networked microgrid (N-MG) using the deep reinforcement learning (DRL) paradigm. This framework aims to minimize operational and environmental costs during normal operations while enhancing critical load supply indices (CSI) under emergency conditions. Additionally, we introduce a multi-temporal dynamic reward shaping structure along with the incorporation of an error coefficient to enhance the learning process of the agents. To appropriately manage loads during emergencies, we propose a load flexibility classification system that categorizes loads based on its criticality index. The scalability of the proposed approach is demonstrated through running multiple case-studies on a modified IEEE 123-node benchmark distribution network. We also test the proposed method with different DRL algorithms to demonstrate its compatibility and ease of application. We compared the results with the traditional metaheuristic algorithms namely particle swarm optimization (PSO) and genetic algorithm (GA). To gain a deeper understanding of the developed model, we conducted a sensitivity study. The key findings from this study align with the mathematical foundation of the approach outlined in this dissertation, providing further support.