CEOE Ph.D, Student Seminar Series

ABSTRACTS

Improving weather models over The New York City Area and their applications by Jorge Bravo, Ph.D. Candidate in Civil Engineering

The New York City Metropolitan Area (NYCMA) is one of the most densely populated and globally diverse urban regions in the United States. As a major economic and cultural hub, it is highly vulnerable to complex weather systems originating from both tropical and high-latitude regions. These weather events can trigger widespread flooding that severely affects transportation and cause extensive property damage. Global-scale models often fail to capture local-scale processes accurately. To address this, we employ dynamic downscaling models based on physical equations, including both traditional and widely used models, as well as next-generation ones. As a case study, we have examined the remnants of Hurricane Ida, which affected New York City between September 1 and 2, 2021.

Enhancing AI Reliability and Practicality for Civil Infrastructure by Seyed Amirhossein Moghaddas, Ph.D. Candidate, Department of Civil, Environmental & Ocean Engineering

Machine learning applications in civil infrastructure face critical reliability gaps including achieving high laboratory accuracy often lack interpretability or fail during field deployment. This presentation presents systematic approaches to overcome these barriers through two research projects. The first project addresses the interpretability challenge in concrete durability prediction by employing Artificial Bee Colony Expression Programming (ABCEP) and Gene Expression Programming (GEP) to formulate chloride migration coefficients and derive explicit mathematical formulas that match black-box model accuracy while providing engineering transparency. The second project tackles the generalization problem in automated pipeline inspection using an ensemble YOLOv8 architecture specifically designed for instance-level class imbalance. Integrating DCGAN-based synthetic data generation with domain-informed preprocessing, the framework achieves successful outdoor validation across diverse environmental conditions. The unified contribution establishes that AI reliability in infrastructure applications demands methodological innovation combining domain expertise with data-driven techniques, advancing both sustainable material design and preventive maintenance strategies.

BIOGRAPHIES

Jorge Bravo is a professional in atmospheric sciences with over a decade of experience in weather modeling, remote sensing, and scientific programming with a focus on applied sciences. Skilled in both basic and applied research, applied to operational activities. Core expertise includes numerical weather and hydrologic prediction, remote sensing analysis, and the development of tools and workflows using programming languages and scientific software.

Seyed Amirhossein Moghaddas is a PhD candidate at Stevens Institute of Technology's Department of Civil, Environmental and Ocean Engineering, advised by Prof. Yi Bao. He earned his bachelor’s in civil engineering and master’s in construction engineering and management from Iran. His doctoral work has contributed to federally funded research (USDOT, NOAA, NSF), resulting in publications in premier journals including Advanced Engineering Informatics and Engineering Applications of Artificial Intelligence. His research interests encompass machine learning, computer vision, structural health monitoring, and sustainable material design, with a focus on creating trustworthy AI systems for real-world civil engineering applications.