Applicable and Generalizable Machine Learning for Intelligent Welding, from Quality Prediction to Robotic Automation

Date: March 28, 2025; Time: 2:30 PM Location: PWEB 175

Abstract: In the last decade, the manufacturing sector has adopted Industry 4.0 innovations, including edge and cloud computing, Artificial Intelligence (AI), and Machine Learning (ML), enhancing production visibility, quality, automation, productivity, and safety. This presentation highlights novel ML applications in welding processes, through case studies in Resistance Spot Welding (RSW), laser welding, and arc welding.

The case study of RSW focuses on process sensing and modeling for quality prediction and defect detection. This study not only employs data-driven modeling but also utilizes ML to uncover physical insights into the RSW process, enhancing feature extraction and developing a more generalizable model for predicting quality and defects. It also introduces a new ML approach to create virtual signals for force and displacement using dynamic resistance measurements, addressing the lack of novel process sensing in facilities due to high costs. The case study of laser welding tackles feature engineering, i.e., from sensing data characterization to feature selection, to improve the model generalizability and decision-making efficiency in a plant production scenario. Transfer learning is also investigated to enable the ML models to adapt to dynamically changing welding conditions. The third case study targets the robotic automation of arc welding. To enable robotic operational adaptivity, a hybrid ML-based process characterization, and online adaptive control framework are developed for robotic arc welding to automatically and efficiently achieve the desired weld pool condition, given any initial conditions.  These case studies showcase significant potential for advancing welding processes to new levels of efficiency and effectiveness.

Biographical Sketch: Dr. Peng (Edward) Wang is currently an Associate Professor in the Department of Mechanical and Aerospace Engineering at Case Western Reserve University (CWRU). Dr. Wang has extensive experience in developing novel ML methodologies for machine condition monitoring and diagnosis, process modeling and quality prediction, and collaborative robots. Dr. Wang is the recipient of the CAREER award from the US National Science Foundation in 2023, Young Investigator Award from the International Symposium of Flexible Automation in 2024, Outstanding Young Manufacturing Engineer Award from the Society of Manufacturing Engineers (SME) in 2022, the Best Paper Award from the 2023 Manufacturing Science and Research Conference (MSEC), Outstanding Technical Paper Award from the SME North American Manufacturing Research Conference (NAMRC) in 2017, 2020, and 2021, and other best paper awards. Dr. Wang is an Associate Editor of the IEEE Sensors Journal and Journal of Intelligent Manufacturing.

Vetting Scaling Laws in Turbulent Reacting Flows: The Case of Damköhler’s Second Postulation

Date: March 14, 2025; Time: 2:30 PM Location: PWEB 175

Abstract: Damköhler’s second postulation has been the foundation of the development of scaling laws for turbulent premixed flames that led to the establishment of regime diagrams and has been used as the principal argument for explaining experimental and computed observables. Damköhler’s arguments are challenged based on direct numerical simulations of vortex-flame interactions and fully turbulent premixed flames under high Karlovitz number conditions. Specifically, the simulations could not prove that sub-flame thickness Kolmogorov eddies can enter the flame due to the high dissipation rate. Local analyses of both configurations showed that frequently used correlations based on the laminar flame structure could not be used to explain, among others, the reported thickening of turbulent flames under extreme turbulence levels. Additionally, laminar flame scales derived using detailed simulations resulted in a wide range of Karlovitz number values of the boundary separating the so-called thin reaction zone and broken reaction zone regimes and are not in agreement with established values in the literature, which have been derived from relatively simple theoretical arguments. Finally, the present results could not support even the existence of the thin reaction zone and broken reaction zone regimes, which have been hypothesized by Borghi and Peters and adopted in numerous computational and experimental studies.

Biographical Sketch: Fokion N. Egolfopoulos is a William E. Leonhard Professor of Engineering in the Department of Aerospace and Mechanical Engineering at the University of Southern California. He obtained his Diploma Degree in 1981 from the National Technical University of Athens, and his PhD in 1990 from the University of California at Davis after spending the last two years of his doctoral research at Princeton University. He is a recipient of the Silver Medal of the Combustion Institute at the Twenty-Second International Combustion Symposium, a Fellow of the Combustion Institute, a Fellow of the American Society of Mechanical Engineers (ASME), and an Associate Fellow of the American Institute of Aeronautics and Astronautics (AIAA). He has authored and co-authored one hundred and fifty-six (156) archival journal publications, eleven (11) editorial comments, three (3) book chapters, one hundred and sixty-two (162) conference proceedings and reports, and has given one hundred and seventy-two (172) invited and contributed scholarly addresses. Since 2009 he has been the Editor in Chief of Combustion and Flame, after serving as an Associate Editor of the journal from 2003 until 2008.

Exploring Modularity for Advancing Space Exploration and Supporting Crew Health

Date: March 7, 2025; Time: 2:30 PM Location: PWEB 175

Abstract: In the quest for exploring new frontiers in space, the design of modular systems has emerged as a potential solution that not only could enhance exploration capabilities but also support crew health and performance. A system is considered modular when its components are designed to function independently, each capable of performing its specific role without reliance on the entire system. At the same time, these components can seamlessly integrate to work collectively, forming a unified whole. This dual capability allows for flexibility, scalability, and adaptability, enabling the system to be customized, expanded, or reconfigured as needed to meet evolving requirements and adapt to different situations. Within a modular system, modules are designed to connect, interact, and exchange resources—either physically or virtually—through standardized interfaces or mechanisms. From modular robotic systems for exploration in unknown environments to modular systems for spacecraft habitats that can support crew health and activities.

Biographical Sketch: He is an Associate Professor in the Department of Engineering at Texas A&M University-Corpus Christi (TAMU-CC), USA. His research interests include the development and integration of Modular Robots and Modular mechatronic systems across different domains such as in Unmanned Autonomous Systems, Space, Agriculture, Industry, HealthCare, and Education. Dr. Baca has worked in the Autonomous Systems and Modular Robotics fields for over a decade and his work has led to multiple publications in leading conferences and journals, as well as organized and co-chaired international conferences and workshops. He has been involved in projects funded by federal agencies such as DoD, NSF, NASA, ED, and USDA-NIFA. He is co-founder of CORAL (Collaborative Robots and Agents Lab), and Faculty member of the NSF CREST-GEIMS (Center for Geospatial and Environmental Informatics, Modeling and Simulation) and the IUCRC (Industry-University Cooperative Research Center) Center for Growing Ocean Energy Technologies and the Blue Economy (GO Blue) at TAMU-CC.

Statistical Mechanics of Light- and Field- Responsive Soft Materials

Date: February 28, 2025; Time: 2:30 PM Location: PWEB 175

Abstract: Light- and electric field- responsive polymeric materials are important for emerging technologies in fields ranging from soft robotics to biomedical devices. However, engineering models of these materials are largely phenomenological, which inhibits systematic materials design. I will present our recent work on formulating statistical mechanical models that account for the coupling between light and electric fields to entropic polymer elasticity. First, we study polymers with photo-responsive mesogens that show spontaneous deformation when illuminated, due to a trans-cis bending of the mesogens. A statistical mechanical model that exploits a separation of energy scales between entropic elasticity and photoswitching is developed and shows the emergence of a broken symmetry in the coupling between light and deformation, which agrees with our experimental measurements of photoswitching and shape evolution. Second, we study the role of nonlocal electrical interactions in polymer chains. We develop a consistent non-perturbative model of electrical fields interacting with polymer chains, and show that the nonlocal nature of the dipolar self-interactions drives the collapse of a polymer chain above a critical field, providing a pathway to understand instabilities and failure mechanisms in polymer chains subjected to large electric fields.

Biographical Sketch: Kaushik Dayal is a professor in the Department of Civil and Environmental Engineering at Carnegie Mellon University. Dayal’s research interests are in the area of theoretical and computational multiscale methods applied to problems in materials science, with particular focus on bridging from atomic to continuum scales in the context of functional behavior, non-equilibrium response, and electromagnetic effects.

Dayal received his B.Tech. degree from the Indian Institute of Technology Madras (Chennai) in 2000. He earned his M.S. and Ph.D. in Mechanical Engineering at the California Institute of Technology in 2007.

Innovative Synergy: The Power of Design-Engineering Teams

Date: February 21, 2025; Time: 2:30 PM Location: PWEB 175

Abstract: In today’s fast-paced product development landscape, collaboration between engineers and industrial designers is more crucial than ever. However, misaligned goals, communication gaps, and process inefficiencies often hinder progress. In this talk, Robert explores practical strategies to bridge the divide between design and engineering, enabling teams to work faster and smarter without sacrificing creativity or functionality. Drawing from real-world case studies, he’ll discuss tools and methodologies—such as iterative prototyping, cross-functional workflows, and shared digital platforms—that foster synergy and streamline the product development process. Attendees will leave with actionable insights to enhance collaboration, reduce time-to-market, and create innovative, user-focused products. Whether you’re a designer, an engineer, or a product manager, this session offers a fresh perspective on building better products together.

Biographical Sketch: Robert is a distinguished Principal Industrial Designer of irwindesigned LLC, founded in 2007. He also works for Eureka in R&D as a Sr. Industrial Design Engineer in Advanced Development. Renowned for his expertise in sustainable product development and industrial design strategy, Robert has collaborated with leading clients like Amazon, Hoover, PepsiCo, and Hershey’s. His work spans user research, CAD modeling, prototyping, and environmental impact assessment. Beyond his firm, he has led groundbreaking projects, including the Amazon Dash Cart and net-zero energy homes, and secured numerous patents for his inventions. Robert is also an educator, founding the Learn Industrial Design platform for on-demand courses, and hosts the “Designing In The Wild” podcast. With a Bachelor of Arts in Industrial Design from The Art Institute of Colorado and certifications in sustainability and design thinking, he is a recognized thought leader in circular systems and life cycle thinking. His accolades include Consumer Product of the Year at the Colorado Inventor Showcase and features in prominent publications and media.