讲座专家：QIAN Xudong（钱旭东）教授，新加坡国立大学
讲座题目：Digital Twins for Fatigue Crack Assessment from Specimen Validation to Structural Applications
讲座时间：2026年11月3日 09:30-11:30
讲座地点：海创大厦A座32层沙盘会议室
专家简介
Prof QIAN Xudong is currently a Professor in the Department of Civil and Environmental Engineering at the National University of Singapore (NUS), where he also serves as Director of the Centre for Offshore Research and Engineering (CORE) and Director of the Centre for Advanced Materials and Structures (CAMS). His research focuses on structural integrity and fracture mechanics of steel structures, digital twin-enabled infrastructure systems, and advanced manufacturing for infrastructure applications. His work integrates structural mechanics, data-driven monitoring, and cyber-physical systems for offshore, marine, energy, and cold-region infrastructures. He has secured more than S$24 million in competitive research funding and has led numerous large-scale industry-linked projects related to offshore structures, fatigue and fracture, digital twins, and coastal resilience systems. His research has received broad international recognition in the areas of fatigue, fracture mechanics, structural health monitoring, and digital twins. Prof Qian serves on the editorial boards of several leading international journals, including International Journal of Fatigue, Engineering Structures, Engineering Fracture Mechanics, Fatigue & Fracture of Engineering Materials & Structures. He is a member of the technical committee in American Society of Testing and Material E08 – Fatigue and Fracture and F42 – Additive Manufacturing.
内容摘要
Fatigue crack assessment requires reliable diagnosis of evolving damage and timely prediction of remaining life from limited monitoring data. This seminar presents a sensor-interfaced digital twin framework that integrates strain measurements with fracture mechanics and machine learning, progressing from specimen validation to welded components and potential offshore structural applications. The framework combines crack diagnosis, crack growth prognosis and model updating to support structural integrity assessment. At the specimen scale, strain-based approaches identify mixed-mode crack trajectories and surface crack dimensions, while computationally efficient surrogate models estimate fracture driving forces for crack growth prediction. Bayesian updating incorporates observed crack evolution to refine growth parameters and remaining-life estimates. Physics-informed learning further ad-dresses data scarcity, with examples demonstrating prediction of subsequent crack growth using only the initial 20% of the monitoring history. For welded components, the seminar examines how transfer learning extends crack diagnosis across joint configurations and loading conditions. A strain-based similarity indicator helps assess model transferability, while domain alignment addresses discrepancies be-tween simulated and measured strain fields arising from idealized geometry and fabrication variability. Experimental studies on cruciform and tubular joints illustrate these approaches under high- and low-cycle fatigue conditions. The seminar concludes with a proposed global-to-local workflow for offshore jack-up structures and a laboratory-scale demonstration, highlighting the integration needed to connect local damage assessment with asset-level models and inspection decisions. These developments provide a basis for more scalable digital twins for predictive maintenance.
