Roger Ghanem is the Tryon Chair of Stochastic Methods and Simulation at the University of Southern California (USC), where he is also a Professor in the Departments of Civil & Environmental Engineering and Aerospace & Mechanical Engineering. He joined USC in 2005, following faculty positions at Johns Hopkins University and SUNY-Buffalo.
Dr. Ghanem’s research focuses on stochastic analysis and computational science. For the past thirty years, he has explored the scientific, mathematical, and algorithmic foundations of uncertainty quantification, with applications spanning diverse fields in science and engineering. His recent work addresses challenges related to modeling errors and complex, interacting systems, particularly those involving multiscale and multiphysics behaviors.
He has co-authored over 200 journal articles in the areas of stochastic systems and predictive science and has supervised the research of more than 20 postdoctoral associates and 40 PhD students.
Dr. Ghanem has held several leadership positions, including serving as President of the Engineering Mechanics Institute (EMI), a member of the Executive Council of the U.S. Association for Computational Mechanics (USACM), Chair of the SIAM Activity Group on Uncertainty Quantification, and a member of the U.S. National Committee on Theoretical and Applied Mechanics (USNCTAM). He is a fellow of AAAS, SIAM, USACM, IACM, and EMI.
Dr. Ghanem has received numerous awards recognizing his contributions to research and teaching. His work has been supported by the National Science Foundation (NSF), the Office of Naval Research (ONR), the Air Force Office of Scientific Research (AFOSR), the Defense Advanced Research Projects Agency (DARPA), and the Department of Energy (DOE). He has also collaborated with industry leaders such as General Motors (GM), General Electric (GE), Daikin, Taisei, Takenaka, and Shimizu.
Condition assessment of spent nuclear fuel prior to permanent storage is an important component of their life cycle assessment. Given their radioactive nature, only external inspection is possible for these systems while damage and conditions of interest are unobserved. The anatomy of containment structures for spent nuclear fuel is quite complex, consisting of a hierarchy of subsystems interconnected at a finite number of joints, thus greatly filtering wave motion and restricting the flow of information throughout this system. In this talk I will describe our effort at prognosis for these systems. Specifically, we develop a highly detailed 200M DOF finite element model. Nested dynamic reduction techniques are used to facilitate the numerical evaluation of this model. Even with such level of detail, modeling errors are still unavoidable and are represented through a combination of non-parametric models (random matrix models for stiffnesses) and parametric models (random variables for joints and fasteners). A six hundred dimensional polynomial chaos expansion (PCE), that also captures non-parametric effects is constructed using projection-pursuit basis adaptation methods. Statistical updating is then implemented to assimilate experimental dynamical measurements made on the external casing with the aim of inferring the nature and location of damage inside the containment structure. The talk will detail the nested structural dynamics construction, the adapted PCE representation, and the specialized updating techniques.