Dr. Mohammad Abdo
Senior Staff Scientist
Idaho National Laboratory

Abstract

Nuclear Digital Twins (NDTs) are virtual/CAD replicas of nuclear assets that utilize continuous communication with sensory networks along with deep analytics to autonomously monitor, diagnose, predict and control all aspects of the physical assets over the whole life cycle. Nuclear applications lack the luxury of overloading the assets with enormous numbers of sensors and instrumentation for the inherited harsh conditions, cost constraints, and physical difficulties of reaching specific locations. Other challenges facing NDTs include the urge of having faster than real-time predictions to avoid and mitigate any undesirable events or accident scenarios. Moreover, continuous validation and real-time optimization of model parameters is the only guarantee that the discrepancy between the ROMS in the digital asset is not growing with time to render the models no longer representative of their physical counterparts. In this presentation, several capabilities and projects conducted at INL that are directly related and required for digital transformations are presented. Including types, components, and challenges facing NDTS.

Biography

Dr. Mohammad Abdo joined Idaho National Laboratory in 2019 coming from Kansas State University in Manhattan KS, where he served during the period 2018-2019 as an instructor in the mechanical and nuclear engineering department. In 2017-2018 he conducted research as a postdoctoral research associate and participated in constructing a time-dependent surrogate model for the TRIGA II reactor, the first step to propagate uncertainties over 70 years of reactor operations. Prior to that, Mohammad earned his Ph.D. from NCSU and spent a year as a postdoctoral researcher.

At INL, as a senior staff scientist in the department of digital reactors technology and development (C160), Mohammad is involved in several projects related to machine learning, deep earning, and digital twins, for instance, he participated in a surrogate-based sensitivity-informed Sodium loop conceptual design for the TREAT reactor. He is also the PI of a project on building metamodels to predict effective properties for advanced materials (i.e., porous materials and additively manufactured materials). In addition, Abdo is leading the AI portion of an LWRS-RISA project that uses machine learning to accelerate revolutionary algorithms for plant optimization under uncertainties. Furthermore, Abdo is the PI of Sparse sensing and Sparse learning in Nuclear Digital Twins, and the PI of a scaling validation and interpolation of a mock experiment to assess the representatively of such experiments to the target plant.

Wednesday, July 20. 2022
3:00 pm seminar
Room 1202 Burlington Labs