Curtis Smith

Curtis Smith is the KEPCO Professor of the Practice of Nuclear Science and Engineering. Prior to joining MIT, he was the Director for the Idaho National Laboratory’s Nuclear Safety and Regulatory Research Division. While at INL, he led several risk-informed activities including the Risk-Informed Systems Analysis (RISA) Pathway under the DOE Light Water Reactor Sustainability Program and the Nuclear Regulatory Commission’s SAPHIRE risk analysis software development. Smith has published over 300 papers, books, and reports on risk and reliability theory and applications. He holds a PhD in nuclear engineering from MIT and a B.S. and M.S. in nuclear engineering from Idaho State University.

• 2022 Idaho National Laboratory Director Leadership award (2022)
• Idaho State University’s Professional Achievement Award from the College of Science and Engineering (2023)
• Awarded level of Fellow of the American Nuclear Society (2023)

Risk and reliability methods, teaching, and tools development

Risk and reliability methods provide a vital way to understand and manage complex systems such as nuclear fission and fusion reactors. These methods are based upon the science of probability and involve the prediction of hazards, failures, and consequences. Key approaches include fault tree analysis and Monte Carlo simulation. The application of risk and reliability tools enable practices including predictive maintenance, real-time risk monitoring, and digital twin forecasting. Applying these methods provides a way to analytically improve system performance, enhance safety, and understand why things fail.

Risk-informed decision making for nuclear and aerospace systems

Risk-informed decision making, or RIDM, is a way to proactively understand and manage performance shortfalls in nuclear and aerospace systems. Research in this area focuses on developing advanced probabilistic risk assessment, computational methods for risk models, uncertainty quantification, and risk knowledge-bases to better predict and manage complex system behaviors. A holistic approach to RIDM includes applying human factors to predict human reliability, integrating organizational safety, quantifying dependencies including causal factors, and integration of new technologies such as machine learning and autonomous controls of advanced systems.

Bayesian analysis

Bayesian analysis is a probabilistic approach to infer probabilities as new evidence emerges. Application of research in this area includes ways to extending computational algorithms such as Marko-Chain Monte Carlo and improving Bayesian methods for high-dimensional engineering modeling. While using established Bayesian tools such as MultiBUGS, R, and Python, we focus on applying these quantitative approaches to complex nuclear science and engineering problems.

Simulation and uncertainty quantification for complex systems

As nuclear systems have increased in complexity, our ability to model them has also increased. Consequently, simulation and uncertainty quantification have become a predominant approach for understanding and predicting the behavior of these complex systems. Our research focuses on developing and applying advanced simulation models to describe, characterize, and quantify where, when, how, and why systems deviate from idealized performance.

Human reliability modeling

Human reliability modeling focuses on understanding human performance when interacting with complex nuclear systems. Our research focuses on exploring data-informed approaches, probabilistic representation of human actions, and quantification of human reliability including extending the widely-used SPAR-H human reliability assessment method for risk assessment purposes.