MS-32 Uncertainty Quantification in Engineering Mechanics: Theory, Methods, and Practice

Michael D Shields, Johns Hopkins University
Alex Taflanidis, University of Notre Dame
Edoardo Patelli, University of Liverpool
Bruno Sudret, ETH Zurich

The development and application of methods for uncertainty quantification (UQ) are now widely recognized as an essential component to investigations (both computational and experimental) in engineering mechanics. The ability to quantify uncertainties plays a critical role in our ability to assess the performance and behavior of structural materials, soils, and fluids. These myriad uncertainties originate from a variety of sources including system excitation/loading, system properties (e.g. material properties), and the form of the mathematical abstraction of the system (i.e. model-form uncertainty). Moreover, they may be caused by inherent randomness (i.e. irreducible) or from a lack of knowledge or data (i.e. reducible) and this distinction may be important in the mathematical treatment required for their quantification. This minisymposium aims to bring together international researchers to present solutions to the challenges of UQ in engineering mechanics ranging from the fundamental theory of UQ, to new methodologies, and their application to wide-ranging problems in mechanics.

Topics of Interest Include:

• New methodology for efficient uncertainty propagation in expensive computational models (e.g. surrogate models, Monte Carlo methods, numerical methods, etc.)
• Theory and methodology to quantify model-form uncertainties
• Approaches to assess reliability and system performance with quantified confidence
• Inverse methods for quantification of uncertainty from limited data
• Novel applications of UQ, particularly in a large-scale, multi-physics, or multi-scale setting