Accelerating Uncertainty Quantification for Expensive Models: Application to Aerospace Problems

Description

CI Seminar - with Dr. James Warner

ABSTRACT
Scientific computing has undergone extraordinary growth in sophistication in recent years, enabling the simulation of a wide range of complex multiphysics and multiscale phenomena. Along with this increase in computational capability is the growing recognition that uncertainty quantification must go hand-in-hand with numerical simulation in order to generate meaningful and reliable predictions for engineering applications. If not rigorously considered, uncertainties due to manufacturing defects, material variability, modeling assumptions, etc. can cause a substantial disconnect between simulation and reality. Packaging these complex computational models within an uncertainty quantification framework, however, can be a significant challenge due to the need to repeatedly evaluate the model when even a single evaluation is time-consuming.

This talk discusses methods developed to accelerate uncertainty quantification for problems with expensive computational models, with a focus on leveraging reduced-order models (both in the physical and stochastic space) and high performance computing (HPC) to speed up analysis times. While the techniques presented are broadly applicable, they will be demonstrated in the context of applications that have particular interest at NASA, including structural health management, trajectory simulation, and additive manufacturing. Results from numerical test cases as well as experimental validation in the laboratory will be presented to show the effectiveness and efficiency of the methods presented.
 

BIO
James Warner joined the Cornell Civil and Environmental Engineering (CEE) Department as a PhD student in 2008 after receiving his B.S. in mechanical engineering from Binghamton University. His doctoral research focused on inverse problems and uncertainty quantification (UQ) for biomedical imaging applications. In 2012, James relocated to Duke University with his advisor and former Cornell CEE professor, Wilkins Aquino, where he completed his PhD two years later. Shortly after graduation, he accepted a position at NASA Langley Research Center in Hampton, Virginia as a Computational Scientist. At NASA, James develops and applies probabilistic methods for a range of applications including structural health management, radiation shielding design, and additive manufacturing. In particular, he focuses on scalable and efficient approaches for UQ that leverage high performance computing and machine learning. James is also the lead developer of the Scalable Implementation of Finite Elements by NASA (ScIFEN), a parallel finite element method code that is publicly available, and contributes to the development and release of several other open-source codes for UQ.