Skip to main content

Algorithms for Tangent and Adjoint Sensitivity Analysis of Chaotic Flow Simulations

Speaker: Qiqi Wang, Massachusetts Institute of Technology

Presentation

Seminar Slide Deck (PDF-10.1MB)

Abstract

Many DNS, LES, and DES of turbulent flows exhibit chaotic dynamics. Algorithms for sensitivity analysis of these simulations can benefit design optimization, error estimation, adaptive meshing, and uncertainty quantification. It has been found that traditional algorithms for sensitivity analysis, either in its tangent or adjoint forms, may diverge exponentially, at a rate corresponding to the leading Lyapunov exponent of the chaotic simulation. A new algorithm has been developed to overcome this issue. The algorithm, based on the Least Squares Shadowing formulation, finds a non-diverging solution of the linearized governing equation. We will share the recent experience of applying this algorithm to several flow simulations.

Biography

Qiqi Wang got his B.S. in mathematics from University of Science and Technology at Hefei, China. He then got his Ph.D. in Computational and Mathematical Engineering at Stanford University, under the supervision of Parviz Moin and Gianluca Laccarino. From 2009 to 2015, he was an assistant professor of Aeronautics and Astronautics at MIT. Since 2015, he has been an associate professor in the same department.