Assistant Professor of Aeronautics and Astronautics

MIT, Cambridge, MA, USA

 

Abstract: Unsteady flow simulations are playing an increasingly important role in aerospace engineering. Many types of flows, including turbulent mixing and 3D flow separation, can only be accurately simulated by capturing their unsteady, often chaotic dynamics. While simulation of these flows (using LES and hybrid RANS/LES) has been enabled by rapid growth of computer power, computational methods for design of unsteady flows is still at the frontier of research.

 

This talk presents my venture beyond just performing simulation of chaotic unsteady flows, towards aspects of computational design. These include optimization, control, uncertainty quantification, and data based inference. We will discuss challenges encountered in attempting this for chaotic unsteady flows. In particular, we will discuss sensitivity analysis -- a common method used in various aspect of computational design. Sensitivity analysis computes derivatives of simulation outputs to input design or uncertain parameters; existing methods for sensitivity analysis fundamentally fails in chaotic unsteady flows. This is primarily due to exponential growth of small perturbations, known as the "butterfly effect" of chaos. New computational algorithms for solving these challenges will be presented with preliminary results.

 

 

Qiqi Wang is an Assistant Professor of Aeronautics and Astronautics at MIT. He obtained his B.S. in 2004 at University of Sciences and Technology of China (中国科学技术大学), and Ph.D. in 2009 at Stanford University. His research interests focuses on sensitivity analysis, design optimization, error estimation and uncertainty quantification of unsteady, chaotic fluid flows.(qiqi（at）mit.edu)

 

 

 

主持人：黄迅研究员

时间：8月10日（周五）下午1:30

地点：工学院1号楼212会议室