Date: Friday, March 15, 2013
Location: 1084 East Hall (3:00 PM to 4:00 PM)
Title: Optimal Mode Decomposition: a new technique to analyse fluid flow data
Abstract: Many techniques exist to extract coherent information from a large sample of flow-field data such as an ensemble of PIV snapshots. Examples include Proper Orthogonal Decomposition, introduced to the fluids community in the 1960s, and Dynamic Mode Decomposition (DMD), developed by Schmid in 2010. Such techniques identify intrinsic mode shapes from the data set which can be used, for example, as a basis to construct a Galerkin approximation for the underlying system.
In this talk a new modal decomposition technique, Optimal Mode Decomposition, is presented and it is shown that it can be viewed as an improvement to and generalization of DMD. For a given dimension, the technique provides the optimal (in a sense to be defined in the talk) linear approximation to data's evolution by a system of that dimension. Such low-order flow representations are important, for example, if a control system is to be designed to influence the flow.
Speaker: Andrew Wynn
Institution: Imperial College
Event Organizer: Charlie Doering
|
