Low-Dimensional Modeling of a Driven Cavity Flow with Two Free Parameters

By applying Proper Orthogonal Decomposition (POD) one is able to extract a limited amount of data which characterizes a flow of interest. The modes resulting from the decomposition form a basis in the phase space on which a Galerkin projection of the equations of motion can be performed. By carrying out such a procedure one obtains a low-dimensional model consisting of a reduced set of Ordinary Differential Equations (ODEs) which models the original equations. A technique called Sequential Proper Orthogonal Decomposition (SPOD) is developed to perform decompositions suitable for low-dimensional models. SPOD is capable of transforming data organized in different sets separately while still producing orthogonal modes. A low-dimensional model is constructed and used for analyzing bifurcations occurring in the flow in the lid-driven cavity with a rotating rod. The model allows one of the free parameters to appear in the inhomogeneous boundary conditions without the addition of any constraints. This is necessary because both the driving lid and the rotating rod are controlled simultaneously. Apparently, the results reported for this model are the first to be obtained for a low-dimensional model based on projections on POD modes for more than one free parameter. Received 29 October 2001 and accepted 27 August 2002 Published online 13 January 2003 Communicated by P. Hall