On a New Model Reduction Method for CFD Systems Used for Flow Control Design

Abstract

The use of active control to get better characteristics of unsteady internal and external flows is the ultimate goal of the research presented in this paper. Usually, unsteady flows are calculated using Euler and/or Navier-Stokes solvers. The efficiency of numerical simulation of an unsteady flow dramatically increases if the unsteady solution is a small perturbation about a steady-state flow, due to disturbances occurring at the boundaries of the flow domain. The main difficulty related to the flow simulation is that any CFD (Computational Fluid Dynamics) technique generates discrete systems with a very large number of states. In order to design an efficient control, the flow solver must be not only accurate and numerically effective, but also it must have a low number of states. The aim of this paper is to present a new method for model reduction of CFD systems using representative governing equations. The focus is on descriptor type systems resulting from the spatial discretization of the CFD governing equations.