An efficient approach to solve nonlinear aeroelastic problems including thermal effect

The purpose of this dissertation is to develop a methodology for identification of computationally efficient reduced order models for aeroelastic problems. The methodology involves employing a system identification technique to develop a state-space based model from the output of a computational aeroelasticity code. The particular computational aeroelasticity code employed in this dissertation solves the nonlinear Navier-Stokes equations using a time-accurate finite-difference scheme. Linear structural dynamics equations are integrated simultaneously with the computational fluid dynamics equations to determine the time responses of the aerodynamic generalized forces due to a step excitation. Volterra based reduced order model kernels are constructed from the output of the aeroelastic solvers. An investigation of different approaches for identifying a technique to find the optimal linearized Volterra-based kernels to include the nonlinear effect of the aerodynamic system is presented, altogether with a scheme for extracting the first and second-order Volterra kernels. The output of Volterra based kernels to impulse responses are employed as the input to a modern identification technique that determines the Markov parameters of a linearized aerodynamic system. The Eigensystem Realization Algorithm (ERA) is then employed to develop an explicit state-space model of the equivalent linearized aerodynamic system. Flutter analysis of a typical section model, a rectangular wing, and swept-back wing are used to illustrate the identification methodology. Thermal effect and geometrical nonlinearities are added to illustrate the implementation of this effect for the reduced-order model. The Reduced Order Model (ROM) identification technique is extremely efficient and can model the type of nonlinearities commonly present in the aeroelastic analysis. The increase in computational efficiency is about four orders of magnitude. The reduced order model integration technique has a format suitable for multidisciplinary applications and it converges in one iteration.