Efficient Analysis For Aeroelasticity Based On Computational Fluid Dynamics

Aeroelastic analysis is very important in aircraft design. There are mainly two current aerodynamiccomputing methods for aeroelsatic analysis. The first is linear model based on lifting surface theory. It is a veryefficient method which is widely used in aircraft design. The accuracy of the linear model is low and it isn'tsuitable for solving nonlinear problem such as in transonic region or high angle of attack problem. The second isnon-linear CFD method. Complex non-linear flow can be simulated by this method. The disadvantage is the highcomputational cost which prevents it to be fit for qualitative analysis or parametric design. In order to solve thecontradiction between computational efficient and computational quality, and the contradiction betweencomplexity and convenience for analysis and design, the following works have been finished in this PhD thesis.A CFD based local piston theory, for supersonic unsteady aerodynamic calculating is developed and is usedin aeroelsatic analysis. In this method, a steady flow solution is first obtained by Euler method. and then pistontheory is applied locally at each point from its mean location. This method connects the high quality of CFD andhigh efficient of piston theory, greatly reduces the limitations of the classical piston theory on flight Mach number,airfoil thickness, and angles of attack, and is suit for supersonic or hypersonic flutter analysis.CFD based non-linear aeroelastic numerical simulation codes are developed. Numerical results for somegeneral aeroelastic models demonstrate their correctness and quality. In this work, two fluid/structure couplingsimulation methods based on aerodynamic extrapolation—improved Runge-Kutta scheme and hybrid linearmulti-step scheme—are brought up. Only one computation of the unsteady flow filed is needed at a physics stepwhile still keeps the high order time accuracy. It makes the coupling simulation of the fluid and structure not onlyefficient but also convenient and the existent unsteady flow solver needs few modifications.A technique called dynamic approximate boundary conditions is developed for solving the unsteady inviscidflow. Different from the traditional way in which unsteady flow field is calculated by deforming meshes at eachtime-step, dynamic approximate boundary conditions are satisfied on the stationary inner boundary so that onlyone set of meshes is needed in present numerical method. Only one set of static meshes is needed for the unsteadyflow solving and improves the efficient and robustness. The unsteady aerodynamic results and the flutter resultsverify the correctness and quality of the method.Using input-output difference model and least squares method, ROM (Red(?)ced Order Model) of unsteadyaerodynamic loads based on unsteady Euler codes is constructed. The input-output difference model is then turnedinto continuous-time model in state space. Coupled structural state equations and aerodynamic state equations,state equations of transonic aeroelastic system are constructed, An interesting S-type flutter boundary at thetransonic Mach number range is analyzed particularly. It is just because the flutter branch changes with theincreasing of dynamic pressure. By this method, the effect of control surface on airfoil flutter in transonic flow isstudied. We find some classical structural design rules are unfavorable for transonic flutter problem. Perhaps itaffects transonic aircraft structural design greatly.Coupling aerodynamic state equations, structural state cquations and servo state equations, a model fortransonic aeroservoelastic analysis is built. The closed loop aeroelastic characteristics of a typical missile arestudied. The results show the effects of the sensor's locations and structural filter on the aeroservoelastic system.Based on the work mentioned above, transonic flutter suppression of active control is carried out. BACT model isused to demonstrate the methodology. The control law designed by the suboptimal control method based on outputfeedback can successfully suppress the transonic flutter and an increase of up to 15% of the velocity index is easilyachieved.