Parametric modelling for determining aircraft stability & control derivatives

Abstract

This thesis is part of a project that focuses on developing an optimisation framework for dynamically scaled flight testing. The optimisation framework must design the scale model and the flight test such that the performance of the scale model is representative to the full-scale aircraft. The similarity in the performances between the scale model and the full-scale aircraft is achieved by altering the geometry, the mass distribution and the structure of the model, which are related to the aerodynamic, flight dynamic and structures disciplines respectively. This thesis contributes to the aerodynamics optimisation framework by focusing on the development of a parametric model that is capable of deriving the stability and control derivatives. The research goals of the thesis are mostly related to the construction of a parametric aircraft model for the aerodynamic solvers that are based on a first order panel method like VSAERO. The most important goal is how to model the trailing edge moveables for a first order panel method for deriving the control derivatives. This thesis will investigate three options for creating the moveables. The first option (’normal rotation’) is a mathematical operation in VSAERO that rotates the normal vectors of the body panels that represents the moveable. The second option (’transition surface’) is a wing with a moveable model that has transition surfaces between the wing and the moveable in the spanwise direction. The third option (’gap’) is the same model as method two, but instead of transition surfaces, there is a gap between the wing and the moveable. The second research goal is to compute the stability & control derivatives of the parametric model and investigate the accuracy of these derivatives. Out of the three modelling options, the wing with a moveable model that has a gap between the sides of the moveable and the wing was useless in VSAERO. The model was unstable due to the presence of the gap. The gap creates a very low pressure locally, which accelerates the surrounding airflow to a ridiculously high value. The difference between the ’normal rotation’ and the ’transition surface’ models is the location of the moveable suction peak. The suction peak of the ’normal rotation’ model is located on the wing in front of the start of the moveable, while the suction peak of the ’transition surface’ is located at the start or slightly behind the start of the moveable. The effect of different locations of the suction peaks is that a suction peak on the wing will generate less induced drag than a suction peak located on the deflected moveable. The results of stability & control derivatives for the ’normal rotation’ and the ’transition surface’ modelling options were quite acceptable with the static derivatives as the most accurate derivatives with an average error of seven percent. The control derivatives, on the other hand, was the least accurate with a mean error of 40 percent. The overall performance of the two moveable modelling options is that they were performing equally well when only the accuracy of the derivatives was observed. The ’transition surface’ modelling option was more accurate in the prediction of lateral stability derivatives and the longitudinal control derivatives than the ’normal rotation’ option, while the ’normal rotation’ option was more accurate in predicting the longitudinal stability derivatives and the lateral control derivatives. But the deflection of the rudder produces side force, yaw moment and roll moment coefficients that are less accurate for the ’normal rotation’ option then for the ’transition surface’ option. The ’transition surface’ modelling option is the better option for modelling the moveable because it produced more accurate aerodynamic results than the ’normal rotation’ modelling option. Both moveable modelling options have its strengths and weakness when concerning the prediction of the stability and control derivatives, and neither of the two was considerably better. Thus based on the accuracy of the aerodynamic characteristics, the ’transition surface’ modelling option is the best option for modelling the moveables of a first order panel method solvers like VSAERO.