Aerodynamic Optimisation Based on Stability and Sensitivity Analysis

Abstract

Aerodynamic optimisation has become a valuable tool the the deign of efficient configurations. However, the usage of aerodynamic optimisation to delay the onset of unsteadiness has remained elusive, even for simple flow.Flow stability analysis predicts how small flow perturbations grow or decay with respect to an equilibrium flow solution (base flow), providing information about the onset of physical mechanisms responsible for unsteadiness, which can help to determine means by which to control it. The linear stability community has developed tools to analysis the onset of least stable modes, which have been popularised as adjoint-based sensitivity analysis. Sensitivity analysis identifies the regions of the flow that, if properly modified, lead to the greatest damping (or amplification) of the feature responsible for the instability.In this work, taking advantage of aerodynamic optimisation, stability analysis and sensitivity analysis, an aerodynamic optimisation tool is developed aiming at seeking optimal flow control strategies (both passive and active), based on stability analysis and sensitivity analysis. The main idea of this work is to identify an instability, find the mechanism (mode) that is causing the instability, locate the most sensitive region of the least stable mode, modify the boundary or flow field to manipulate and delay the unsteady onset of this mode. With optimisation methods, it is possible to find a most suitable strategy for the modification to the boundary or the flow field.To achieve this, stability analysis is used to identify the least stable modes, sensitivity analysis is used to locate the region in the flow field where modifications will most affect these modes. Aided with a variety of techniques, optimisation investigation is conducted on a stability problem. In this thesis, geometric parametrisation is used to represent the geometric with a limited number of parameters, a mesh deformation tool based free form deformation (FFD) is developed, avoiding generating large numbers of similar meshes from scratch, a dynamically updated surrogate model is developed to ensure the accuracy of the surrogate model, while reducing the sampling points in the design of experiments, a mode tracking scheme is developed to avoid manual checking and idenfying modes of interest, which reduces computational cost and alleviates manual labour, making the optimisation investigations more feasible.In this thesis, the stability analysis and sensitivity analysis are applied to different problems and identified the modes that are causing instabilies, and the structural sensitivity map corresponding to the modes are obtained, which point out the regions where these modes are most affected. The location of the most sensitive region, which provides guidance of where the deformation of the shape should be vii viii included, and hence supports and guides the geometric parametrisation for an optimisation investigation. Through optimisation, which is aided with a dynamically updated surrogate model, without excessive...