On the Development of Topology Optimisation Algorithms for Aerospace Applications

Structural topology optimisation has been identified as the most challenging and economically rewarding task in structural design, as it is the least restricted by predefined design variables. The main advantage of topology optimisation over other types of structural optimisation is that no specified initial topology needs to be presumed a priori. Nevertheless, topology optimisation does not have a strong history in the least-weight design of aircraft and aerospace structures. The aim of this thesis is to develop topology optimisation methods for use in the aerospace industry through three main contributions. Firstly, algorithmic development is considered. The main driver for aerospace design is weight minimisation, often resulting in thin frame-like structures that are inherently prone to buckling and dynamic instabilities. Despite this, topology optimisation applied to buckling and dynamic stability is seldom seen. This is likely due to the computational burden and numerical instabilities present. In this thesis novel sensitivity analyses and constraints are implemented in current algorithms to circumvent such issues. Secondly, multi-physics problems are solved. A likely factor that has impeded the use of topology optimisation in the aerospace industry is the complex load environment present. For example, a wing represents a classic multi-physics problem where structural and aerodynamic design requirements are conflicting and intimately related. In this thesis, topology optimisation algorithms that can solve multi-physics problems are developed. Finally, multi-objective topology optimisation is investigated. Aerospace design problems typically involve several objectives, most of which counteract each other. Therefore, it is desirable to present the designer with a set of Pareto optimal solutions that capture the trade-off between these objectives, known as a smart Pareto set. In this thesis the Smart-Normal Constraints Bi-directional Evolutionary Structural Optimisation (SNC-BESO) algorithm is developed and shown to produce smart Pareto sets