Multi-objective design of a composite cylinder considering buckling and creep

Abstract

This thesis seeks to create a multi-objective design optimisation (MODO) method to optimise the buckling resistance and the long-term creep behaviour of three-ply filament-wound composite cylinders, simultaneously. The cylinder design space comprises of a near-hoop ply located at either the outer, the middle or inner through-thickness position, and a pair of 2 helical angle plies that can be designed within the manufacturing constraints where the parameterisation is directly connected to the stiffness distribution and selection of the fibre angle. A finite element (FE) model is developed, and an artificial neural network (ANN) surrogate is learnt to approximate the structural responses. This surrogate model is then integrated into a multi-objective genetic algorithm (NSGA-II) optimisation loop which seeks a Pareto optimal set of combinations of ply angles that maximise buckling resistance and minimise creep deformation. The optimal designs consistently feature three distinct ply roles; a near hoop ply for circumferential stiffness, a near axial ply that directly counters matrix creep and intermediate ply that balances buckling load. This ply combination served better to preserve the buckling capacity and reduce creep of the composite cylinder.