An investigation and optimisation study on the structural performance of filament wound GRP pressure vessels

Abstract

Filament wound composite pressure vessels and pipes are tailored to cater for specific needs and the design phase is usually the crucial factor for the successful implementation of these novel materials. According to the European standard EN 13923:2005, the ultimate pressure of glass-fibre reinforced polymer (GRP) filament wound pressure vessels is defined by the first occurrence of failure. There is no definitive way how to detect the onset of failure during a pressurisation test therefore, a methodology involving the monitoring of the mid-span expansion, backed up by strain measurements, to capture both the first ply failure and the successive behaviour leading to the ultimate failure of the vessels, was proposed. The methodology was successfully implemented in a set of pressurisation tests performed on e-glass reinforced polyester filament wound pressure vessels. From an analytical point of view, it was shown that utilising the classical laminate theory (CLT), together with the stress resultants as indicated by the standard EN 13923:2005, is not adequate to predict the stress-state of unsymmetric and/or unbalanced cylindrical laminates. Ideally, the use of finite elements software or closed solutions, where available, should be encouraged. Furthermore, modelling the actual failure progression of a filament wound cylindrical pipes post the FPF load utilising a progressive failure analysis (PFA) based on a failure mode dependent sudden degradation methodology, is not sufficient. A novel method to overcome this shortcoming was proposed where initial imperfections are introduced systemically in the geometry or the material properties based on the void fractions of the specimens used in the material characterisation tests. The novel method creates a link between the manufacturing induced inhomogeneities and the numerical model, thus improving significantly the predicted behaviour both quantitatively and qualitatively. A study on the implementation of the novel Big Bang – Big Crunch (BB-BC) optimisation procedure to optimise filament wound pressure vessels composed of different types of layers is presented. The BB-BC methodology is able to find the global extremum although care should be taken when choosing the population size and the β parameter. Higher population sizes increase the algorithm robustness in finding the global extremum, whilst favouring the fittest solution while creating subsequent populations is essential for objective functions with local extrema.