Cross-sectional optimisation of I-shape pultruded FRP beams using search-adaptive micro-population genetic algorithm

Despite the growing utilisation of pultruded fibre-reinforced polymers (PFRP) profiles in infrastructure applications, there is still a scarcity of design tools with a feasible computational cost. Such shortage deprives these profiles of competing with economic cost and optimised performance against other construction materials. This research investigated the numerical optimisation of cross-sectional geometry and anisotropy ratio of I-shape PFRP beams for the design against local buckling of flange under bending and web under transverse compression. A new search-adaptive micro-population genetic algorithm (SA-µP GA) tool was proposed to achieve accurate results with low computational cost. This tool was verified and compared with other GA codes from the literature. The artificial neural networks (ANNs) machine learning tool was used to maximise the use of data and generate practical and economic design charts considering the varying interactions between the design parameters. This optimisation approach represents an efficient design tool to achieve the design requirements of stability, strength, and serviceability (stiffness) limits with minimum cost and optimal structural performance. The optimisation approach was addressed using four case studies from the literature.