Design optimisation of hollow box pultruded FRP profiles using mixed integer constrained Genetic algorithm

Hollow Pultruded Fibre-Reinforced Polymer (FRP) profiles are increasingly used as structural members in many infrastructure applications. However, there is still a lack of coherent design methodology considering local buckling. This research presents a fast-converging numerical approach combining the Finite Element Modelling (FEM) and the Genetic Algorithm (GA) to design the optimal configuration of the geometry and layup design parameters against local buckling under two separate loading conditions of axial compression and four-point bending. The FEM model was validated experimentally. The mixed-integer constrained optimisation GA code (MI-LXPM) was used to solve the problem. The optimisation objective was to minimise the manufacturing cost per metre of pultrusion while maintaining the same stiffness and strength properties of the control profile. The Kriging model was used to interpolate the design space based on the intermediate optimisation data output and produce a practical design chart linking the profile geometry to the ultimate load capacity. An experimental case study on the design of a hollow rectangular pultruded FRP girder demonstrated the proposed optimisation approach. The new design saved 10.6% of the material cost and enhanced the ultimate strength by 41%.

Automated summary

This research proposes a fast-converging numerical approach that combines Finite Element Modelling (FEM) and Genetic Algorithm (GA) to design the optimal configuration against local buckling. The study validates the approach experimentally and generates a practical design chart linking the profile geometry to the ultimate load capacity.