Multi-objective robust design optimization of a two-dimensional tri-axial braided hollow pillar using an evolutionary algorithm

Abstract

In this work, multi-objective robust optimization of a two-dimensional tri-axial braided hollow pillar was presented. Two conflicting objectives, weight and cost, were simultaneously minimized under the constraint of mechanical safety requirements based on Probability theory. The material properties were taken into account with three design variables and two uncontrollable variables. The three design variables were the total fiber volume fraction, the relative volume fraction of carbon fiber in the total fiber volume fraction and the braid angle. The two uncontrollable variables were the fiber volume fraction of the carbon fiber tows and of the glass fiber tows. A mesoscopic analytical model of the two-dimensional tri-axial braided composites (2DTBC) was adopted to obtain the necessary mechanical parameters. Appropriate surrogate models were used to investigate the approximate relationship between the mechanical properties and the input variables, which could significantly reduce the calculation cost. The optimization problem was solved with Pareto optimal solutions through a multi-objective evolutionary solver. The results indicate that, in general, the uncertainty in the total fiber volume fraction was the main decisive factor in the robust optimization and mechanical analysis of the monolithic construction.