Gaussian process homogenisation and design of technical textiles

Abstract

A method for computational homogenisation of thin membranes is proposed to assist in the mechanical analysis of technical textiles. Conventional computational homogenisation schemes have a limited scope in analysing the complex nonlinear behaviour of textiles where the interlooped or interlaced fibres undergo large deformations. Also, the higher computational cost of the existing homogenisation schemes inspires the inception of a data-driven multiscale computational homogenisation scheme. In this paper, the efficient integration of statistical learning in computational homogenisation is proved not only to result in the efficient mechanical analysis but also to open avenues in material design of knitted and woven technical textiles.