Experimentally validated stochastic geometry description for textile composite reinforcements

Abstract The uncertain quality of composites, due to variability in the mechanical response, forces design engineers to employ high safety margins to ensure that the design requirements are met. Especially for textile composites, an improved assessment of the quality of any composite material is achieved by identification and simulation of the inherent uncertainty in the reinforcement geometry. This paper presents such a comprehensive multi-scale strategy to develop realistic stochastic replicas of a composite material, with emphasis on the identification step. First, the scatter in the tow reinforcement is characterised on the short-range (meso-scale) and long-range (macro-scale) from high-resolution images. Next, a probabilistic uncertainty quantification method is proposed to analyse the variability of each path parameter in terms of average trend, standard deviation and correlation information. This set of statistical information is essential to reproduce the random textile geometry in a numerical simulation approach. The multi-scale framework delivers representative models in the WiseTex format and is demonstrated for a carbon-epoxy 2/2 twill woven composite produced by resin transfer moulding.