Influences of fibre shape on the transverse modulus of unidirectional fibre reinforced composites using finite element and machine learning methods
Haowei Huang, S.A. Hadigheh, Keyvan Aghabalaei Baghaei
Abstract
Unidirectional fibre reinforced polymer (FRP) composites are becoming increasingly popular in industries such as aerospace and automotive. Although longitude modulus predictions using traditional methods are accurate, the prediction of transverse modulus using current methods returns scattered results. The lack of precise information on the transverse modulus has limited the modelling performance for practical applications. Moreover, with the advances in manufacturing technologies, more irregular shaped fibres are on the market while the traditional analytical methods often ignore the fibre shapes. In this work, finite element method is used to create micromechanical unit cells and to compute transverse modulus. A python framework is established to generate and store data based on the finite element analysis. The data is then passed on to train the machine learning models which provide an efficient way of predicting the transverse modulus of unidirectional composites with the consideration of fibre shapes. Fibre shapes are found critical in these predictions. Machine learning method is proven to be a reliable tool in this task. Based on the best performance machine learning model, the importance of each parameter for the prediction are analysed using out of bag error method, which also quantified the importance of fibre shapes.