Identifying the Most Effective Data Processing for Fatigue Delamination Growth in FRPs

Abstract

This paper investigates the fatigue-induced delamination growth in carbon fibre-reinforced polymer (CFRP), considering different fibre orientation combinations. The study explores the application of Artificial Neural Networks (ANN) in the simulation of fatigue delamination behaviour to reduce the number of experimental tests required for fatigue evaluation and eventual certification. The research aims to evaluate the effectiveness of ANN at different stages of data processing, including raw data simulation and final curve estimation. The results show that applying ANN at the raw data stage provides flexibility in modelling, with error < 10%. In addition, when ANN is applied directly to the final Paris curve, it minimises errors and increases reliability, allowing for a more cost-effective fatigue evaluation process. The study highlights the importance of the data processing stages in determining the accuracy of fatigue delamination predictions with AI modelling, thus informing strategies for efficient fatigue evaluation of CFRP components in structural applications.