Review on Methodologies of Fatigue Property Prediction for Carbon Fiber Reinforced Polymer
Aijia Li,Qianzhu Mao,Jinke Li,Youming Li,Xi Li,Jia Huang,Jun Xing,Chao Zhang
DOI: https://doi.org/10.1016/j.compositesb.2024.111659
2024-01-01
Abstract:With the increasing application of CFRP in engineering structures, researchers are devoting more attention to its fatigue issues. Due to the greater complexity and diversity of composite materials compared to metals, the utilization of conventional experimental methods to assess their fatigue performance incurs higher costs. Consequently, enormous efforts have been made to seek accurate and reliable methods for the prediction of fatigue properties of composites in a short time. The present review paper summarizes various fatigue life prediction methods for CFRP composites, including semi-empirical methods, finite element methods, non-destructive testing technology (NDT) based methods and data-driven methods. The advantages, limitations and application scopes of those methods are discussed in details. The semi-empirical methods predict fatigue life expediently and quickly, yet it exhibits limited adaptability to different material types and structural configurations. Finite element methods are applicable for predicting fatigue life in various complex composite engineering structures, but they require extensive experiments for parameter calibration. NDT based methods enable the rapid acquisition of a substantial amount of data relevant to fatigue damage for life prediction; however, accurately correlating NDT data with different types of fatigue damage still needs further investigation. Data-driven methods can integrate extensive historical data for predicting fatigue performance of composite materials; however, effectively filtering and cleansing data related to fatigue remains a challenging task. This review paper aims to provide the most relevant and up-to-date information on the fatigue property prediction methods for CFRP composite, and the potential and development of newly proposed fast prediction methods are also prospected.