Low cycle fatigue prediction of short fibre composites at elevated temperatures

Fibrous composite materials are being extensively used not only in the industrial application, but also in the daily life. The specific use of short fibres with thermosetting polymer has broadened its application owing to its economical constituents and simple manufacturing process. However, engineering designers and maintenance engineers have lacked confidence in using short-fibre composite on a long term basis. Hence, the development of fatigue model for short-fibre system is difficult due to the complexity of interaction between fibre ends and matrix. Furthermore, the sensitivity of thermosetting polymers to heat augments the difficulty in damage modelling of short-fibre reinforced composites. The exhaustive literature review on the fatigue modelling of composite materials is conducted, regarding a new finding that Poisson’s ratio plays a significant role in modelling damage in fatigue. The literature review identified the most suitable damage modelling approaches to predict remaining life of short-fibre composites under cyclic loading and elevated temperatures. Two types of modelling approaches are suggested in the present study, viz. phenomenological approach and the surrogate-assisted optimization modelling. First of all, a combined phenomenological mode is proposed using residual strength and stiffness degradation model combining with a temperature model as a function of strength variation. Second, a function approximation is developed in the present study using optimization technique assisted by a surrogate algorithm. Based on a considerable number of experimental data, the surrogate model was trained to define the non-linearity of the function to model the fatigue damage of short-fibre composites at elevated temperatures. Two sets of experiment were conducted with an objective of acquiring the material constants for the phenomenological model to train the data sets for the surrogate model as well as to verify the models proposed. The first set of experiment was carried out with 200 specimens of E-glass Chopped Strand Mat (CSM) / polyester, used with AVE (Advanced Video Extensometer) to measure Poisson’s ratio. E-glass CSM / vinylester, widely applied in the maritime industry due to the superior water resistance, was adopted for the second experiment using the traditional bi-axial strain gauges which can measure the longitudinal and transverse strains at the same time. Very good agreement between theoretical models and experimental data was found. Especially the surrogate model is able to predict the residual stiffness of the specimen very accurately with an average deviation of 3%. Poisson’s ratio is used as an input variable for the training purpose of the surrogate-assisted optimization model. It is proved that Poisson’s ratio is the critical damage variable and the inclusion of Poisson’s ratio has led to a drop in a prediction deviation by more than 1% on average. The design engineers dealing with composites can benefit from the present study to estimate the residual fatigue life of short-fibre composites. The final cost might be reduced at the design stage by adopting a proper design concept and at the maintenance stage by establishing an appropriate maintenance strategy