A stochastic fatigue damage model for composite single lap shear joints based on Markov chains and thermoelastic stress analysis

Abstract

A B S T R A C T Quantitative damage detection in composite single lap joints during fatigue is difficultand not practical with many traditional experimental methods. The objective of this study is to combine a new experimental analysis method with existing stochastic models that is capable of taking the scatter inherent in composite fatigue data into account for better representation of damage evolution in single-lap joints. This study presents a new quantitative approach where thermoelastic stress analysis (TSA), a remote sensing and non-destructive evaluation technique, is used for in situ continuous monitoring for fatigue damage detection of Fibre Reinfoced Polymeric (FRP) S2-glass/E733FR singlelap joints. Image analysis algorithms are developed to capture the debonding fronts, and Markov Chain models (MCMs) are developed to capture the fatigue damage evolution in composite lap-joints based on the experimental analysis. The TSA damage index at the last fatigue cycle is used to calibrate the MCMs, and damage predictions using the MCMs are then examined at different fatigue cycles. Methods to construct a predictive stochastic S-N curve utilizing the MCMs are discussed. The proposed IR-TSA with MCMs is shown to be very effective in predicting the damage evolution and allowed constructing S-N curves that take the variability into account for damage in composite single-lap joints.

Fatigue of short‐natural‐fiber‐reinforced high‐density polyethylene: Stochastic modeling of single‐gear‐tooth bending

“…The Markov chain model used in the present study includes a probability transfer matrix (PTM), shown in Equation 1, that represents a load cycle. The variables p 1 and q 1 are, respectively, the retention probability and the transition probability 25,26,28,39 . A sample can therefore pass from an initial state b 1 to the next state b 2 with probability q 1 but cannot return to a previous state.…”

“…It has been applied to composite materials 25–27 while showing life prediction skills. The models can be made more complex by using a time transformation, 25,28 by optimizing the number of parameters, 29 and/or by calculating them iteratively 30 . All three of these techniques increase the precision of an already well‐fitted model.…”

Fatigue performance of adhesively bonded single lap joints with non‐flat sinusoid interfaces

“…It was almost exactly from the beginning of the test that the crack initiation occurred at both ends of flat joints. According to the literature, crack propagation dominates the fatigue life of bonded SLJs, where the crack is initiated after 15% of the entire fatigue life …”

“…According to the literature, crack propagation dominates the fatigue life of bonded SLJs, where the crack is initiated after 15% of the entire fatigue life. [49][50][51][52][53] After appearing and extending rapidly for a short distance, this crack grew in a stable manner until the final failure of the joint. When around 0.4-0.6 of the fatigue crack propagated along the bond length, sudden fracture occurred in all the specimens.…”

Predicting fatigue damage in composites: A Bayesian framework

“…Markov chains were first applied to fatigue modeling in composites by Rowatt et al [17] as an extension of the pioneering work of Bogdanoff et al [16]. Other relevant examples of Markov chains models for fatigue in composites are found in [18][19][20][21] while for metals in [22][23][24][25][26][27][28]. More applications and theoretical insight about stochastic cumulative damage models are provided in [29][30][31].…”