Cyclic Ratcheting Strain‐Driven Non‐Homogeneous Markov Framework for Probabilistic Fatigue and Remaining Useful Life Prediction of Composite Single‐Lap Adhesive Joints

ABSTRACT The stochastic nature of fatigue damage evolution in composite adhesive joints limits the reliability of conventional deterministic life prediction approaches. This study presents a physically interpretable probabilistic framework based on a cyclic ratcheting strain (CRS)‐driven Nonhomogeneous Markov Chain (NHMC) for predicting fatigue life and remaining useful life (RUL) of composite single‐lap joints (SLJs). CRS is employed as an irreversible, ratcheting‐driven strain‐based damage indicator that directly reflects progressive deformation under cyclic loading, overcoming the limitations of stress‐ or stiffness‐based metrics. Fatigue experiments conducted at multiple load levels were used to calibrate the model parameters through back‐calculation and regression, together with a load‐scaling formulation that captures load‐dependent damage rates while preserving intrinsic transition kinetics. The calibration results reveal a consistent critical CRS threshold of approximately 18%–20% at failure. Model validation demonstrates strong predictive performance, with prediction errors below 3% across all loading conditions. Moreover, more than 80% of the experimental fatigue lives are captured within ±1 standard deviation of the probabilistic predictions, confirming uncertainty‐aware reliability. The results indicate that the critical ratcheting strain acts as an effectively geometry‐insensitive failure metric within the investigated range, consistent with previously reported trends for double‐lap joints. Overall, the proposed framework provides a robust and transferable basis for probabilistic fatigue prognosis and real‐time structural health monitoring, beyond the limitations of traditional S–N‐based approaches.