PROBABILISTIC MODELLING USING EXPONENTIAL DISTRIBUTION FOR ENHANCED FATIGUE BEHAVIOUR UNDERSTANDING IN PLAIN CONCRETE

This paper presents an in-depth study combining exponential probabilistic distribution and finite element analysis (FEA) to accurately predict the fatigue life of plain concrete subjected to cyclic loading. Existing fatigue life prediction models do not adequately address the variability and uncertainties inherent in concrete properties under cyclic loading. This is due to the complex nature of concrete and the time-consuming and challenging laboratory fatigue testing. Hence, the finite element method (FEM) was adopted as an alternative method in fatigue testing and the implementation of the probabilistic approach can account for these uncertainties and enhance the reliability of predictions. The research follows three primary stages: constructing S-N (stress-number of cycles) curves derived from experimental data across varying loading conditions, developing and validating a FEM in ABAQUS, and refining these S-N curves using exponential probabilistic techniques to improve predictive accuracy. Initial S-N curves from the experimental data confirmed the expected inverse relationship between applied stress levels and cycles to failure, consistent with classical fatigue behaviour in plain concrete. The FEM, validated using an optimal mesh size of 12 mm, achieved a high correlation with experimental data, evidenced by a minimal percentage error of just 0.80%, indicating high model fidelity. The validated FEM produced additional data points, complementing the experimental dataset. These were subsequently integrated into a probabilistic model using an exponential distribution, which enhanced the statistical representation of fatigue life predictions. This probabilistic model, driven by the exponential distribution, enabled the estimation of fatigue life across multiple confidence levels, with failure probabilities ranging from 99% to 20%, offering insights into the reliability of concrete structures under cyclic loads. The analysis further demonstrated that higher stress levels directly correlate with shortened fatigue lives and elevated rate parameters, indicating an accelerated rate of material degradation and highlighting the need for conservative stress design in concrete structures. This study underscores the essential role of integrating experimental data, FEM, and probabilistic analysis in advancing fatigue prediction methodologies by offering a comprehensive framework for improving reliability and structural safety, particularly under variable cyclic loading conditions.