Enhancing the real-time prediction of fatigue damage in offshore structures: A novel method integrating frequency-domain approaches

Abstract

Offshore structures are continuously subjected to wideband random loads, rendering the real-time prediction and assessment of their fatigue life a challenge. Thus far, a high-confidence method for predicting fatigue damage, referred to as rain-flow counting (RFC), has failed to perform real-time damage prediction in actual offshore operation scenarios owing to its large computing time and input data size. In this study, an improved fatigue damage estimation method is proposed for real-time prediction in the frequency domain by constructing standard stress spectrum to determine the distribution parameters, weight factors, and distribution correction coefficients associated with the spectrum width parameters. A new assembly distribution is applied using a weighted combination of exponential, half-Gaussian, and two double-parameter Weibull distributions. Based on the established standard stress spectra, the proposed method was compared with RFC to preliminarily verify its prediction accuracy. Subsequently, the proposed method was applied to actual ship monitoring and compared with nine typical prediction methods; the results confirmed the satisfactory prediction accuracy and stability of the developed model. In conclusion, the high-precision predictions made by the developed method closely match the RFC results, significantly improving the fatigue damage prediction over various bandwidth processes.