Predicting the residual fatigue life of a cargo hull tank using a deep-learning technique

Abstract

ABSTRACT Unexpected loads are commonly applied in cargo ships when workers fail to comply with applicable regulations while loading and unloading cargo. This important factor can shorten the design life of a ship by increasing the possibility of structural damage. One way to improve the reliability of the fatigue life of a ship is to check for fatigue damage by observing real-time stress response depending on marine environmental conditions. Modern stress monitoring systems are installed in critical spots to capture stress signals and used to minimize the risk of crack occurrence. In the present study, a hull residual life prediction procedure was introduced by integrating traditional fatigue analysis and the stress response obtained from strain gauges installed in the hull. A three-cargo-hold oil tanker model was used to calculate the residual fatigue life using Harmonized Common Structure Rules (CSR-H). For evaluation, data obtained from the strain gauges and predicted data using a deep-learning technique were compared. According to the results, this process could be applicable to predicting residual fatigue life during a ship’s operation in real time if a long period of statistical data is available. The developed techniques can help analyzing the fatigue integrity of operating ships and ship equipment.