This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

The fatigue behaviour of 2024-T351 aluminium alloy, commonly used in aeronautical structures due to its favourable strength-to-weight ratio, is analysed under variable-amplitude axial loading. This study leverages advanced Continuum Damage Mechanics (CDM) simulations implemented through a UMAT (User-Material) subroutine in Abaqus, alongside a cutting-edge simulation environment governed by a UEXTERNALDB subroutine. The UEXTERNALDB subroutine incorporates two essential functions that enhance the simulation'™s efficiency and accuracy: 1. Loading Cycle Control: The subroutine reads a binary input file containing the loading sequence during the initialization phase of the analysis, enabling precise control over each loading cycle applied in the simulation. By calculating the necessary information to assess fatigue damage per loading cycle based on interpolated stress states within each analysis step, UEXTERNALDB minimizes the computational resources required. This cycle-by-cycle optimization notably reduces the total simulation time, making it particularly suitable for long and complex loading sequences, unlike conventional methods that require stepwise evaluation. 2. Enhanced Material Calibration for CDM: UEXTERNALDB accesses an external database that contains a neural network specifically trained to model the fatigue behaviour of the 2024-T351 alloy under complex stress states. This neural network is subsequently integrated with the UMAT subroutine at the integration point level, providing a sophisticated material model capable of accurately predicting fatigue under variable loading conditions. This approach is particularly beneficial for evaluating complex stress scenarios, including high stress ratios, which are typical in aeronautical applications. The simulated results derived from this CDM-based virtual testing framework will be validated against benchmark physical test results to assess accuracy. The feasibility of applying this novel simulation environment to predict fatigue under variable loading sequences will be critically evaluated to support its potential use in the aeronautical industry. In conclusion, this simulation framework could represent a significant advancement in fatigue prediction, reducing costs and time associated with physical testing, thereby supporting safer, more efficient aircraft design.