PSI - Issue 75

Laurent Dastugue et al. / Procedia Structural Integrity 75 (2025) 334–343 Laurent Dastugue et al. / Structural Integrity Procedia (2025)

341

8

Fig. 8. Location of side rails in chassis and damage result.

Simulations were conducted on a high-performance computing setup with 28 cores and 175 GiB of memory. Figure 9 presents a comparative overview of runtime performance, disk space usage, and disk I/O load for both the classic solver and the MLDR solver across various modal spaces.

Fig. 9. Run times, disk space consumption and disk I/O for truck chassis fatigue analysis (including vibration analysis – modes – and modal time history analysis). The simulations evaluated three modal space configurations: • Default: 515 modes (up to 49 Hz) • Extended: 835 modes (up to 100 Hz) • Further Extended: 1539 modes (up to 200 Hz) The ability to extend the modal space significantly enhances the accuracy of fatigue predictions, particularly for high-frequency phenomena, which are increasingly relevant in modern lightweight and electrified vehicle designs. Key Findings: • Integrated fatigue analysis with MLDR delivers a breakthrough in runtime efficiency: Fatigue, vibration, and time history analysis phases saw drastically reduced runtimes compared to the classic solver. • Scalable with mode count: fatigue analysis in combination with MLDR maintained minimal increases in runtime and disk usage, even as the number of modes increased by a factor of three. • Sustainable performance under resource constraints: Simulations with extended modal spaces remained within typical disk limits — a feat previously unattainable without compromising model detail or analysis fidelity. The ability to incorporate higher-frequency modes into fatigue analysis without incurring prohibitive computational costs marks a paradigm shift. With the fully integrated fatigue solver, it is now feasible to: • Perform high-resolution durability analysis for complete vehicle systems,