PSI - Issue 75

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

340

7

The core principles of the integrated approach are on-the-fly generation and cleanup of analysis prerequisites. Instead of precomputing and storing extensive intermediate datasets — such as stress results across all model regions — only the relevant subsets are computed when needed. This selective computation strategy ensures the solver uses minimal disk space, which is a growing concern in large-scale simulations. The benefits of full integration are especially evident in large modal-based transient dynamic simulations, which typically produce vast amounts of time-domain and modal-space data. Through several key innovations, these jobs are now far more manageable: Management of transformation matrices and modal stresses: Historically, these components represented significant storage burdens. Integrated workflows now allow users to bypass these bottlenecks through advanced solver configurations. Two primary measures illustrate the strategic shift toward efficiency: Use of MLDR (Multi-Level Dynamic Reduction): When using the MLDR solver for eigenmode computation, the need for storing large transformation matrices is eliminated entirely. This contrasts with classic modal solvers. One-part-at-a-time processing: Another space-saving measure is the ability to process the model in individual parts rather than as a whole. This segmented approach significantly reduces the required memory and storage footprint. It is made feasible through the integrated, on-the-fly workflow that allows localized processing without needing to access the full dataset.

Fig. 7. Left model as a whole; right part by part solution.

6. Industrial application example: Truck chassis To demonstrate the industrial applicability and performance benefits of fully integrated fatigue analysis in combination with the newly developed MLDR solver, a comprehensive fatigue life prediction was performed on a full-scale truck chassis model. The analysis focused on the side rails, a critical structural component of the chassis, subjected to a wide range of dynamic loads encountered during vehicle operation. The finite element (FE) model (Fig. 8) used in this study is of significant complexity and realism, representing a state-of-the-art industrial application. The model contains: • 7.9 million nodes • 5.7 million elements • 82,000 fatigue-relevant solid nodes • 54,000 fatigue-relevant shell nodes • 515 structural modes (up to 49 Hz) for the default setup • 174,000 total time steps, across 16 torture tracks representing harsh driving conditions These characteristics reflect a highly detailed model of the truck chassis, pushing the limits of modern fatigue analysis methods in terms of both scale and fidelity.