PSI - Issue 75

Marco Bonato et al. / Procedia Structural Integrity 75 (2025) 677–690

679

3

/ Structural Integrity Procedia (2025)

PSD

Power Spectrum Density

PV Product Validation RMS Root Mean Square SN

Stress vs Cycles Fatigue Life Curve

SSoR Sine Sweep on Random Signal UTS Ultimate Tensile Strength

1.2. FEA Fatigue Simulations The potential of FEA simulation for fatigue life predictions has developed in the last years thanks to the implementation of dedicated software solutions. Such tools permit estimating the nominal time to failure of the component by assuming its underlying material characteristics: the damping, the stiffness, the effect of temperature and surface treatment, and most importantly, the material cards (UTS, yield strength, Basquin coefficients, etc.) Simulation fatigue analysis of automotive components, especially under vibration loadings, offers great potential but faces challenges. Component complexity, from simple brackets to complex systems like cooling modules, introduces non-linearities affecting accuracy. While FEA can predict primary frequencies and local stress, fatigue simulation is hampered by the difficulty in obtaining component-level fatigue life curves due to high scatter and cost of vibration tests. Commercial software uses specimen-level material data, yielding promising comparative results, but access to specific alloy data is limited. Correlating simulated fatigue life with physical tests requires extensive strain and acceleration measurements. Despite these challenges, FEA fatigue analysis provides valuable insights into product robustness, and understanding key parameters is crucial for reliable results and validation. FEA Fatigue in Vibration tests Previous studies on FEA methods applied to fatigue life prediction and vibration stress applied to the validation of an engine cooling module had shown how the main parameters of the model highly influence the nominal results. The extensive work, which involved roughly 30 000 simulation runs as a combination of the variables, was not correlated with real material data available for the products [Czerlunczakiewicz et al. (2023)]. In the following paper, Bonato [Bonato et al. (2023)] highlighted the consequence of performing fatigue simulation using complex vibrational signals. In the study, a validation vibration test provided by two European carmakers assumed the nature of a sine sweep on random and show-plus-random type, applied to the design validation of two automotive components (a high voltage heater and an alternator). The paper shows how such signals can be transformed into purely random or sinusoidal singles, based on fatigue equivalence criteria. However, the choice of the fatigue damage variables, i.e. the damping factor Q and the Basquin fatigue exponent plays a major role in the representativeness of the results. In this framework, Yang [Yang et al. (2025)] carried out a more thorough investigation by calculating the vibration fatigue life of the specimen with simple geometry (having one or two modal resonances). The sine-on-random signal is computed for fatigue life simulation by adopting the original signal via transient analysis, and after its transformation into a purely sinusoidal or random signal. In previous works, only nominal results were obtained. The variability of the FEA models and the material data were kept fixed. In reality it is known that each product (from complex heat exchangers to simple specimens) shows an intrinsic variability associated with its time to failure. The source of this scatter might have originated from the variability in geometries, testing conditions, material etc. An improved simulation test plan is needed to include those parameters affecting the variability of the FEA simulation, and to further correlate the model to experimental results from a statistical point of view (fatigue life distribution). The simulation process moves from a “deterministic” model to a “stochastic” one, which is able to include all the uncertainties of the input parameters. 1.3.