PSI - Issue 75

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

680

4

1.4. FEA Stochastic Simulation The variability of FEA parameters was firstly investigated by Halfpenny [Halfpenny et al.(2019)] in the seminal paper on stochastic simulations. The Latin hypercube sampling method was chosen as the best compromise between number of simulation runs and the number of variables of the simulation model. Indeed, the Latin Hypercube Sampling (LHS) better explores the parameter space with fewer samples than Monte Carlo, and unlike Design of Experiments (DOE), handles many variables efficiently for parametric analysis. In this study, The FEA simulation regarded pressure loadings affecting the durability of an automotive heat exchanger (an air cooled intercooler). Bench tests on five components showed an excellent level of correlation with the stochastics simulation (in total 29 runs). Probabilistic fatigue simulation, using Monte Carlo with Latin hypercube sampling and factorial sampling with response surfaces, improves on traditional deterministic design by enabling reliability tests and robust design. This paper has paved the way for investigating the application of stochastic simulation for more complex testing scenarios, such as vibration tests. In this current investigation, we use the concept of FEA fatigue and stochastic simulation to predict the correlation of FEA models used for the validation of automotive components undergoing vibration loadings. By tuning the FEA model to specimen with simple geometries (one or two modal shapes), we introduce the variability associated to the simulations are related to three main variables: - The damping (Q factor) - The Basquin Fatigue coefficient (slope of the SN curve) - The intercept of the SN curve (expressed as the ratio between the UTS and the intercept) Moreover, the study paves the way to the continuation of a previous investigation concerning the feasibility and representativeness of FEA simulation on complex vibration signals (sine sweep on random) once they are simplified as purely sine or purely random. 2. Methods and Tools The goal of this study is to perform a statistical analysis on the time to failure of simple geometry specimens undergoing vibration loads by considering: - The results obtained from the stochastic FEA calculations - The time to failure obtained by destructive shaker tests performed on physical parts. 2.1. The Specimen Model A simple model (a cantilever beam with one main resonance in the frequency range of the excitation signal) serves as both physical and virtual prototypes. The virtual specimen has the same geometry as the prototypes of previous study [Yang (2025)]. The physical prototypes are made of an aluminum alloy and the nominal material parameters are listed in the table here below (Figure 1). Note that the authors intended to include as simulation variables, the scatter of the geometrical measurements of the prototypes, such as: i ) the radius of the root; ii ) the depth of the vertical slot, iii ) the thickness of the narrow bridge and iv ) the length of the narrow bridge. A statistical analysis of the measurements has shown that the variability