PSI - Issue 57

Sven Maier et al. / Procedia Structural Integrity 57 (2024) 731–742

732

2

S. Maier et al. / Structural Integrity Procedia 00 (2023) 000±000

1. Introduction

Vehicle structures and integrated components are subjected to various types of loads during their lifetime. In par ticular, for the fatigue lifetime of automotive components which are mounted to the body-in-white, vibration loads due to external excitation are of major importance. The components themselves are local vibration systems that are influenced by their mass, moment of inertia and mounting attributes. In case the peak of external excitation coincides with the vibration behaviour of the subsystem, resonance could occur and eventually lead to significantly reduced fatigue lifetime. A pronounced mass force excitation can lead not only to a failure of the component itself but often also to damages at the connection points of the chassis. Di ff erent excitation sources can be the reason for vibrations of the automotive components. The vibration of the body-in-white structure is mainly induced by the road unevenness and can be considered as loading with a random characteristic Decker et al. (2018). Components submitted to random vibration loads can be advantageously analysed using a frequency based fatigue calculation approach. The random load is typically represented by a Power spectral density (PSD) Bishop (1999). The fatigue and oscillation behaviour of the components depend on a great number of uncertain influence parameters. Uncertainty can occur e.g. in the load assumptions, material parameters, geometry or modelling parameters. Uncer tainty can generally be divided into two di ff erent types: aleatoric or epistemic. Aleatoric uncertainty is random based and cannot be reduced, while epistemic uncertainty is reducible and based on a lack of knowledge Pelz et al. (2021). In structural durability analysis, this uncertainty is traditionally covered by a deterministic approach using safety factors. These concepts are often based on intuitive approaches or experience Schueller (2007). With the aim of improving numerical prediction accuracy and robustness evaluation, probabilistic concepts are more frequently proposed Daouk et al. (2015). Applications of probability methods for consideration of uncertainty in dynamic vehicle simulations can also be found in the acoustic and crash domain Luegmair and Schmid (2020); Jehle et al. (2022). As mentioned above, the aim is to prevent an overlap of the natural oscillation behaviour of the components and peaks of the external exci tation. Even minor uncertainty or inappropriate modifications by development changes in the characteristic properties of the component or mounting can lead to significant changes in the oscillation behaviour of the system. In order to ensure the structural durability of these subsystems, it is necessary to determine the e ff ect of changes in the relevant parameters by means of a probabilistic concept. This study presents an investigation of the actual scatter of the oscillation behaviour and vibration fatigue of a current mass force excited automotive component mounted to the body-in-white structure by experimental measurements. The experiments are performed using a modal analysis and vibration tests on a shaking table. In addition to recording the deviations induced by uncertainty in the manufacturing and assembly process for the original configuration, the e ff ect of a modification of major influencing parameters (mass, sti ff ness) by two supplementary variants is also investigated. For variant 2 the mass of the component is increased and for variant 3 the sti ff ness of the system is reduced by using a thinner chassis structure. These two variants represent typical modifications of the early development phase e.g. due to requirement changes from other structural dynamic fields, such as acoustics. The entire experimental investigation is accompanied by a numerical simulation. A simulation chain is constructed using commercial CAE programs to consider the scatter of relevant input parameters using Monte-Carlo-Simulation (MCS) and to compare the fidelity of the simulation results with measured input data and unknown scatter bounds, which is usually the case. Finally, the experimental and simulation results are discussed.

Nomenclature

Confidence level

C

Material constant of SN-Curve

C s

E ( D )

Expected damage Eigenfrequency n Slope of SN-Curve

f n

k

Required sample number for Monte-Carlo-Simulation

M

Concentrated mass Number of cycles

m c N a