PSI - Issue 38

Denis Chojnacki et al. / Procedia Structural Integrity 38 (2022) 362–371 Author name / Structural Integrity Procedia 00 (2021) 000 – 000

367

6

approach is proposed (Debarbouillé et al., 2021) based on a twin numerical model of the car (chassis and car body) and a specific Constrained Extended Kalman filter taking into account the nonlinear behavior of the suspension for the data fusion from measurements. Then, with numerical approach based on MBS, we can produce from WFT time histories all the deliverables for chassis components and chassis/Body interfaces required by delivering the appropriate specifications (Maxi, Fatigue, Load spectrum…) 2.4. Proving Grounds used for endurance validation tests As illustrated Fig. 2, the target to cover with RLDA campaigns is the normal Use and some Misuse Levels. All data post-processing methodologies (Fig. 4, Fig. 6 & Fig. 9) set the Target Client (§2.2) that define the Fn level of severity, representative of a severe customer and used both to specify and validate our designed parts. An example of a signal set at that level is obtained by mixing signals measured on a vehicle driving our Proving Grounds (PG) Fig. 7. Some new developments are in progress to consider multiaxial variable amplitude loading for automotive parts fatigue life assessment. This method aims to a more precise comparison between PG and RLDA campaigns (Bellec et al., 2021). This PG tuning methodology won’t be developed in this paper.

Fig. 7. Belchamp Proving Grounds

3. RLDA campaign methodology A historical practice to survey real-life loads was to lend fully instrumented cars to real customers for a duration of a month. While this approach delivered realistic samples of actual car uses, they mostly surveyed home to work travels. Measurement campaigns would deliver low amounts of measured miles despite their large durations. Moreover, a large number of loans were required to distribute properly the variety of client loads, further increasing either instrumentation costs or campaign durations. In the last decades, new sampling approaches were applied to increase the time versus cost efficiency of RLDA. Different selective sources of information are now exploited to gather a richer knowledge of client loads, by handling differently the diversity of drivers and the diversity of their runs. The theoretical developments to manage this distinction are presented in (Baroux et al. (2021)). In this section, we present the approach to acquire and label data on car runs. 3.1. RLDA campaign improvements : On the Field Sampling efficiency Paid mission drivers perform these new measurement campaigns, allowing larger distances in shorter campaign times. Their driving behavior is fixed in advance and compared to real drivers’ curve distribution, thus we have imposed a controlled bias on the driver’s profile. The variety of drivers among our customers is investigated differently. The composition of the mission’s run, the MPF, is designed to meet the whole variety of clients’ trips. We transform a survey on travelled road types into segmented road load data. See Fig. 8 for an example of a recent mission.