Fatigue tests for automotive design: optimization of the test protocol and improvement of the fatigue strength parameters estimation

摘要

Despite of continuous improvement of numerical computations, experimental fatigue tests are still of prior concern for high safety level parts, in particular for the automotive industry. Destructive fatigue tests are required in order to prove to authorities an effective fatigue design by highlighting the weakest link and its failure mode, and possibly taking account for the manufacturing process influence. Actually, experimental tests need for expensive means (i.e. rigs and specimens) and fatigue phenomenon is expected to take long time to get a crack issue. Thus, cost and delay time reduction always calls for more effective methods. Within this framework, we address here some recent improvements dealing with, respectively: a) the optimization of the fatigue test protocol, i.e. how the load time history is applied to the specimen in order to minimize the overall fatigue test duration; b) the reliability of the fatigue strength estimation, i.e. how the experimental data are considered in a reliable statistical model. The first point let us focus on a so-called Locati fatigue test protocol, whereas the second point let us make use of Maximum Likelihood and Baycsian techniques. These improvements are currently and successfully applied to chassis system parts at PSA Peugeot Citroen.