Please choose your delivery country and your customer group
In this paper, we proposed a variability hierarchy approach to understand the global variability of the booming noise. In this hierarchy, we focused on the variability of the body in white. Some measurements showed a high level of variability on the body in white, which represents the first step of production in the automotive industry. Then, it appeared that a way of reducing the booming noise variability was to understand and predict the variability of the body in white in the development process. For that, a numerical FE model of a body in white was considered to estimate the variability level. First calculations on this model showed that the nominal configuration (i.e. without random parameters) did not correspond to the mean value of the measurements. Then, a numerical software was proposed to introduce random parameters into industrial finite element models. The advantage of the approach is to manage hundreds of parts and parameters simultaneously and run automatically hundreds of calculations in an acceptable CPU time in a development process. The originality of the methodology lays into the stability of the modal basis that is calculated once. Each random parameter only induces local perturbations into this global modal basis. This procedure allows fast calculations and FRF estimations for each iteration instead of recalculating the modal basis. Thus, this methodology is well adapted for the development process. Based on that software, a methodology has been developed to introduce variability both into the different panels of the body in white and the connections between all parts. The introduction of those parameters was driven by the knowledge of the process. By introducing those parameters, variability calculations were launched and the calculated results were very similar to the measured results: both for point inertances and vibration transfer functions, we obtained comparable levels of variability on the body in white model. In conclusion, this methodology is very interesting in order to understand variability propagation mechanisms. It is fully integrated in the development process and directly applicable on real industrial-size finite element models. These results represent the first application for Renault of running a complete variability computation on a project industrial model. Such models can lead into targets adjustments in order to take into account variability as early as possible in the development process of a vehicle. It allows to have an idea of the risky frequency ranges and to forecast mass production problems. Finally, even if other key parameters are still to introduce into the model to improve the accuracy of the model, this article gives first guidelines to predict variability.