Design Improvement on Plastic Fuel Tank System with Model Bias Prediction

With the increasing development in automotive industry, finite element (FE) analysis with model bias prediction has been used more and more widely in the fields of chassis design, body weight reduction optimization and some components development, which reduced the development cycles and enhanced analysis accuracy significantly. However, in the simulation process of plastic fuel tank system, there is few study of model validation or verification, which results that non-risky design decisions cannot be enhanced due to too much consuming time. In this study, to correct the discrepancy and uncertainty of the simulated finite element model, Bayesian inference-based method is employed, to quantify model uncertainty and evaluate the simulated results based on collected data from real mechanical tests of plastic fuel tanks and FE simulations under the same boundary conditions. The advantages and disadvantages of the applied method are presented, and the effectiveness of the proposed approach is also demonstrated. It is shown that the accuracy of FE simulations coupled with model bias prediction increases apparently.