Prediction of Life Distribution and Design Robustness of Converter Joint Durability Using CAE Techniques

A variety of parameters influence the durability of a converter to pipe joint of an automotive exhaust system. Some of the parameters are design variables and some factors are related to manufacturing. The design parameters include the thickness of the components, diameter of the pipe, sleeve length of the cone etc. While the variables like the weld penetration and the fit-up of the joint are related to manufacturing. Traditional durability simulations utilizing computer aided engineering (CAE) methods are conducted using nominal values of the design and manufacturing variables. In reality scatter and randomness in parameters are present due to the tolerance in components and limitations of the manufacturing process. In this paper a CAE based stochastic approach to determine the life distribution for a converter joint of an automotive exhaust system is presented. During the first step of the study, design of experiments (DOE) using finite element analysis (FEA) were conducted to identify the critical parameters and interactions. Results from the DOE study were utilized to generate the response surface model to predict the durability of the joint. Finally, the response surface model was used to conduct Monte Carlo simulation and six sigma analysis to understand the robustness and reliability of the design. Correlation between the life of the converter joint predicted from this approach and life distribution obtained from physical tests are also presented. The main advantage of this approach is that the joint life is predicted using probabilistic distributions instead of the mean life.