Today in the automotive industry, there is a continual reduction in available development time. There is also an urgent need to reduce cost and weight, to adapt to customer and legislation which drives to an increase in design complexity. These challenges are sometimes made harder by the late availability of hardware and this creates the need to extend and continually improve the established CAE methods which are used to develop automotive parts. This holds especially true in the field of exhaust systems and their components, which experience loads from various sources like temperature, engine or road.
In the field of road excitation the use of dynamic transient simulation and subsequent damage calculation is state of the art in terms of simulations methodology. Nevertheless the problems of this method are the long calculation times and large scratch data requirements as well as the necessary precise knowledge of the test track profile, which is generally not available in early project phases or sometimes cannot be communicated by the OEM to suppliers. This lack of information results, in many cases, in the usage of simplistic, experience based static approaches that do not display the real dynamic behaviour and lead to over engineering.
Frequency domain methods for fatigue analysis and/or for general random response analysis have experienced resurgence over the last years due to improved technology and better computational processes. In previous papers, frequency based fatigue methods have been successfully applied displaying good correlation with the time based approach. This paper will propose a possible alternative to the use of the transient method or other, simplified static approaches for the development of complete passenger car exhaust systems using the capabilities of a frequency based fatigue method in a realistic scenario, i.e. under multi-position and multi-directional excitation.
The key steps of such a frequency domain analysis will be discussed in the following. The first step relates to the conversion of the multi-channel load time histories followed by the second step related to how boundary conditions are placed. Then, the frequency domain results are compared in detail with traditional time based results. At the end of the comparison, a detailed overview of the calculation time and disk space needed for both methods will be shown in order to demonstrate the real advantage in the use of the frequency base fatigue method.
As a conclusion and outlook, the huge opportunities to improve the cooperation between OEM and suppliers without compromising any precise information about track profiles using the power spectral density PSD of the signals will be illustrated.