Many organizations are increasingly relying on design simulation rather than expensive and time-consuming prototype testing for product evaluation. However, uncertainties in analytical and computational methods need to be understood in order to improve confidence in their use, and models need to be validated. This paper presents a case study of a MacPherson strut automotive suspension analysis, and evaluates the uncertainties in the modelling of this complex dynamic problem using a simplified analytical model and a complex computational model. In both cases, variability in design variables is characterized using probabilistic design methods. As a first step, the model variables are described by assumed datasets, which are collated from several sources such as tolerances specified in drawings, expert opinion, published data, etc. Measurement of the properties of the suspension system components is then performed (spring stiffness, damping coefficient, etc.), and the statistical parameters so obtained are used in probabilistic calculations for specified time sequences from measured test track road load data. The results are used to accumulate evidence of uncertainties in analytical and computational methods, to correlate predicted results to experimental data for vehicle chassis top mount force, and to derive sensitivity measures. A response surface function is approximated which is useful for parametric studies for new variants of the system studied. Sources of uncertainty in this case study and methods for improving the correlations are then suggested.
|Number of pages
|Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
|Published - 2005