Reducing the uncertainties in the achievable vehicle performance targets through design optimization

In this paper an alternative method is proposed for defining the suspension performance targets through the use of full-vehicle modelling consisting of a ride model and a handling model. These models are derived with the use of a non-linear damper, suspension kinematic characteristics and basic vehicle dimensions. The vehicle performances can be explored using the design-of-experiments method. The non-sorting method is then employed to sort for non-dominated solutions, where these samples represent the Pareto front of the vehicle performances in ride comfort and handling. The k-means clustering method is used to classify further the solution into different unique optimum characteristics. The expectation–maximization algorithm is developed to compute the allowable variance of design parameters required to achieve the specific optimum design targets. This method can be a very useful tool in the earliest design stages where vehicle data are inadequate. This methodology potentially reduces the uncertainty in the achievable vehicle performance targets by allowing engineers to compare the optimum limit of the suspension with those of benchmark vehicles in the early suspension design and development process