Abstract
The automotive industry has a vast amount of accumulated experience of assessing vehicle handling through subjective assessments using skilled test drivers and objective measurements carried out on test tracks often utilising international standard techniques. The third technique for understanding vehicle handling is to predict behaviour at the design stage, typically sing sophisticated multibody dynamics computer packages. The focus of this paper is on the inter-relationships between these three aspects of achieving good vehicle handling behaviour; subjective, measured and predicted.
The key issue in these inter-relationships is understanding the link between subjective evaluation and objective measures, whether they are obtained experimentally or predicted. This was tackled by devising a series of tests based on two nominally identical vehicles, one a control vehicle and one for which 16 different set-ups could be arranged. Subjective experiments were based on steady state, J turn, double lane change, pseudo random steer and impulse tests. The subjective / objective comparisons were then explored using multiple regression techniques.
Given the context that previous work in this area has historically always proved challenging, some of the results have begun to show important trends and have confirmed the usefulness of the new methodology developed. For example, for individual drivers it was possible to highlight questions which correlated very well with objective data. However, of much more importance was whether similar correlations could be achieved over the group of drivers; in this case, a subset of objective responses e.g.lateral acceleration gain, yaw rate gain, steering torque during J turns and sideslip gradient from the steady state tests were highlighted as fitting well with the subjective views of all drivers.
The paper argues in further detail the potential benefits of this new methodological approach to an age-old problem in vehicle dynamics.