This paper presents a statistical method to predict NCAP (New Car Assessment Program) star ratings at early stages of the vehicle development process when only limited vehicle information, such as weight and a few basic dimensions, is available. The heart of this method is a probabilistic model, in which the probability distribution of NCAP star ratings, conditional on vehicle characteristics, is estimated using residual analysis. Seventy-nine NCAP tests from model year 1999 to model year 2002 were collected from NHTSA public database and used for the analysis. Given a vehicle’s weight and basic dimensions, the estimated probabilities reflect variations among similar existing vehicles due to detailed vehicle structure design, interior geometry layout, restraint system design, and the inherent variability of hardware and crash testing. The statistical method developed in this study provides a fast, probabilistic assessment of a concept vehicle’s potential NCAP performance.

This content is only available via PDF.
You do not currently have access to this content.