Alternative Modeling Framework for Pavement Serviceability Analysis

Abstract

This paper presents an alternative modeling framework for serviceability analysis that is conceptually based on fuzzy set theory and uses discriminant analysis as the estimation tool. The proposed approach respects the ordinal characteristics of serviceability ratings because it does not assume that the ratings are a ratio scale, which is implicitly required when using regression analyses. The practicality of the proposed approach is tested in a case study, which indicates that the proposed approach provides a consistent classification of the pavement sections, though it underestimates pavement serviceability. Since the serviceability ratings are not needed for the estimation of the model (ratings are only required for validation), assembling a panel of users is not required to create the calibration data set. This feature may produce significant savings in the data collection process.