Development of Prediction Models for Performance of Flexible Pavements in Kansas with Emphasis on the Effects of Subgrade and Unbound Layers

Abstract

This study resulted from the need for better consideration of subgrade and unbound layers on the performance of flexible pavements in Kansas. Thus, the main objective was to develop pavement performance prediction models with emphasis on the effects of subgrade and unbound layers. To this end, pavement distress data, which were collected over several years across the state of Kansas, including rutting, fatigue cracking, transverse cracking, roughness and core analysis, served as the input data into statistical models. The effects of subgrade and unbound layers were represented by the corresponding results of dynamic cone penetrometer (DCP) tests and thickness of the unbound layer. In addition, traffic volume was represented by average annual daily truck traffic (AADTT). Multiple statistical analyses identified positive correlations of dynamic cone penetration index (DPI) and rate of total rutting, and DPI and percent of good core. Negative correlation was discovered between DPI and fatigue cracking code one, and DPI and percent of poor core. AADTT was positively correlated with transverse cracking codes one and two while it had no correlation with transverse cracking code zero. Thickness of the unbound layer was negatively correlated with pavement roughness and percent of poor core, while it was positively correlated with the percent of good core. Finally, the recommendation for minimum acceptable value of California bearing ratio (CBR) was provided based on the correlation between DPI and rate of change of rutting code. The recommendation enables the selection of a CBR value based on the number of years required for unit increase in the rutting code.