Ordinal Logistic Regression Model for Predicting AC Overlay Cracking

Abstract

Predicting performance of asphalt concrete (AC) overlay is important for both pavement design and management. Based on 328 AC overlay sections from the long-term pavement performance (LTPP) program, ordinal logistic regression models were developed in this study to predict the probability of severity levels for alligator cracking. It was found that the alligator cracking is significantly affected by alligator cracking of the existing pavement, thickness of overlay, thickness of the existing AC, age of pavements after resurfacing, truck volume, freeze-thaw cycles, and the amount of precipitation per wet day. In addition, it was found that the use of recycled asphalt pavement in the amount specified in the SPS 3 experiment and preoverlay treatment significantly affect overlay cracking. Both statistical conformance tests and empirical verification using independent data indicate that the developed models are reasonably accurate. The model can generate the probability of a pavement staying at a certain distress level and the odds ratio, which enables highway agencies not only predict probability of cracking but also assess the confidence of making such predictions.