Prediction model of the damage state for semi-rigid base layer in asphalt pavement based on pavement performance index

It is difficult to evaluate the damage state of the base layer through direction observation, and conventional methods, like Falling Weight Deflectometer (FWD) testing, have not yet been widely implemented in practice. Therefore, developing a practical method for evaluating the damage state of the base layer in asphalt pavement remains a challenge. This paper aims to propose a prediction model for evaluating the damage state of semi-rigid base layers based on pavement performance index (PPI). First, the detection of five commonly-used PPIs, including Pavement Condition Index, Roughness Quality Index, Rutting Depth Index, Skidding Resistance Index, and Pavement Structure Strength Index, FWD testing and core sampling were conducted on the in-service pavement, with the FWD data serving as the basis for determining the damage state ( D FWD ). Then, the relationship between PPIs and D FWD was analyzed by linear regression for the establishment of the model. Finally, the predicted damage state ( D PPI ) was compared with the actual damage state observed in field samples to validate model’s accuracy, together with three conventional indices. The results indicate that Pavement Condition Index and Pavement Structure Strength Index are recommended as input variables of the prediction model, achieving a correlation coefficient of 0.908. Based on the observed damage states, the comparison between D PPI and conventional indices reveals that D PPI provides more accurate assessments, particularly in identifying severe and bad conditions, demonstrating its superior applicability. The findings of this research are expected to provide an effective method for evaluating the damage state of semi-rigid base layers. • Integrate PPI detection, FWD testing and core sampling. • Establish a prediction model between PPI and base layer’s damage state. • High assessment accuracy is validated by the observed damage state of samples. • The proposed method is advanced in identifying different damage states.