Deterioration Modelling of Visual Condition Index for Calgary’s Highway Pavement Management System

The project is to evaluate pavement performance measures and to provide a deterioration model to improve the current asset management practice of the City of Calgary. The performance evaluation compares the pavement surface condition measurement methods used by the City and the Pavement Condition Index (PCI) approach recommended by American Society for Testing and Materials (ASTM) D6433. It includes two steps: re-calculating from raw distress data and a ground truth validation survey using images from Google Street View. The study concludes that the Highway Pavement Management Application (HPMA) Visual Condition Index (VCI) method outperforms the ASTM PCI method. Therefore, VCI decrements are the outcome of the deterioration model developed by a machine learning approach, decision tree regression. The robustness of the proposed model is validated by examining overfitting and the selected variable's consistency by the k-fold cross-validation method. The result of the project assured that the current practice of the HPMA system is justifiable, and the implementation of the deterioration model contributes to future practice by providing more accurate information on monitoring the network's life expectations and vulnerable communities.