Development of Pavement Condition and Roughness Evaluation Models for Asphalt Concrete Pavements

Abstract

In order to maintain the functional and operational condition of the road, the deterioration in the pavement's condition must be properly evaluated and dealt with at the earliest. To suggest maintenance activities, DoR presently employs the SDI measure in general which is subjective and inadequate pavement performance measure. Due to the incorporation of all forms of distresses, their density, and their severity, PCI is thought to be one of the most comprehensive and widely recognized global method of pavement condition evaluation and is the focus of the study. The IRI, an indicator of perceived road roughness and ride quality is also in practice by DoR in some major national projects as evaluation measure. Therefore, an IRI model and its correlation to PCI is also a focus of this study. The PCI and IRI pavement evaluation models is developed using regression analysis and ANN. The distress data is collected and quantified as per ASTM 6433 whereas, the IRI data is collected using RoadRoid application after validation. A total of 503 and 468 data is collected and used for the evaluation of PCI and IRI respectively based on the distresses. The regression models yielded a R2 of 0.600 and 0.621 for PCI and 0.599 and 0.614 for IRI for grouping set 1 and 2 respectively showing moderate fit of the data. In order to further improve the results, ANN model is developed using python 3.9 for PCI and IRI evaluation. Based on the ANN output, R2 of 0.857, 0.715 and 0.747 for training, validation and testing for grouping set 1 and 0.852, 0.810 and 0.670 for grouping set 2 is obtained which indicating improvement in result when comparing to the regression models for PCI whereas the R2 value in training, validation and testing 0.559, 0.518 and 0.536 for grouping set 1 and 0.699, 0.597 and 0.575 for set 2 during ANN for IRI evaluation. Similarly from sensitivity analysis, high severity potholes are found to be most significant parameter of both PCI and IRI. Finally, the PCI-IRI relationship was established in excel showed negative correlation & yielded maximum R2 of 0.7858 .