Structural Evaluation of Flexible Pavement from FWD Deflections, Deflection Bowl Parameters & Surface Moduli using Machine Learning

Abstract

A pavement’s evaluation is necessary so that the required maintenance and rehabilitation schedule can be made in order to maintain a certain level of serviceability. Numerous structural and nonstructural methods can be employed for determination of structural and functional evaluation of a pavement. Measuring both the structural and functional adequacy of pavement is time and cost intensive. In this Dissertation, a cost-effective approach for Structural evaluation of pavement by the strains developed in the pavement, Horizontal tensile strains at bottom of bituminous layer and Vertical compressive strains at top of subgrade, is presented by modelling them from the FWD (Falling Weight Deflector) Deflections, Deflection Bowl Parameters (DBPs) and Surface Modulus (SM) using various Machine Learning Techniques. Parameter FWD deflections were determined from FWD testing, from them DBPs and Surface Moduli were calculated. Data was collected from a 50km stretch of flexible pavement of a 4-lane National Highway in state of Haryana in India. For modelling the strains, Regression models and Machine Learning approaches, i.e., Artificial Neural Network, Gaussian Process Regression, Support Vector Machine, Regression Trees, etc., were used. The developed models presented good results for both horizontal and vertical strains and can be employed by stakeholders for analysis of pavement life, distress analysis in pavements and further guide the future detailed investigation at the project level and is less resource and time intensive with relatively simpler approach. Applicability of the developed models for preliminary determination of maintenance and rehabilitation schedule of pavement at both Project and Network levels is possible using this.