Prediction of critical strains of flexible pavement from traffic speed deflectometer measurements

Abstract

Structure indicators of flexible pavements can be categorized into material moduli, deflections, distresses, or responses. Since field measurements of pavement responses, such as strains and stresses, require embedded sensors, predictions from easily measured surface deflections are desired. This study aims to predict critical strains of flexible pavements from traffic speed deflectometer (TSD) measurements. Semi-analytical finite element method (SAFEM) was developed and validated for simulation of truck-pavement system by comparison with traditional three-dimensional (3-D) FEM. Results indicated that SAFEM saved significant calculation time while ensuring the same accuracy. The efficient advantage makes it possible to develop an extensive databased of TSD measurement considering a wide rangte of pavement layer thickness, material moduli, temperature, and speed. Traditional regression analysis and artificial neural networks (ANN) models were applied to predict critical tensile, shear, and compressive strains from surface deflection parameters. The developed models were further validated with field measurements. The regression models with explicit forms are easier to understand but require asphalt layer thickness as an input for prediction of tensile or shear strains. Nevertheless, the ANN model offers greater accuracy and only requires the inputs of TSD-measured deflection slopes, which is more applicable for pavement management applications.