Prediction of Flexible Pavement Layer Moduli from Dynaflect and FWD Deflections

Abstract

Analytical investigations of Dynaflect and falling weight deflectometer (FWD) were performed using a linear elastic multilayer computer program (BISAR) to generate deflections for different combinations of layer thicknesses and moduli. The generated data base was used to develop layer moduli prediction equations for each NDT device. Prediction equations from multiple linear regression analysis of the FWD data were dependent upon all sensor positions except for the subgrade modulus prediction, which required only the use of either the sixth or seventh sensor. However, it was found that the Dynaflect with modified sensor locations provided separation of deflection response between the upper pavement layers (asphalt concrete and granular base), the subbase, and the subgrade. Although the Dynaflect prediction equations were reasonably accurate on the basis of the analytical evaluation, they were considered too complex for practical use. NDT data were collected on flexible pavements at sites exhibiting a wide range in deflection response. The standard sensor positions were used for both FWD and Dynaflect testing of the pavement sections. However, the modified sensor positions for the Dynaflect were also used to collect deflection data. Mean pavement temperature, cores of asphalt concrete pavement, and cone penetration test data were obtained concurrently. Asphalt recovered from the cores Was tested to establish the asphalt viscosity-temperature relationship. Asphalt layer modulus values corresponding to pavement temperature during NDT testing were computed from a previously established relationship between resilient modulus of asphalt mixtures and asphalt viscosity. Cone penetration tests provided information on stratigraphy, soil type, and cone-bearing value. Plate-bearing test data were also obtained at several test sites. Analyses of the field deflection basins were performed using the viscosity predicted E1 value and the analytically predicted values for E2, E3, and E4. These moduli values were used in BISAR and adjusted (tuned) to give the best possible fit to the measured deflection basins.Multiple linear regression analyses were performed with the FWD tuned moduli to establish new prediction equations which were similar to those originally developed from the analytical study. Log-Log plots of layer moduli and deflections from the Dynaflect modified sensor configuration indicated that a simple power law equation was adequate for defining the composite modulus of asphalt concrete and granular base (E1.2), E3, and E4. The resulting Dynaflect prediction equations appear to give reliable layer moduli within the established layer thickness and deflection constraints. However, the predicted Dynaflect layer moduli are usually greater than the FWD predicted moduli, especially for the base and subbase layers. The cone penetrometer data were used to establish a relationship to the resilient moduli for use in predicting moduli of layers within the subgrade support system. Since pavement response and performance is highly dependent upon subgrade soil-moisture regime, the cone penetrometer provides data suitable for elastic layer modeling and pavement distress evaluation.