The mechanistic analysis of pavement deflections on subgrades varying in stiffness with depth

Nondestructive deflection testing (NDT) has become an integral part of the structural evaluation of pavements. Interpretation of the measured deflection data is extremely complex and the analyzed pavement is often modelled as a multilayered elastic system. In this model the subgrade is usually defined as uniformly stiff and infinitely thick, or a rigid layer is placed at an arbitrary depth. The actual subgrade on which the tested pavement structure is founded varies considerably from this model. It is not infinitely thick and whether the subgrade is sedimentary or residual in nature, its stiffness normally changes with depth. This change in stiffness can be due to shallow bedrock, material differences, the stress history, or an apparent increase in stiffness due to the stress dependant behavior of most soils.; In this study a method to determine the apparent depth to a rigid layer from surface deflections is developed. This method is based on Bousinesq's equation and is related to a three layer linear elastic system through an extensive regression analysis. The procedure is validated by comparing the predicted rigid layer depths from surface deflections on five pavement sections to that obtained through penetration testing and seismic refraction analysis.; The methodology is further extended to pavement systems where the subgrade stiffness increases with depth. A nonlinear elastic backcalculation technique, based on a finite element approach, is used to illustrate the change in apparent stiffness with depth on a sandy and clay subgrade. It is shown that a three layer linear elastic system with an apparent rigid layer can be used to model the increasing stiffness with depth. This developed procedure is compared to existing backcalculation models. Monthly collected falling weight deflectometer (FWD) deflection data on ten in-service pavement sections are analyzed and the results are compared in terms of the available laboratory data. The inclusion of an apparent rigid layer into the pavement model led to considerable improvements in the backcalculated layer moduli. The procedure is also evaluated on two pavement sections instrumented with multidepth deflectometers. The backcalculated moduli are used to predict deflections within the pavement. These predictions are compared to actual measured deflections to determine how accurate the pavement was modelled.