A Rapid Approach to Interpretation of SASW Results

Abstract

Nondestructive testing (NDT) of pavements has made substantial progress during the last two decades. Most algorithms currently used to determine the remaining life of pavements rely on stiffness parameters determined from NDT devices. One major area of continual improvement is the reliable extraction of stiffness parameters from nondestructive field data. The Spectral analysis of Surface Waves (SASW) method is one of the NDT methods that is used more frequently because of its capabilities in characterizing the near-surface layers more effectively. In this method, time records obtained with vibration sensors are used to obtain an experimental dispersion curve, which provides, through an inversion procedure, an estimate of the elastic modulus profile of the pavement. The inversion process requires a significant computational effort or frequent operator's intervention. To improve the user-friendliness of the inversion process, a new algorithm for the rapid reduction of the SASW data has been developed. Thickness and modulus of each pavement layer are estimated in real time using artificial neural network models. The training and validation of models are done using an axisymmetrical full-waveform forward model to minimize the approximations associated with simpler models used in the inversion algorithms. This paper provides an overview of the proposed inversion and its practical use and limitations in pavement analysis and design. The reduction algorithm seems to be robust and to yield consistent results in almost real time. For the covering abstract see ITRD E118503.