Dynamic backcalculation of pavement properties using optimization in nondestructive evaluation

Evaluation of the properties of the existing pavement is of importance to the prediction of the future pavement performance and to pavement management. Currently, two evaluation methods are used: destructive evaluation and nondestructive evaluation, nondestructive evaluation generates great interest due to its fast and economical operation. Falling Weight Deflectometer (FWD) is one of the most popular nondestructive pavement testing devices used today.;This dissertation presents a dynamic backcalculation approach to determine pavement layer properties using FWD experiment data. The dynamic backcalculation method consists of two components: forward model and inverse model. The forward modeling adopts the theory of transient wave propagation in a multilayered medium based on the elastodynamic equations. Laplace-Hankel mixed transform and transfer matrix techniques are employed to solve these elastodynamic equations. Efficient numerical evaluation algorithms of integral transformations and verification of forward modeling are also presented.;The inverse model is the optimization process. The general optimization problem is stated by its objective function and constraints. The optimization related issues, including objective functions, optimization methods, constraint handle techniques, and numerical differentiation techniques are discussed in details. The dynamic backcalculation can be performed in time domain or complex frequency domain. For the dynamical backcalculation in time domain, the inverse model is based on time histories of displacements. For the dynamical backcalculation in complex frequency domain, the inverse model is based on transfer function.;By comparing the outputs from dynamic backcalculation with the theoretical values and with the results using static backcalculation programs, the proposed dynamic backcalculation method proves to be effective.