Determination of pavement layer moduli from Falling Weight Deflectometer (FWD) test data is known as backcalculation analysis. Generally, backcalculation analysis is an unstable procedure which is greatly influenced by several types of errors. These errors may be categorized according to modeling error in the forward analysis, error in deflection measurements, modeling error and instability in the backcalculation procedure, etc. Because of all the problems mentioned, selection of seed values for layer moduli would highly influence backcalculation results. In order to reduce effects of measurement error, truncated singular value decomposition is utilized in backcalculation for regularization purpose. Scaling of variables, which is often used in optimization algorithm, is implemented to improve numerical accuracy. In dynamic backcalculation, Ritz vector reduction method is employed to efficiently solve a large system of dynamic equations. Various other means are also introduced to cut down computation time. This paper presents recent updates of DBALM (Dynamic Back Analysis for Layer Moduli), whose solver is axi-symmetric FEM and was first developed in 1993. Examples on airfield pavement are also presented. The results are compared with the results from the authors' static backcalculation software BALM (Back Analysis for Layer Moduli) where the solver was developed using multilayered linear elastic theory. From the authors' experience, they believe a dynamic backcalculation is superior to static backcalculation. The difference between the results from the two methods is presented in this paper.
Abstract