Soft Computing Methodology to Determine Pavement Thickness from Falling Weight Deflectometer Testing

Abstract

This paper describes an advanced pavement analysis application based on a soft computing methodology, SOFTSYS, recently developed at the University of Illinois for identifying pavement geometry and layer properties through various nondestructive testing methods. Specifically, it is capable of backcalculating pavement layer properties as well as determining thickness of full depth asphalt pavements using results of the Falling Weight Deflectometer (FWD) test. SOFTSYS uses a combination of Genetic Algorithms and Artificial Neural Networks together with a structural pavement analysis model based on the Finite Element Method to obtain accurate and reliable backcalculation solutions. In this paper, basics of SOFTSYS, its numerical schemes and computer code implementation are described first within the scope of soft computing methods. An illustrative example of pavement backcalculation is then presented using the synthetic FWD field data obtained from nonlinear ILLI-PAVE finite element program representing typical fulldepth asphalt pavements.