01373734

Component

Various backcalculation methods have been proposed to calculate pavement structural properties from surface deflection measurements by falling weight deflectometer (FWD) nondestructive monitoring devices. However, no unique technique has proved to yield a globally optimum solution to this complex, non-differentiable problem. The most successful methods generally calculate elastic moduli of pavement layers based on known layer thicknesses and interface bond conditions, which may not be available accurately. This study explores development of an effective and reliable backcalculation strategy with attention to variable layer thickness and interface bond condition factors. An artificial neural network (ANN) was initially trained and tested using synthetic databases for typical three-layer flexible pavement structures, created using the 3-D finite element EverStressFE®. To minimize the error between the FWD-measured deflections and ANN-calculated deflections, a restart covariance matrix adaptation evolution strategy (CMA-ES) was implemented. Preliminary testing of the developed methodology (RCMA-BC) on the synthetic database demonstrated its effectiveness and reliability in backcalculating moduli and AC layer thickness. However, RCMA-BC cannot reliably backcalculate base thickness and interface shear stiffness because the forward calculation routine does not have significant sensitivity to these parameters. While the backcalculated subgrade modulus appeared to be insensitive to change in interface bond, base layer modulus was highly affected by erroneous assumptions for interface stiffness. The RCMA-BC models and Evercalc® backcalculation software were applied to data from FHWA’s Long-Term Pavement Performance (LTPP) database; RCMA-BC exhibited consistently lower errors in deflections. The backcalculated layer thicknesses were also compared to available ground penetrating radar (GPR) and coring information.

This paper was sponsored by TRB committee AFD80 Strength and Deformation Characteristics of Pavement Sections

01362476

12-4127

English

20p

2012

Transportation Research Board 91st Annual Meeting

Digital/other

Figures; References; Tables

Backcalculation; Covariance; Falling weight deflectometers; Flexible pavements; Ground penetrating radar; Matrices (Mathematics); Neural networks; Stiffness; Thickness

Structural properties

Highways; Pavements; I30: Materials

Transportation Research Board Annual Meeting 2012 Paper #12-4127

TRIS, TRB

Feb 8 2012 5:22PM