Neural Network-based Backcalculation Models for Non-Destructive Evaluation of Rigid Airfield Pavement Systems

Heavy Weight Deflectometer (HWD) tests were routinely conducted on three Portland Cement Concrete (PCC) test items in the Federal Aviation Administration’s (FAA's) National Airport Pavement Test Facility (NAPTF) to verify the uniformity of the test pavement structures and to measure pavement responses during full-scale traffic testing. A six-wheel aircraft landing gear with dual-tridem configuration was trafficked on the north side of the test pavement while the south side was trafficked by a four-wheel landing gear with dual tandem configuration. HWD tests were conducted before and during traffic testing at slab center, at longitudinal joints and at transverse joints. Substantial corner cracking occurred in all three of the rigid pavement test items after 28 passes of traffic had been completed. Trafficking continued until the rigid items were deemed failed. This paper presents an Artificial Neural Networks (ANN) based approach for backcalculating the NAPTF rigid pavement properties. The study findings illustrate the complexity of backcalculating properties of rigid pavements subjected to full-scale dynamic traffic testing. Apart from other complicating factors such as the slab curling and warping behavior, the test pavements exhibited corner cracks within few passes of traffic loading which further complicated the interpretation of HWD test results.