Comparison between mechanistic analysis and in-situ response of full-depth flexible pavements

Accurate prediction of tensile strain at the bottom of hot-mix asphalt (HMA) is crucial for mechanistic pavement analysis. In this study, transverse tensile strain at the bottom of HMA was calculated using mechanistic analysis and compared to measured values from test sections exposed to accelerated pavement testing (APT). Three full-depth flexible pavement sections having 152, 254, and 420mm of HMA placed on 300mm of lime-stabilized subgrade were constructed and exposed to APT using the Advanced Transportation Loading ASsembly (ATLAS), housed at the University of Illinois at Urbana-Champaign. Transverse tensile strain histories at the bottom of HMA under various loading conditions (load, tire pressure, and speed) were measured through embedded strain gauges at the HMA-stabilized subgrade interface. Indirect tensile complex modulus tests for three HMA materials were conducted using an Instron Universal Test Machine (UTM). The elastic modulus of subgrade was backcalculated from FWD testing conducted on the test sections. The pavement responses of test sections were calculated using the NCHRP 1-37A Mechanistic-Empirical Pavement Design Guide (MEPDG) procedure as well as a finite element model. In general, the MEPDG approach underestimates the transverse and longitudinal strains when the pavement thickness is equal to or smaller than 254mm; the difference is more manifested at shallow depths below 152mm. Hence, the pavement thickness design based on MEPDG procedure could underestimate the pavement fatigue damage for thin pavements. Inaccurate calculation of the loading pulse period and the empirical conversion between loading period and frequency are among the assumptions that contribute significantly to the ill-prediction of pavement responses when current NCHRP 1-37A is used. A 3D finite element model, on the other hand, was developed in order to predict pavement response to tire loading. The model considers measured tire-pavement contact stresses, continuous moving-wheel loading, implicit dynamic analysis, and HMA viscoelastic characteristics. Difference between measured and predicted strains was found to be within 5%.