This research aimed to investigate the attenuation mode of the layer modulus of asphalt pavement in accelerated pavement testing (APT). A full-scale experimental section was constructed and tested using the APT facility. Two non-destructive testing (NDT) methods, named falling weight deflectometer (FWD) technique and portable seismic property analyzer (PSPA) test, were used to obtain the layer moduli of asphalt pavement during the APT test. The variation patterns of layer moduli obtained by FWD and PSPA tests were calculated and compared after the temperature was corrected to 20 ℃. It was found that the variation pattern of surface layer modulus based on field FWD measurements was consistent with the one measured from PSPA tests. That is the modulus of the surface layer increases with the APT load repetitions firstly and then decreases with the rise of the repetitions. The modulus values of the surface layer measured from PSPA tests are obviously larger than those backcalculated based on deflection basins. The ratio of the measured surface layer modulus based on the PSPA test to the backcalculated one based on the FWD test ranges between 2.06 and 2.71. The backcalculated base layer modulus always declines with the increasing loading repetitions. The attenuation patterns of the surface layer modulus and the base layer modulus in the damage stage are described as E_a=421100*N^-0.6119 and E_b=128000*N^-0.1096, respectively.

The modulus of asphalt mixture is traditionally measured from the laboratory dynamic modulus test. However, different laboratory test methods often lead to obviously different test results. To evaluate the moduli of asphalt mixtures as constructed in field pavements, this study back-calculated the moduli of the pavement layers, based on measured strain data in Accelerated Pavement Testing (APT). Field tests were conducted to measure strains at different locations of the built pavement section at different temperatures and wheel motion speed. The loading frequencies of the asphalt layer subject to different motion speeds were calculated based on the duration of measured strain pulses. Subsequently, the relationship between the motion speed and the loading frequency was established. An finite element (FE) model of the pavement section was created, which was used to back-calculate the moduli of asphalt pavement layers using the measured strain data. Based on the back-calculation results at different loading conditions, the master curve of the field asphalt layer was determined. This master curve was further used to compare with that obtained from laboratory uniaxial compressive test. The relationship between the field and laboratory moduli was determined to be E_Laboratory= 1.298E_Field.

The asphalt-based pavement performance evaluation includes the derivation of the dynamic moduli and loading frequencies of pavement layers under various traffic- and climatic-induced loading conditions. The traffic-induced strain pulses and loading frequencies of commonly used (semi-rigid, flexible, and steel deck) asphalt pavements were experimentally determined by vehicular loading field tests with embedded strain gauges for different axle loads, motion speeds, and temperatures. It was found that the axle load values had no noticeable effect on the pavement loading frequency, which was mainly controlled by the vehicular motion speed. The transverse frequencies were found to be higher than longitudinal ones, while the distributions of loading frequencies by pavement depth differed for three pavements under study. The frequency values at temperatures over 35 °C exceeded those at lower temperatures, while in the temperature range from 4 to 31 °C, the motion speed vs. loading frequency relations for three pavements were nearly identical. The loading frequency f increased approximately linearly with the motion speed V, according to the unified fitting equation for three types of pavements under study, namely f = 0.127 × V. This unified equation was further proved valid to predict the dynamic modulus properties of field asphalt pavement layers. Moreover, several previous prediction models for loading frequency, including the Brown model, Ullidtz model, MEPDG procedure and Ulloa model, were compared to the results in this study. These previous models were found to overestimate the loading frequencies within the asphalt layer. The prediction errors of the Brown model and the Ullidtz model were pronounced. The loading frequencies calculated by the MEPDG procedure and the Ulloa model need to be modified by dividing 2.8 and 1.7, respectively.