Improving the Accuracy of Dynamic Modulus Master Curves of Asphalt Mixtures Constructed Using Uniaxial Compressive Creep Tests

Abstract

Uniaxial monotonic or constant loading tests have been used to efficiently construct master curves of the magnitude and phase angle of the complex modulus of asphalt mixtures. However, when using monotonic or constant tests, the test data fitting model and time-temperature shift factor equation were usually arbitrarily chosen to construct the master curves, which were not validated at all or were validated using only a couple of data points. This study developed a composite test protocol consisting of uniaxial compressive creep tests to construct the master curves and dynamic modulus tests to validate the constructed master curves of the magnitudes of the complex moduli of asphalt mixtures. The loading force and duration were carefully chosen for each test segment to assure that the test specimen was retained in the linear viscoelastic stage during the entire test protocol. Commonly used data fitting models and time-temperature shift factor equations were compared, respectively, when constructing and validating the master curves. The Prony series model was determined to be the best fitting model for the creep strains because it not only provided satisfactory modeling accuracy but also complied with the actual material properties of asphalt mixtures. The Williams–Landel–Ferry (WLF) equation was demonstrated to be the most appropriate time-temperature shift factor equation because the master curve constructed with the WLF equation made the most accurate predictions in a fairly wide range of frequencies. This study refined the master curve construction method for asphalt mixtures using uniaxial monotonic or constant loading tests. With properly selected and validated data fitting model and time-temperature shift factor equation, the constructed master curve had the capability of accurately predicting the magnitude of the complex modulus in a wide range of loading frequencies with an R2 value of approximately 0.97 or higher.