Fatigue cracking is considered one of the main damage mechanisms in asphalt pavement design. Design methods use fatigue laws obtained by laboratory testing of the materials involved. Typically, these tests consist of subjecting the asphalt mixture to cyclic loading until failure occurs. However, failure is associated not with specimen fracture (which is unusual), but with a slight decrease in the mechanical properties of the material, usually in the complex modulus. As a consequence, it is important to differentiate between real damage to the material and changes in its viscoelastic behavior and thixotropy. It is also crucial to account for the healing that occurs in asphalt material after rest periods. The above considerations are important in the fatigue testing of asphalt binders because these materials show pronounced viscoelastic behavior and thixotropy, especially when subjected to cyclic loading. This paper demonstrates that in many cases what is taken for fatigue failure during testing (i.e., a decrease in the complex modulus below half of its initial value) is actually thixotropy. Thus, the complex modulus can be recovered by reducing the loading or, as in this study, the strain applied. In contrast, asphalt mixtures experience irreversible damage, and depending on the asphalt binder, the thixotropic effects are more or less pronounced. This paper analyzes the failure criteria currently used in the fatigue testing of asphalt mixtures and binders and evaluates the parameters chosen, namely, complex modulus (G*) and phase angle (δ) to characterize asphalt binders (G*sin δ). A cyclic uniaxial tension–compression test under strain-controlled conditions was performed. Three test modalities were used: time sweeps (constant strain amplitude until total failure), increasing strain sweeps (increase in strain amplitude every 5,000 cycles), and up-and-down strain sweeps (alternating increases and decreases in strain amplitude).
