Various phenomena other than fatigue (so-called “biasing effects”) occur during laboratory fatigue tests on asphalt mixes because of cyclic loading applications, thus altering experimental results and leading to misleading conclusions. The purpose of the study is to isolate and quantify biasing effects, therefore isolating real fatigue damage. In particular, non-linearity, self-heating and thixotropy (defined as a recoverable viscosity reduction after shear application) were evaluated. Six different mixes were produced using three distinct asphalt binders. Tests were performed in tension/compression mode on cylindrical samples. A particular test procedure was followed, consisting of two parts. In the first part, complex modulus measurements were performed at temperatures from 8°C to 14°C and strain amplitudes from 50 to 110 µm/m, at 10 Hz. Regression equations were fitted in order to evaluate variations of norm of complex modulus and phase angle caused by temperature and strain-level changes around common fatigue test conditions (10°C, 100 µm/m). In the second part of the test, five partial fatigue tests (each one consisting of 100,000 cycles at a 100 µm/m strain amplitude) were performed at 10°C, 10 Hz. After each fatigue lag, a 24 hour rest period was imposed. During rest periods, short complex modulus measurements were performed (10°C, 10 Hz) in order to monitor the recovery of mechanical properties. Surface and internal temperature of samples were constantly measured throughout the entire test, in order to monitor self-heating due to repeated loading. A significant temperature increase was observed during each fatigue lag, while, during rest periods, temperature rapidly decreased to the initial value. Self-heating was observed to be correlated to viscoelastic energy dissipation. The procedure used in the study allowed quantitatively estimating biasing effects. Therefore, unrecovered mechanical properties, due to damage accumulation, were obtained. Ninety per cent of total complex modulus and phase angle variations observed during each fatigue lag were found to be completely reversible. Non-linearity and thixotropy appear to influence mechanical properties variations more importantly than self-heating.
Read full abstract