Discrete element modelling of constant strain rate and creep tests on a graded asphalt mixture

This thesis investigates the use of Discrete Element Modelling (DEM) to simulate the elastic and viscoelastic deformation behaviour of an asphalt mixture. A numerical specimen preparation procedure has been developed to produce specimens with an isotropic stress and correct volumetrics. Stone mastic asphalt has been chosen in this project because of its high content of coarse aggregate and binder-rich mortar. A range of uniaxial compression tests have been undertaken in the laboratory under various loading speeds and stresses. The axial stress, axial strain and radial strain were recorded during the tests. The peak stress is found to be as a power-law function of the strain rates for the asphaltic material. Elastic contact properties have been used to investigate the effect of particle number and location, loading speed, normal and shear contact stiffness. The Poisson's ratio was found to increase with the ratio of normal to shear contact stiffness but was independent of the stiffnesses. The Young's modulus was found to be dependent on both normal and shear stiffnesses, in agreement with previous work on idealised asphalt mixtures. The Burger's model was introduced to give time-dependent stiffness for the viscoelastic modelling. The Burger's model was implemented to give moment and torsional resistance as well as in direct tension and compression. To reduce the computational time in the creep simulations, the effect of scaling both viscosities in the Burger's model to simulate a shorter time, have been investigated. The effects of each parameter on the deformation of asphalt mixture were also investigated. The stress-strain response for the laboratory tests and the simulations were recorded. The results show good agreement when the bond strengths in the model are made to be a function of strain rate for both constant strain rate and creep conditions. Both normal and Weibull distributions have been used for the bond strengths between the aggregate particles. The constant strain rate tests results were proved to be independent of the bond strength variability and position of particles, while the creep tests results were dependent on the hand strength variability and position of particles. This is in good agreement with experimental tests: different specimens at the same stress level gave variability in the creep behaviour, at higher stress levels. Bond breakages were recorded during the simulations to investigate the micromechanical deformation behaviour of asphalt mixtures. It was found that a well-defined rupture at higher stress levels coincided with the maximum rate of bond breakage. The modified Burger's model has therefore proven to be a useful tool in modelling the moment and torsional resistance at particle contacts in an asphalt mixtures, in order to correctly predict observed behaviour