Response Modelling of Bitumen, Bituminous Mastic and Mortar

Abstract

This research focuses on testing and modelling the viscoelastic response of bituminous binders. The main goal is to find an appropriate response model for bituminous binders. The desired model should allow implementation into numerical environments such as ABAQUS. On the basis of such numerical environment, Delft University of Technology (TU) is developing mechanistic asphalt mixture design tools. These tools are based on Meso scale mechanics. For Porous Asphalt (PA) performance, such a tool is readily available in ABAQUS. Implementation of an accurate viscoelastic response model for bituminous mortar will improve the tool’s capability in explaining PA performance at various temperatures and it is therefore a primary goal of this research. In addition the response model is also thought to be of equal importance for other meso mechanics tools for asphalt concrete mixtures that will be developed in the near future. To realise the main objective of the study, first an extensive Dynamic Shear Rheometer (DSR) testing program was carried out on bituminous binder, mastic and mortar. The program was carried out to get a better understanding of the response behaviour of binders for various loading conditions. For the pure binder and mastic testing, a cone and plate setup was developed. For the mortar testing, a specially designed mortar column setup was utilized. The frequency and time domain response of the binders was first analysed in the linear viscoelastic range. Hereafter the frequency domain response of the binders beyond the linear range response was investigated. The results showed that binders exhibit nonlinear behaviour at higher levels of shear stress. At relatively high temperatures, in the range of 30°C and above, mortar and mastic show nonlinear behaviour at shear stresses as low as 10 kPa. At low temperatures of 0°C and below, high shear stresses in the range of 1 MPa were observed to cause nonlinear behaviour. The second part of the study covers extensive modelling work. After a literature survey, two response models were first selected as a basis for further research; i.e. the Huet-Sayegh (HS) and the Burgers’ model. These models were then utilised to describe the frequency domain response data. It was observed that the Burgers’ model requires a number of Kelvin-Voigt elements to accurately describe experimental data. The HS model on the other hand presented an accurate description of response data. However, the HS model lacks the capability of explaining viscous deformation. For this reason the HS model was extended by adding a linear dashpot in series. The modified Huet-Sayegh (MHS) and the generalized Burgers’ model were then used to describe the frequency domain response of various materials. Results have shown that the MHS description of the response data excels that of the generalized Burgers’ model. For time domain use in numerical environments, incremental formulations of the response models were obtained. The formulations were coded and the numerical outputs were then validated by performing various simulations. The formulations were used to simulate time domain response tests. In this process parameter determination was first performed on the basis of frequency domain data. The parameters were then used to simulate time domain creep and relaxation tests. The simulation results showed that the frequency domain master curve data provided accurate material information for simulating the time domain response. The result further underlines the fact that binder’s behaviour is intrinsic, and as such their behaviour in frequency and time domain is related. It is this intrinsic behaviour of the binders that was described by the generalized Burgers’ and the MHS models. Hereafter the models were finally implemented into ABAQUS, and they are made available for use in the meso mechanics PA design tool. The results from the PA design tool showed that both models lead to comparable results. The pros and cons of the models for practical application were evaluated. For relatively small numerical models, the MHS model is suggested because of its simplicity in the number of model parameters and its high accuracy in describing material response. However, for computationally intensive numerical models, the use of the generalized Burgers’ model is suggested because of its high computational efficiency in numerical environments. Finally, the nonlinear response of binders was analyzed using Schapery’s nonlinear theory. Numerical formulation of the theory that incorporates the generalized Burgers’ model was adopted. The formulation was coded into a User Subroutine Material code (UMAT) for use in ABAQUS. In the UMAT code an iterative scheme for obtaining correct stress state was incorporated at the material level. Results from the code were verified by performing various simulations. Application of the Schapery’s nonlinear theory in the PA design tool showed that the effects of nonlinear behaviour are negligible at temperatures of 10°C and below. However, at 20°C and above, distinct and significant differences between linear and nonlinear simulations are observed. From the results it is concluded that common binders may be modelled as being linear viscoelastic for temperatures of 10°C and below. At 20°C and above nonlinear response becomes significant and, it needs to be considered in meso mechanistic computations.