Development and Analysis of High-Modulus Asphalt Concrete Predictive Model

Abstract

The main purpose of this paper is to present the development of a new predictive model intended for the calculation of stiffness modulus |E*| determined by a four-point bending beam test (4PBB or 4PB-PR). The model developed, called model A, was based on the Witczak model, which was developed for the dynamic-modulus (DM) method. Most of the asphalt mixtures used to develop the model were high-modulus asphalt concrete (HMAC). The most commonly used methods for determining the stiffness modulus |E*| of asphalt mixtures were also discussed. The paper presents the results of the study for 10 asphalt mixtures but 8 of them were used to develop the predictive model. In addition, the results of complex shear modulus G* tests on neat and modified bituminous binders carried out in a dynamic shear rheometer (DSR), necessary for the development of a predictive model, are presented. The tests carried out in the dynamic shear rheometer had significant measurement uncertainties. The results of the volumetric parameters of the asphalt mixtures are also reported. The developed model A has maximum absolute errors e = 1930 MPa (p = 95%) and maximum relative errors re = 50% (p = 95%). The distribution of the absolute errors of the model, after discarding outliers, has a normal distribution as in the development of other models of this type, which was confirmed by appropriate statistical tests. On the basis of the tests and calculations carried out, it was concluded that, in order to increase the precision of the predictive models, it is advisable to reduce the measurement uncertainty of the bitumen complex shear modulus G*. For the developed model A, the limiting values of the stiffness modulus |E*| are also shown, within which the determined stiffness modulus should fall.