Virtual Mechanical Research On Mesoscopic Structure Of Asphalt Mixture Based On X-Ray CT Technology

Asphalt mixture has been widely used in highway pavement. However, with the popular application of asphalt pavement, more problems have appeared. For a long time, the design system of asphalt mixture is mainly based on simplified models or tests in laboratory, which could cause a great waste of time and energy and still not explain the complex mechanical properties of asphalt mixture. Finite element method has been employed by many researchers to study asphalt mixture from meso-level, but it has the deficiency on describing the contact between materials. The relationship between internal structure and macro tests of asphalt mixture was discussed in this research based on X-ray CT and discrete element method.Firstly, three typical kinds of gradations, namely AC20, SUP20 and SMA20 were studied with the help of X-ray CT scanning and improved digital image process., from which we got the image of internal structure in asphalt mixture. Through building a relationship between two dimensional gradations and three dimensional gradations, proper vertical two dimensional images were chose according to which, numerical models were established in Particle Flow Code 2D. Afterwards, uniaxial creep test of asphalt mastic was carried out to get the values for parameters in burger’s model. After all the parameters in PFC2D were obtained, uniaxial compression tests were simulated successfully.Secondly, In order to find out the influence of aggregate stiffness and mastic viscoelasticity on the uniaxial compressive strength, orthogonal tests have been designed. What’s more, BP neural network has been employed to build the relationship between macro mechanical property and contact parameters. The effect of different gradations on such relationship has also been discussed. The orthogonal tests show that the normal stiffness of aggregates has significant influence on uniaxial compression strength. The combination of BP neural network and PFC demonstrated that the relative error of network for SUP20 and AC20 is below 6% while for SMA20 it reaches 15%.Meanwhile, the movement of aggregates inside asphalt mixture during loading was studied and index has been proposed. Center-of-mass coordinate and long axis were used to describe the rigid movement and rotation. Aggregates were found to move vertically at first and then move at horizontal direction. The distribution of contact force among mastic was analyzed at the same time. The amount of contacts in mastic for three specimens had the same pattern while the values developed differently. We found that mastic in SUP20 bore tension while in AC20 bore compression, but for SMA20 mastic didn’t carry force.It was found that the not only the property of aggregate and mastic but also the gradation has an impact on macro performance of asphalt mixture. It is hoped that with this study, more new ideas about research on internal structure of asphalt mixture could come out. Deep understanding on asphalt mixture could be achieved based on digital image process, discrete element method and BP neural network.