Laboratory and computational evaluation of compact tension fracture test and texas overlay tester for asphalt concrete

Abstract

Reflective cracking is the primary mode of failure for pavements rehabilitated with asphalt overlays in many instances. The Texas Overlay Tester (OLT) has been utilized by several researchers and practitioners to evaluate the reflective cracking resistance of asphalt overlays. The OLT is a simulative test procedure that emulates the portion of asphalt overlay located directly on top of the crack or discontinuity in the underlying pavement. The testing involves cyclic horizontal displacement of the underlying layer to initiate and propagate the crack. The number of cycles required to form the crack through asphalt overlay is typically utilized as a performance parameter indicative of cracking resistance of the asphalt mixture. This paper describes a comprehensive analysis of the OLT through comparative laboratory fracture testing and computational modelling. The compact tension (CT) test geometry has been recently adapted to characterize the fracture properties of asphalt concrete, and can be used to extract useful mode I (tensile) local fracture properties such as material strength and fracture energy. Laboratory creep and fracture testing was conducted for two hot-mix asphalt samples. Both OLT and CT tests were conducted for each mixture, and both tests were simulated using the finite element technique. The simulation results and the laboratory findings demonstrate the relative pros and cons of each approach (fracture test versus simulative test). Reasons for the significantly higher variability found in the OLT as compared to the CT test are hypothesized and discussed. The development and implementation of a phenomenological cohesive zone fatigue (CZF) model specifically tailored for this study is presented. The CZF model utilizes fracture properties obtained from CT test along with a functional degradation of those properties under cyclic straining as calibrated using OLT results. The calibrated model was shown to be in favourable agreement with laboratory testing results. Extensions and limitations of model are also discussed. © RILEM 2012.