Road structures are made by a superposition of different layers of heterogeneous materials (bituminous materials, cement concrete, composite grids, etc). The durability of such composite structures depends on severe climate conditions that occur during their lifetimes and multiple heavy loads that move on them. Various damages are identified such as delamination mechanisms at the interface between surface pavement layers. Studying debonding phenomenon is a necessity in order to propose innovative solutions to maintain an old efficient transport road network. This also helps to develop new concepts for turning road infrastructure with new added functionalities such as those that integrates into the road surfaces a dynamic charging box for electric vehicles. Nowadays, this type of cracking in pavement structures is still not well understood. As for edge delamination in composite field, edge effect of existing joints or vertical cracks in a pavement layer create such a high concentration of both normal and shear interface stresses that the crack could propagate along the interface between the two different layers before penetrating through one of materials or even debonding elsewhere far from them (Chabot et al., 2013). Appropriate fracture opening pure mode I or II and mixed mode laboratory tests under several static and fatigue load conditions and various environmental conditions (mainly temperature and moisture) need to be adapted or developed for the specific bending study of such multilayered structures.

The main objective of this paper is to compare the results from different techniques in the aim to characterize experimentally the interface crack initiation and propagation by way of interfacial fracture energies obtained on composite specimens coming from road construction. Shear fracture mode tests in the pavement field are not so easy to modify and it is not so simple to compare their "pure" mode test results to a part of those coming from fracture mixed mode tests. In the present work and following many previous works, the Wedge Splitting Test (WST) developed by Tschegg (1986) is chosen. This test is a quite convenient test method to study the fracture behaviour at the interface of heterogeneous materials in mode I. The WST test has been adapted for specimens extracted from full scale pavement sections (Gharbi et al., 2017). Cubical shape specimens of important size (compared to the granular maximum size of the materials) are prepared with a cylindrical groove instead of a traditional rectangular groove. A slim wedge of 14° is used to ensure principally the studying of the interface debonding phenomenon in a quasi-pure opening Mode I. A notch of 5 mm thick is sawed at the interface between the two material layers to guarantee the initiation of the crack at the interface. A finite element study is conducted to analyze the effect of variation of the specimen dimensions. Static tests controlled with a constant displacement speed rate (0.7mm/mn or 2 mm/min) and constant temperatures (~20°C) are performed in order to determine the specific fracture energy (GF) of interface between bi-layered specimens. Different techniques are used to evaluate initiation and propagation of the fracture with the modified WST test. The tensile strength and the fracture energy are firstly estimated from the splitting force (FS) - Crack Mouth Opening Displacement (CMOD) curve. To obtain the displacement field at points between the two crack lips, the Digital Image Correlation (DIC) technics are used. Then these displacement fields are introduced in the elastic model proposed by Dunder (1969) to calculate, under 2D conditions (plane stress and deformation), the strain energy release rate evolution. This energy depends on the interface crack length estimated also by DIC measurements. Results found from the two methods are finally discussed.