Multi-scale study on the deformation and fracture evolution of clay rock sample subjected to desiccation

International audienceThe aim of this paper is to compare and discuss the values of strains and crack apertures associated with desiccation cracks measured in Tournemire clay rock at different scales (micrometer to decimeter). Experimental investigations in the laboratory were conducted on one clay rock sample subjected to a desiccation process. Two faces with dimensions of 20 × 20 mm2 (i.e., macroscopic scale) and 5.1 × 4.1 mm2 (i.e., mesoscopic scale) were analyzed. The induced hydric strains and desiccation cracking were monitored by digital image correlation combined with a new algorithm (H-DIC). The results were compared with the data of Hedan et al. (2014) at the gallery scale (decimeter) and those of Wang et al. (2013) at the microscopic scale (micrometer).Our laboratory study yielded the following phenomenological results. First, the displacement fields revealed the presence of sub-horizontal cracks associated with the direction of bedding planes and sub-vertical cracks, as previously observed in a gallery front in Tournemire Station. Second, when the relative humidity (RH) decreased between 98% and 33%, the crack aperture kinematics at the macroscopic scale (centimeter) was divided into three steps: (i) a phase of opening and closure, (ii) a phase of only gradual closure, and (iii) a final phase in which the desiccation cracks closed. Only phases (ii) and (iii) were observed at the mesoscopic scale (millimeter), revealing that the kinematics of cracks depends on the scale observed. The comparison of the strains at the mesoscopic and the macroscopic scales also highlights that their values depend on the study scale: the presence of cracks at the mesoscopic scale leads to a large overestimation of the values of the strains calculated at the macroscopic scale.In contrast to the observations in the laboratory, the desiccation cracks detected in the gallery systematically open when RH decreases. This difference and the differences observed in the geometrical organization of crack networks are explained by the different boundary conditions prevailing in both cases (i.e., free swelling/shrinkage in laboratory versus constrained swelling/shrinkage in the gallery).The interpretation of the entire dataset emphasizes the need for a multi-scale approach to understand and model desiccation cracking mechanisms and the associated hydric strains in clay rocks