Normal stiffness of fractures in granitic rock: A compilation of laboratory and in-situ experiments

Fractures and the networks they form have a major impact on the mechanical, thermal and hydraulic behaviour of a rock mass. For this reason, practically all geotechnical engineering projects conducted in fractured rock require quantitative descriptions of the properties of the fractures. A case in point is the project that motivated this study: the simulation of the surface subsidence resulting from the drawdown in pore-pressure about a deep tunnel in crystalline rock due to drainage into the tunnel [1,2]. Hydro-mechanically coupled discontinuum models of the medium required properties to be assigned, amongst which was the stress-dependent normal stiffness characteristics. Since there were no estimates of such for the fractures within the rock mass in question, a survey of published studies of normal stiffness was undertaken to guide the model parameterisation. Most studies measured fracture closure as a function of a change in applied effective normal stress. In laboratory experiments, the change was usually cyclical. A key problem in rendering such diverse data to be collectively interpretable is posed by the fact that the normal stiffness of fractures, kn, is strongly stress dependent, particularly at low effective normal stress levels (Fig. 1). This means that the closure-normal stress curves must be approximated by some function that fits the curves, preferably with as few free parameters as possible.