Faults and fractures form the largest contrast of fluid flow in the subsurface, while their permeability is highly affected by effective pressure changes. In this experimental study, fractured low-permeability Flechtingen (Rotliegend) sandstones were cyclically loaded in a MTS tri-axial compression cell.

Two different loading scenarios were considered: “continuous cyclic loading” (CCL) and “progressive cyclic loading” (PCL). During continuous cyclic loading, a displaced tensile fracture was loaded hydrostatically from 2 to 60 MPa in several repeated cycles. During progressive cyclic loading, the load was increased with a step-wise function (15, 30, 45 and 60 MPa) and unloaded after every loading step. For full elasticity of rock matrix deformation each rock sample has been preconditioned up to 65 MPa. After that, an artificial tensile fracture was introduced into the sample using the Brazilian Disk test. The fractured sample was installed into the MTS triaxial cell at a given offset of 0.5 mm and hydrostatic loading was applied accordingly. The fracture permeability was measured continuously using the cubic law calculated from the hydraulic aperture. Fracture closure was measured using LVDT extensometers during the entire experiment and the resulting fracture closure and stiffness was calculated accordingly. The total deformation of the sample was corrected by the amount of elastic deformation of the rock matrix to obtain the fracture closure only. Potential changes to the fracture surface topography before and after the experiments were analysed from high-resolution surface scans obtained by a 3D profilometer using the fringe pattern projection.

The scale-independent roughness exponent was calculated using power spectral density method assuming self-affinity. The fracture aperture distribution and contact-area ratio was calculated by matching the best fitting principal planes of the bottom and top surface and applying a grid search algorithm. The results showed a “stress-memory” effect of fracture stiffness during progressive loading that can be used to identify previous stress states in fractures. This effect is characterized by a transition from a non-linear to a linear (reversible to non-reversible) behaviour of specific fracture stiffness when a previous stress-maximum is exceeded. Furthermore, the evolution of fracture permeability shows less reduction during progressive cyclic loading compared to continuous cyclic loading.

The data measured during the flow-through experiment under varying effective pressure are provided in the file “MTS_data.zip”. The data are provided as separate text-files as well as in Excel format with different spreadsheets, such that each figure in the paper can be recalculated and that the underlying data is comprehensive. The name of all three rock samples is given in the file name including the type of the experiment (CCL or PCL). The fracture surfaces and the fracture aperture distributions are found within the file “Surface_data.zip”. This file contains the fracture data of each of the three rock samples as point cloud data (text-files), as well the data calculated from the surfaces.