Geothermal projects relying on superhot rocks (SHR), such as the Japan Beyond-Brittle Project, Iceland Deep Drilling Project, and Newberry Volcano, aim to harness heat from geothermal reservoirs where water reaches a supercritical state (temperature ≥ 400 °C and pressure ≥ 22 MPa). Such projects could multiply the power output of geothermal power plants by a factor of ten, positioning them at the forefront of the energy transition. However, a major challenge hindering the widespread application of SHR is the fact that supercritical water resources are often found in regions of the crust where rocks exhibit ductile behavior, a rheological regime where the formation of large-scale fractures and cracking is hindered. However, these fractures are crucial for enabling water flow, and currently, the evolution of rock permeability and other hydraulic properties in this context remains largely unknown.

This study presents experiments conducted in TARGET, a newly designed gas-based triaxial apparatus located at EPFL, CH. Cylindrical cores of Lanhélin granite (40 x 20 mm) were deformed under an effective confining pressure of 100 MPa, temperatures ranging from 200 to 800 °C, and a strain rate of 10^-6 s^-1. Continuous recording of sample permeability using the pore pressure oscillation method was carried out during deformation. Moreover, post-mortem samples were retrieved and scanned at the ESRF synchrotron facility (Grenoble, FR) and the tomographs were used to reconstruct the 3D crack network. Flow in the sample was then modelled using the Avizo XLab extension and permeability was computed in the x y and z direction.

We report that Lanhélin granite transitions from being in the localized regime with the formation of a sample scale fracture to becoming ductile between 600 and 800°C. In the brittle, localized regime, sample permeability remains relatively constant throughout deformation. In the ductile regime, sample strength is halved, and beyond the initial decrease upon loading, permeability increases monotonically by more than an order of magnitude. These results suggest that sample bulk controls the sample permeability in our experiments. In localizing samples, fractures do not connect the ends of the rock core but concentrate all strain after nucleation, limiting permeability improvement through micro-cracking in the bulk. In the ductile regime, where no localization occurs, the bulk permeability of the rock continuously improves with strain. Flow modeling in post-mortem samples yielded permeability values up to seven orders of magnitude greater than in-situ measurements. This substantial difference is attributed to the effect of confining pressure on the crack network aperture. Despite this absolute difference, our modeling results confirm that flow in nominally ductile samples is controlled by bulk cracking rather than macroscopic fractures. Our study demonstrates that low-porosity rocks in the ductile regime can be more permeable than often anticipated. These results hold significant implications for the engineering of SHR reservoirs, showcasing the potential for permeability enhancement in ductile rocks.