Two-phase flows in open geological fractures: a combined experimental and numerical investigation – 2PhlowFrac

Many geotechnical and environmental engineering problems, such as geothermal energy production or storage of fluids in deep formations, involve flow processes in subsurface fractured media. When two immiscible fluids coexist, the flow involves the displacement of the interfaces between them. This so-called two-phase flow can result in very complex and intricate spatial distributions of the fluids, depending on the mean flow velocity, the viscosities and densities of the fluids, the mechanical properties of their interface, and the geometry of the permeable medium. For porous media, the rich phenomenology of two-phase flows has been investigated at length in the last 20 years. In geological fractures, on the contrary, the characterization of this rich phenomenology is still in its infancy. The first main objective of this project is to systematically investigate it, as a function of the fluids’ properties, flow conditions, as well as fracture geometry and closure.

This objective will be tackled through a combination of (i) laboratory experiments on transparent setups reproducing a realistic fracture geometry but allowing for measurements of the spatial distributions of the two fluids and of their velocities, and (ii) numerical simulation of the two-phase flow in the fracture space, based on the first principles of fluid mechanics. The laboratory experiments will be set up and run at Géosciences Rennes, while the numerical study will be developed and run at University Hannover, thus building on the complementary expertise of the two groups. An innovative so called depth-averaged two-dimensional numerical simulation, which directly computes fluid velocities averaged over the fracture aperture, will also be developed jointly by the two groups. It is expected to provide an excellent compromise between prediction accuracy and computing efficiency. Systematic characterization of two-phase flow in rough fractures from the experimental and numerical data will be done jointly by the two groups.

For practical applications, which involve length scales of tens to thousands of meters, the length scale of fluid-fluid interfaces cannot be resolved. To predict flow on field case studies, so-called continuum scale numerical models, which are not based on the first principles of fluid mechanics, are therefore used. These models are known to not often reproduce the spatial fluid distributions well. For example, the amounts of naturally-occurring fluid that remains in the subsurface after an injected fluid has displaced that previously-resident fluid, or the travel times of fluids over a given distance, are often poorly predicted. The second main objective of this project is to improve such large scale predictions by using the newly-acquired knowledge of two-phase flow phenomenology to find appropriate large scale model parameters that will optimize the prediction of large scale observables such as the amount of displaced fluid remaining in the geological formation.