Unresolved issues in fractured rock hydrogeology

A brief historical review starts from the double porosity approach of Barenblatt in 1960; various versions of it are still used today despite the difficult evaluation of the exchange coefficient between the two media. Later, the homogenisation theory was applied, but it stopped before solving the macroscopic equations. A decisive step was achieved in 1980 by viewing a fracture network as a set of disks randomly distributed in space. This was the starting point of our own works with additional ideas such as the crucial role of the dimensionless density for convex fractures and the use of systematic numerical experiments in order to rationalize the results.

Today, with the powerful computers at our disposal, very complex configurations can be addressed, but the theoretical and methodological gains are limited.

Therefore, the basic issues have not changed for real networks. They can be summarized and illustrated by our simple approach. First, what are the most salient features of the measurements made on oucrops and wells? Some of them provided a basis to the theoretical predictions of quantities such as percolation and macroscopic permeability, but alternative approaches could certainly be developed. Second, how can we reconstruct, i.e., generate realistic networks based on these measurements? Statistical techniques have been extensively used, but more advanced techniques such as AI should be envisioned. Third, experimental results are usually scarce and often not easily compared with numerical predictions.

Despite and/or because all these difficulties, this field is expected to remain very active in the future.