Best practices in surface and subsurface natural fracture characterisation to advance carbonate geothermal reservoirs insights: a spotlight on the Geneva Basin, Switzerland.

Naturally fractured geothermal reservoirs (NFGR) represent a challenging frontier for sustainable energy exploration and production. These reservoirs are characterised by the presence of complex fracture networks controlling hot fluid movement at depth. Unfortunately, these networks cannot be directly observed, and their properties need to be modelled. Classically, these models are based on statistic data obtained from outcrops and borehole data. Outcrops allow characterisation of the geometry of networks at a scale up to 100’s of meters. However, the analogy between surface and subsurface is not trivial and surprisingly, the notion that different fracture sets are genetically related is rarely used. Borehole core data provide the only direct sampling of the subsurface. However, cores are challenging and expensive to obtain. As an alternative, geophysical borehole images are acquired to observe fractures in the subsurface. However, the quality of these images is variable, making their interpretation uncertain. In this study, we aim to minimize the uncertainties related to fracture picking in borehole images to accurately recognise specific part of the network interpreted from the surrounding outcrops. This approach will provide new perspective in the characterisation of NFGRs.

We test our approach on the Geneva Basin, located in westernmost part of Switzerland. There, the Lower Cretaceous carbonate units are expected to host geothermal resources. Recently, an exploration well, GEo-01 was drilled in the Canton of Geneva to evaluate the geothermal potential of the Lower Cretaceous reservoir. The basin is bounded to the north by the Jura and to the south by the Saleve mountain and the Borne massif where the Lower Cretaceous rocks are outcropping. This area extends for about 1500 km².

In these outcrops, we introduce the concept of discontinuity associations where sets of fractures, veins and stylolites which formed under a similar stress regime are grouped together. The characterisation of discontinuity associations allows to map the orientation of the maximal principal paleostress (σ₁) of genetically related discontinuities. This method is a more robust way of reconstructing fracture-forming deformation events than assigning one deformation event per discontinuity set. We consistently identify three distinct associations over the investigated mountain ranges. Those associations are formed before the onset of fold-and-thrust belt and therefore constitute a background fracture network expected to be found in the targeted geothermal reservoir.

To prove this hypothesis, we looked for the same discontinuity associations in borehole images of GEo-01. This well disposes of a unique dataset of five independent interpretations of the same 122 m interval of Lower Cretaceous series. To quantify and reduce interpretation uncertainties, our study involves a comparative statistical analysis of these interpretations. The outcomes of this  analysis facilitate the identification of intervals where interpreters reached consensus and those where discrepancies emerged. We delved into the factors influencing interpretation agreement or divergence, considering fracture attribute variability, image log quality variation, and geology. We define guidelines to interpret fractures in the borehole images of the Lower Cretaceous of the Geneva Basin and ultimately validate the presence of the three discontinuity associations as background fractures in the geothermal reservoir.