Preparation of fracture data and delineation of bedrock domains for 3D DFN modeling of a cavern thermal energy storage site in Southern Finland

Cavern thermal energy storage (CTES) is increasingly recognized as a key technology for integrating renewable energy sources and balancing seasonal heat supply and demand. The implementation of CTES in crystalline bedrock settings requires detailed characterization of brittle deformation zones and fracture networks that both control heat transfer and cavern stability. Site-scale zones of localized brittle deformation, ranging from tens of centimetres to several metres in thickness and tens to hundreds of metres in length, can often be identified from drillcore interpretations and represented deterministically in structural geological models. In contrast, fracture networks are commonly constrained by sparse observations of individual fractures with small apertures, necessitating stochastic approaches to account for limited sampling and uncertainty within poorly constrained subsurface volumes. Consequently, detailed fracture mapping and classification, together with the identification of volumes of bedrock constrained by the deformation zones, i.e. bedrock domains, are essential prerequisites for fracture network modelling that utilize, for example, Discrete Fracture Network (DFN) approaches.

In this study, we evaluate the usability of the conventionally acquired and subsequently classified fracture data for generating fracture sets for 3D DFN models in crystalline bedrock deformed by multiple tectonic events and comprising variably altered granites and gneisses. We also evaluate brittle deformation zones as constraints for determining structurally homogeneous bedrock domains. The study focuses on the planned largest CTES site in the world, VARANTO, located in Vantaa, southern Finland, with an approximate storage volume of 1 million m³ and a heat capacity of 90 GWh. The dataset comprises optical or acoustic borehole images (OBI and ABI) from 36 boreholes, fracture observations from oriented drillcores and field observations, and 3D deformation zone models available from earlier work.

The results indicate that classification of discontinuities from the OBI and ABI images based on filling type can present challenges in distinguishing between brittle fractures and other structures such as dikes without a brittle interface. In addition, variations in the OBI and ABI image quality may lead to intervals with limited or less distinct observations. These findings highlight the value of integrating supplemental data sources, such as the fractures mapped from oriented drill cores and field observations, to enhance interpretation and overall representativity of the fracture data. Moreover, using the deformation zone models to constrain the bedrock domains results in a domain pattern that is challenging in terms of drillhole fracture data availability for DFN modeling. Therefore, we briefly discuss filtering of the domains based on available data.