Item

Abstract

We present a part of the ThermoBase project, where we developed an end-to-end workflow and database effort to transform heterogeneous subsurface data into high-resolution structural models and OGS-ready thermo-physical meshes for nuclear repository site evaluation in Germany. We compile and harmonise lithology and stratigraphy information from multiple institutional and industry sources, covering thousands of boreholes. Where available, borehole temperature logs and laboratory measurements of physical properties were linked to records in the database to populate model units and to support validation. We also utilize a thermal property prediction tool based on geophysical logs, which in many cases first required digitalization, but provides immense value in areas with sparse direct measurements.
For selected salt geometry dominated sub-areas, existing regional 3-D models were augmented with the compiled borehole stratigraphy to construct detailed structural frameworks in Petrel. Horizon surfaces extracted in Petrel were imported into FEFLOW to build conforming finite-element meshes; those meshes were converted to vtu files for OpenGeoSys (OGS) using the feflow2ogs utility from ogstools. This Petrel → FEFLOW → ogstools → OGS toolchain allowed us to produce simulation-ready meshes suitable for conductive thermal modelling and subsequent envisioned basin-scale upscaling.
Parallel to modelling, ThermoBase supported a targeted sampling campaign focusing on clay and salt sealing units — the repository-relevant lithologies. Laboratory measurements of thermal conductivity, heat capacity and related physical properties were used to parameterise and ground-truth numerical models at locations where cores were available.
Because data volume and heterogeneous quality dominated the project effort, the presentation emphasises “lessons learned” from ThermoBase: the necessity of rigorous, standardised metadata and stratigraphic lexicons; automated and reproducible quality-control routines; early and explicit alignment of coordinate systems and stratigraphic conventions; transparent gap-filling and uncertainty documentation; robust version control and provenance tracking; and iterative coupling between geological and numerical modellers. We also report practical gains and limitations of the chosen toolchain, and give recommendations to reduce bottlenecks when scaling from sub-area structural models to basin-scale thermal simulations. ThermoBase aims to increase reproducibility, reduce manual screening effort, and provide a scalable foundation for repository-relevant subsurface modelling in Germany.