Automating Core Data Integration: Insights from the Volve Field Case Study

This study presents an integrated, multi-scale workflow for quality evaluation in Middle to Late Jurassic reservoir in the Volve Field, Norwegian North Sea (Hugin Formation), focusing on Volve 15/9-19A and 15/9-19BT2 wells. It aims to streamline core data interpretation from reservoir to pore scale using machine learning algorithms, enhance reservoir characterization efficiency and accuracy through the unification of geological and petrophysical data across different spatial resolutions. The proposed workflow integrates core gamma-ray responses with high-resolution core image analysis and applies a random forest classifier to automate lithological descriptions. These outputs are refined through integration with thin-section petrography for detailed facies analysis. Facies are then correlated with core petrophysical measurements and poral network characteristics derived from microscopy. Finally, K-means clustering is applied to petrophysical well logs into discrete reservoir quality classes. The integration of these classes with facies analysis and petrophysical evaluation enabling a robust, data-driven approach to reservoir zonation and quality assessment. Automated lithological description significantly reduces interpretation time while maintaining consistency and quality. Multi-scale reservoir characterization within the Hugin Formation enabled identification of three distinct reservoir quality zones in the Volve 15/9-19A well, and two zones in the 15/9-19BT2 well. These zones are separated by lithological and diagenetic heterogeneities. Reservoir zone one (23.74 m thick) on the Volve 15/9-19A well, is dominated by mouth bar, tidal bar, and upper shoreface facies. This interval is primarily associated with good reservoir quality, with localized sections of excellent quality. However, the presence of lithological heterogeneities results in localized poor to moderate reservoir quality. Reservoir zone two (19.32 m thick) is dominated by tidal channel facies, with minor contribution from tidal bar facies. This zone is largely associated with excellent reservoir quality, with subordinate intervals of good quality. Reservoir zone three (12.96 m thick), is composed mainly of mouth bar and tidal channel facies. It is predominantly characterized by good reservoir quality, with minor sections exhibiting excellent quality. However, the presence of lithological heterogeneities (represented by tidal flat facies) and diagenetic overprint within the mouth bar facies leads to localized poor reservoir quality. In the Volve 15/9-19BT2 well, two reservoir quality zones were identified within the Hugin Formation. Reservoir zone one (17.43 m thick) is dominated by tidal channel facies. This interval is primarily characterized by good reservoir quality, with minor sections exhibiting moderate quality. Limited diagenetic heterogeneities are observed near the top of the zone, locally reducing reservoir quality. Reservoir zone two (26.70 m thick), comprises a mix of tidal bar and tidal channel facies. This interval exhibits the best reservoir quality in the 15/9-19BT2 well; it is predominantly composed of rock type 4, reflecting good reservoir performance throughout the zone. This integrated workflow illustrates the efficiency of machine learning in automating core interpretation and enhancing subsurface understanding of this mature offshore field.