Pinpointing hydrocarbon-bearing intervals remains a key challenge in reservoir characterization, impacting well placement and development and production plans.
Traditional workflows, relying on seismic inversion and simplified rock models, struggle with uncertainties from limited well data and oversimplified geology, which may reduce confidence in fluid predictions and lead to costly misinterpretations.
At the Dig X Subsurface 2025 conference in Oslo in December, Åsmund Heir, Managing Director at RagnaRock Geo, unveils a deep learning workflow to predict elastic properties like acoustic impedance and Vp/Vs from pre-stack seismic data, well velocities, and geological horizons.
Tested in the decommissioned Volve field, it maps hydrocarbon zones by projecting properties onto a rock-physics template, ensuring geological consistency.
Unlike traditional seismic inversion, which demands extensive tuning and hours of processing, this method delivers precise fluid predictions in minutes. Results show elastic properties correlating with well data at up to 0.92, with blind-well tests at up to 0.85.
Fluid maps reveal clear hydrocarbon plumes, outperforming interpolation methods that blur reservoir boundaries.
This workflow promises a robust seismic-based hydrocarbon detection, with potential for applications also in carbon storage and geothermal reservoir planning.
Join us at Scandic Fornebu, December 03-04, 2025, to learn more. The program can be found on the conference website.
