Reservoir Characterization to Optimize CO2 Injection

Abstract Carbon capture, utilization, and storage (CCUS) projects require detailed reservoir characterization to ensure safe and effective CO2 injection. This study focuses on the Triassic Baldonnel Formation in western Canada as a target for CO2 sequestration. We applied a deterministic Amplitude-Versus-Offset (AVO) seismic inversion using a three-term Aki–Richards linear approximation (Aki and Richards, 1980) to obtain elastic properties in the Baldonnel. Comprehensive seismic data conditioning (including azimuthal alignment, amplitude balancing, and bandpass filtering) was performed to improve inversion fidelity. A robust low-frequency model (LFM) was built by integrating one key well log with seismic processing velocities, adjusted via time shifts obtained from the alignment. Then, a rock physics inversion was conducted using a regression-based model constrained by theoretical bounds (Voigt, Reuss, and Hashin–Shtrikman limits) and multi-mineral fluid substitution using Gassmann’s equation. The rock physics model was calibrated explicitly for the Baldonnel formation’s mixed lithology consisting primarily of dolomite and limestone with clay and siltstone interbeds. The inversion predicts spatial distributions of porosity and mineral fractions (clay, calcite, dolomite) across the reservoir, which serve as critical inputs for reservoir modeling and simulation. Results reveal that higher porosity zones in the Baldonnel are laterally continuous where dolomitization is prevalent, whereas deeper intervals lack significant porosity. A horizontal CO2 injector well drilled post-inversion confirmed the absence of high porosities in the lower Baldonnel, validating the inversion’s predictions. The operator is currently injecting ~40 tonnes of CO2 per day into the Baldonnel and plans to scale up to ~500 tonnes/day in the near future. Our study provides valuable insights into using deterministic seismic inversion for CCUS: by integrating rock physics and inversion products into dynamic models, we optimize injectivity and storage capacity while ensuring containment. New porosity maps based on this workflow demonstrate significantly improved reservoir characterization when seismic inversion is incorporated, compared to maps derived only from available well control. These outcomes underline the impact of advanced geophysical workflows on CO2 injection planning and set a precedent for CCUS best practices in similar reservoirs.